BOSCHFUEL

INJECTIONSYSTEMS

Forbes Aird

HPBooks

-------- ------

HPBooks

are published byThe Berkley Publishing Group

A division of Penguin Putnam Inc.375 Hudson Street

New York, New York 10014

First edition: July 2001ISBN: 1-55788-365-3@ 2001 Forbes Aird10987654321

This book has been catalogued with the Library of Congress

Book design and production by Michael LutfyCover design by Bird Studios

Interior illustrations courtesy of Bosch, Inc. and the author as noted

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form,by any means electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of thepublisher.

NOTICE: The information in this book is true and complete to the best of our knowledge. All recommendations on partsand procedures are made without any guarantees on the part of the author or the publisher. Author and publisher disclaimall liability incurred in connection with the use of this information. Although many of the illustrations in the pages thatfollow were supplied by Bosch and used with their permission, this publication is a wholly independent publication ofHPBooks.

I am grateful to Robert Bosch GmbH in the persons ofGerhard Kopany, for permission to reproduce copyrightphotos and illustrations, and of Wolfgang Boerkel, foranswers to some technical questions. Thanks are also dueto Wolfgang Hustadt of Bosch North America and to BillRoth, for further technical information.

~- -

A;thOUghgasoline fuel injection (FI) has beenaround just about as long as the automobile'tself, it has always been a mysterious

technology. Until about 1970 it was both rare andexpensive, restricted to some aircraft applications, and toa hand-full of exotic, high performance cars and racers.Production cars - even high performance sports cars-made do with carburetors.

Thirty-plus years later, and as a result of the"electronics revolution," that situation has been verynearly turned on its head. While cars in many racingclasses wear carburetors, virtually every production carin the world has FI! Yet it remains a mystery to most.

This book is an attempt to de-mystify fuel injection.While it deals principally with the various electronic fuelinjection systems produced by the Robert Boschcompany, much of what is said in the following pagesalso applies to other systems.

No book of this size-indeed, likely no book of anysize-could fully describe the minor variations in FIinstallations between one vehicle model and another.

Still, it is hoped that sufficient detail is provided to be ofbenefit to mechanics both amateur and professional,

while the general principles described will be useful forthose attempting performance tuning, and informativefor readers who simply seek to understand the mystery.

CONTENTS

CHAPTER 1Food For Engines, Food For Thought

CHAPTER 2Fuel Injection: Then and Now

CHAPTER 3Bosch Intermittent Electronic FI

CHAPTER 4Motronic Engine Management

CHAPTER 5Troubleshooting Bosch Intermittent Electronic FI

CHAPTER 6Bosch Continuous Injection

CHAPTER 7Troubleshooting Bosch Continous Injection

CHAPTER 8Performance Modifications

1

19

37

59

75

85

109

123

If a certain fixed quantity of air-or anyother gas- is confined in a closed con-tainer and then heated, the pressure

inside the container will rise. If one of the

walls of the container is moveable, the inter-nal pressure will push that wall outwardwith a certain amount of force, according tohow much heat was put into the trapped gas.

That, in a nutshell, is the working princi-ple of all internal combustion engines: Eachcylinder is a closed container, and each pis-ton represents a moveable wall of that con-tainer; the heat is supplied by the burning ofa fuel, usually gasoline, and the trapped gasis whatever mixture of gaseous compoundsleft over after the burning.

Meanwhile, the other moving parts of anengine are there for one or the other of justtwo supporting functions. The "bottom end"converts the movement of the pistons intorotary motion and, by returning them to thetop of their strokes, restores the closed con-tainers to their original size; the valve gearand everything else at the "top end" are theresimply to provide for the emptying out ofthe spent gasses and the refilling of thecylinders with a fresh charge of burnablemixture.

This may all seem very obvious to anyonewith even the most basic understanding ofhow engines work, but lurking within thesimple facts outlined above is a wealth ofdetail. Consider the fuel, for example. Somefuels contain more chemical energy perpound than others, and so can produce moreheat when burned. Even limiting the discus-sion to gasoline, the fact is that ordinarypump gasoline is a mixture of hundreds ofdifferent flammable compounds, and eachof those compounds has a different potential

ability to generate heat when burned. Theexact nature of the mixture of these com-

pounds varies from one pump to another andfrom one season to the next, so a pound ofgasoline from one pump on one day mightrelease somewhat more or less heat when

burned than would a pound from anotherpump, or from the same pump on someother day.

While each is unique, all the hundreds ofcompounds that make up gasoline have onething in common-they are all hydrocar-bons. That is, they are all made of just twokinds of atoms, hydrogen (H) and carbon(C). The difference between one of thesehydrocarbons and another lies in either thenumber of hydrogen and carbon atoms, or inthe way in which these two component ele-ments are arranged, or both.

Now, burning is a process of oxidation-acombining with oxygen (O)-so, reduced toits basics, when a hydrocarbon fuel likegasoline burns, individual hydrocarbon mol-ecules from the gasoline combine with indi-vidual molecules of oxygen from the air.The hydrogen (H) in the hydrocarbon com-bines with some of the oxygen (0) in the airto produce water (H2O), while the carbon(C) in the hydrocarbon combines with therest of the oxygen to form carbon dioxide(CO2)' In this process, a large amount ofenergy gets released, in the form of heat.This chemical dance amounts basically to areversal of the processes that went into cre-ating the hydrocarbons in the first place. Seethe box, "Sunlight by the Gallon."

Air, too, is a mixture of substances,although all of them are gasses at room tem-perature. About 78 percent of our atmos-phere is nitrogen (N); only about 21 percent

a: Richmixtureb: leanmixture

..s:;:""

c:

.8"E..

c...E-~=> c:c.. 0-;;'-'0Q) =>LL.

~-~""c..V'I

0-'-'-0V>.!!.!><'i§~c -=>'-'0="".2:"6

Q)eo:::

. a b ..~I.

0.8 1.0 1.2

ExcessAir Factor

co

0.6 1.40.8 1.0 1.2

ExcessAir Factor

Burning fuel with insufficient air results in carbon monoxide (CO) being formed, instead of carbon dioxide (CO2)' Excess fuel also leavesunburned hydrocarbons (HC) in the exhaust. Paradoxically, a too-lean mixture causes misfires, which also lets unburned fuel escape.Oxides of nitrogen (NOx) are produced in greatest abundance when the mixture strength is just a bit lean, when combustion is hottest.(Robert Bosch Corporation)

of it is oxygen. The remaining one percentor so is made up of several rare gasses, likeneon and argon, plus CO2 and water vapor.The chemical reaction of burning gaso-line-especially inside the cylinders of anoperating gasoline engine-is further com-plicated by the presence of these other ele-ments, and particularly the nitrogen.

Nitrogen is a comparatively inert sub-stance-it does not readily react with any-thing much, so in a simplified description ofthe burning of gasoline in air, the nitrogen isignored, on the assumption that it passesright through the whole operationunchanged. In fact, that is not quite true.Exposed to the enormous temperatures andpressures in the combustion chamber of anengine, a little of the nitrogen does end up

2

combining with some of the oxygen, form-ing various oxides of nitrogen- NO2, NO3and so on-known collectively as NOx.While for most purposes the minor involve-ment of the nitrogen does not make muchdifference, these nitrogen oxides are air pol-lutants. Thus, while the idea of "burning" afuel seems a simple business, here is just onefactor that begins to reveal that it is some-what more subtle and complex than it at fIrstappears.

Stoichiometric Mixture

Within narrow limits, a fixed quantity(that is, weight) of air contains a certain spe-cific numberof oxygenmolecules,and anygiven weight of any specificgasolinelike-wise contains some definite number of

Vegetables are a lot smarter than you might think. Millions of years ago, they man-aged to harness the energy of sunlight in order to manufacture themselves from thesimple raw materials available to them, specifically water (H2O) from the ground andcarbon dioxide (CO2) from the atmosphere. A plant-a tree, for instance-grows byusing solar energy to split the CO2 into separate atoms of carbon (C) and oxygen (0),and then to combine the carbon with the water to make molecules of cellulose

(C6HlOOS)'In effect, a tree uses water and carbon to make more tree.From the tree's point of view, the oxygen remaining after the carbon is split from the

CO2 is incidental, and so is cast aside, into the air. From our point of view, this aspectof the process is vital- it is the reason we have a breathable atmosphere. (And youthought those tree-huggers were as dumb as vegetables!)

Usually, once a plant dies, the cellulose breaks down, but the process is not exactlya reversal of the original chemistry. In fact, some of the carbon and hydrogen the treehas spent so much of its sun-energy input combining remain hooked together as mol-ecules having a certain number (say, "X")of carbon atoms and some number (say, "y")of hydrogen atoms, (CxHy), of which there are many hundreds of different ones. Ifexposed and left to rot, the major product is likely to be a gas-methane (CH4)' other-wise known as "swamp gas," or "fire damp.IIUnder certain conditions, however, entireforests can get buried and, over the course of countless millions of years, subject toenormous pressure from the weight above, the carbon and hydrogen atoms get re-shuf-fled into more tightly clustered hydrocarbon molecules, many of which are liquids. Allof the world's oil (hydrocarbon) deposits are the age-old remains of this process ofdecay and chemical rearrangement of vegetation.

When we subsequently extract, refine and bum some of this liquid sunshine, we arerecombining the hydrogen and carbon with oxygen. When we do so, the swapping ofchemical partners that occurs liberates all of the considerable solar energy that, overyears, went into the original separation processes. And it is that energy, in the form ofheat, that makes the wheels go 'round.

hydrocarbon molecules. Because the burn-ing process amounts to individual atomscombining with each other, it follows thatthere is only one particular ratio of gasoline-to-air that can ensure that all of the oxygenmolecules mate up with all of the hydrocar-bon molecules. This theoretical ideal iscalled a stoichiometric mixture.

If there is an excess of oxygen molecules,

some of them will fail to find partners. Interms of the number of oxygen-hydrocarbonpairings, and thus the amount of energyreleased, the effect is as if we had startedwith a smaller quantity of air. At the sametime, if there are too many hydrocarbonmolecules in relation to the amount of air,then some of the hydrocarbons will emergefrom the combustion process unburned.

3

Some of the gasoline is simply wasted. Notonly that, but a shortage of oxygen meansthat there is a likelihood of some of the car-

bon atoms in the hydrocarbon fuel to com-bine with just one oxygen atom, rather thantwo, yielding carbon monoxide (CO) ratherthan carbon dioxide (CO2)' While CO2 isone of the" greenhouse gasses" that are part-1yresponsible for global warming, at least itis only immediately harmful to animal lifewhen its concentration grows so large that itdisplaces much of the oxygen we need tobreathe. CO, on the other hand, is toxic evenin small doses.

It turns out that about 14.7 lbs. of air con-

tains the correct number of oxygen mole-cules to pair up with the number of hydro-carbon molecules in 1 lb. of gasoline. Theratio of air to gasoline to achieve a stoichio-metric mixture, in other words, is approxi-mately 14.7:1, by weight. Note we sayapproximately - there is no single numberthat correctly identifies the stoichiometricmixture for all gasolines. To explain, recallthat gasoline is a mixture of hydrocarbons.Each has its own stoichiometric mixture

strength, ranging from less than 13:1 tomore than 15:1, so the stoichiometric ratio

for the entire blend depends on the propor-tions of the differing hydrocarbons thatmake it up. Apart from incidental variations,the major oil companies deliberately modifythe blend of hydrocarbons in pump gasolinefrom season to season and from place toplace, so the stoichiometric mixture maycorrespondingly vary slightly, according towhere and when you buy the fuel. (See alsothe sidebar "The Oxygen Battery," page 34.)

As we have said, gasoline, strictly defined,contains only hydrocarbons, but oil compa-nies have also begun, fairly recently, toinclude certain additives in gasoline that fur-ther affect the chemically correct mixture.Among the additives commonly found inboth pump and racing gasolines are ethylalcohol (ethanol) and methyl tertiary butylether (MTBE). Both these substances are

4

examples of what are called oxygenates-like the hydrocarbons we have been speak-ing of they contain hydrogen and carbon,but unlike the hydrocarbons they also con-tain oxygen. A fuel carrying its own oxygenadds to the amount inhaled by the engine, sothe presence of oxygenates means that a lit-tle extra fuel is needed in relation to the

quantity of air the engine is breathing in, totake account of the additional oxygen beingcarried within the fuel- the stoichiometric

ratio becomes a little (numerically) smaller.This is yet another reason why it is not pos-sible to specify one exact stoichiometricmixture strength for any and all gasolines.

Note, too, that the stoichiometric mixture

strength is expressed as a ratio of weights-or more correctly, masses-not volumes.(The mass of something is, in effect, a"count" of the number of molecules in it.) Acertain mass of air- that is, a certain numberof molecules - will occupy more or less vol-ume, according to its temperature. A cubicfoot of hot air contains fewer gas molecules,including oxygen molecules, than a cubicfoot of cold air. Other factors, like baromet-ric pressure and altitude also affect the den-sity of air-the weight of a certain volume,in other words. For that matter, the densityof gasoline also varies with temperature,though not nearly as much.

While the ideal of stoichiometry expressesthe chemically correct air-to-fuel ratio forany particular gasoline blend, gasoline will,in fact, burn in air over a spread of ratiosfrom about 6:1 to more than 24:1. Mixturesthat contain more fuel than the theoretical

optimum are said to be "rich," while thosewith an excess of air are termed "lean." For

maximum power production, there is some-thing to be said for mixtures that are some-what richer than stoichiometric.

To begin to explain, consider a four-cycleengine turning 6000 rpm. At that speed, eachpower stroke lasts just 1/400 of a second. Toget an idea of just how short a time that is,sight through the shutter of an unloaded

~Ii:!~0~Ii:!rn~0~

0.7511.0

0.8011.8

0.8512.5

0.9013.2

0.9514.0

1.0014.7

1.0515.5

1.1016.2

RICHER .... LEANER

1.1516.9

A

A/F ratio

Maximum power is produced with a mixture just a bit richer than the chemically "correct" ratio. Power tapers off sharply with increas-ing richness beyond that point, and more slowly as the mixture is leaned-out.

camera set to that speed, and push the but-ton. Even though the combustion eventinvolves extreme turbulence that violentlystirs and mixes the different molecules, it isextremelyunlikelythateach and everyoxy-gen molecule will be able to find a hydro-carbon molecule to react with in such a brief

flicker of time. Yet for maximum power wewant maximum heat, and the heat comesfrom the combining of the hydrocarbonmolecules in the fuel with the oxygen mole-cules from the air.

An engine'scylindersare of a fixed size,however, so the maximumquantityof air,andsothenumberof oxygenmolecules,thateachcylindercaninhaleis limited.Formax-imumpower,we want to make sure that allthe oxygenmoleculesavailablein the fixed

amount of air inside the cylinder react witha hydrocarbon, and the way to do that is toprovide some extra hydrocarbon molecules.And the way to do that, in turn, is to providea mixture that has a little excess fuel- a

slightly rich mixture. As noted, however,that extra gasoline is wasted; it also adds toair pollution. Unburned hydrocarbons, or"HC," are another of the exhaust pollutantsthat environmental laws seek to control.

On the other hand, if we are prepared tosacrifice a little power, we can make maxi-mum use of the quantity of fuel burned byproviding a slightly lean mixture. In thesame way that a little surplus fuel ensuresthat all the oxygen gets used, a little extra airhelps ensure that every hydrocarbon mole-cule finds an oxygen molecule to mate with.

5

This can reduce, if not eliminate, HC emis-sions. Within limits, it also leads to lower

fuel consumption for a given power output.Most fuel injection (PI) systems-and

most carburetors, for that matter- take these

considerations into account in their designand operation. During light load operation,such as occurs when cruising at a constantmodest speed with a comparatively smallthrottle opening, the system leans the mix-ture out a bit, to enhance fuel economy andminimize HC pollution. When the driveropens the throttle wide, demanding fullpower, the system provides a richer mixture,at some cost to fuel economy and HC levelsin the exhaust.

There are other aspects to the rich-mix-tures-equal-maximum-power issue. First,when gasoline evaporates, it soaks up a lotof heat in the process, as you probably knowfrom spilling gas on your hands in coldweather. The internal cooling effect of aslightly rich mixture reduces internal tem-peratures somewhat, especially in criticalareas like the piston crowns and the edges ofexhaust valves. While modem street enginesare boringly reliable, the internal coolingprovided by a surplus of fuel can make aconsiderable difference to the survival of a

race engine that is running on the raggededge of thermal self-destruction.

Also, the heat soaked up in the process ofboiling that excess liquid gasoline into vaporcan reduce the temperature of the air/fuelmixture entering the engine. As we havepointed out, cooler air is more dense thanhotter, so a cylinder full of an intake mixturecooled in this way will weigh more (andthus contain more oxygen molecules) thanotherwise. This accounts for some slightpotential gain in power output.

Detonation

Another consideration relating to the con-nection between mixture strength and poweris the issue of the tendency of a gasoline/airmixture to detonate. To explain, the burning

6

of fuel inside an engine cylinder is oftencharacterized as an explosion, but althoughthe combustion event is extremely rapid, it isnot, technically, an explosion. Once initiatedby the spark, the burning begins as a smallbubble of flame around the plug electrodes.Under normal conditions, the burningprocess then spreads rapidly but smoothlythroughout the rest of the mixture as anexpanding ball of fire.

In some circumstances, however, the com-

bustion may start off smoothly enough, butas the flame-front expands through the com-bustion chamber, the rapidly rising tempera-ture and pressure ahead of it causes complexchemical changes in the unburned mixturefurthest away, called the end-gas. Squeezedand heated by the approaching fireball, itchanges from a predictable, slow burningmix into something far more unstable. As aresult, the overheated end-gas ignites spon-taneously almost all at once, the pressureinside the cylinder rises so fast it is more likean explosion than a controlled bum, and theresulting shock wave rings through themotor. That is detonation, or "knocking."

The sharp pressure spike that results whenthis violent secondary event meets up withthe original flame front can punch holes inpistons. Even if it does not, the turbulencecreated by detonation scours against the sur-faces of the combustion chamber, allowingheat to flow out of the swirling gasses andinto the surrounding metal much faster thannormal. As a consequence, the gasses loseheat, their pressure accordingly falls, andpower drops off immediately. (Although thepeak pressure during detonation is muchhigher than during normal combustion, theaverage pressure is way down, because ofthis heat loss.)

Because the changes preceding detonationare chemical, the ability of a particular blendof gasoline to resist detonation depends onthe chemistry of the blend, and thus in turnon the various hydrocarbons that make it up.Overall, the knock-resistance of any sample

~0.r-!

--j-J

~s~IJl

~~0 H

U~r--i p...,

Q) ~~ ~~~

,DU ~

.r-! ~4--1'F9

UOJ~

UJ

maXImumeconomy

maXImumpower

Horsepower..

of gasoline is expressed by its octane rating(see the sidebar "Doing Octane Numbers"),but the number so obtained also depends tosome extent on the mixture strength. Somegasoline components knock worst when runrich; some others increase substantially inknock-resistance with wholesale enrich-

ment. Predictably, these latter are found inabundance in race gas.

For typical pump gasoline in a typicalengine, the mixture ratio for peak power islikely to be in the area of 12:1. Dependingon the particular blend of gasoline, anythingricher than that may exaggerate detonationproblems, and the cooling effect of the sur-plus fuel, if carried to extremes, may sapsome of the heat that we would rather have

working to raise the gas pressure. For better

mileage and lowest HC emissions, some-thing closer to 16:1 is wanted. Indeed, atcomparatively high engine speeds undervery light load, mixtures as lean as 18:1 mayoffer even better fuel economy. Such leanmixtures burn hot, however, and that extraheat, together with all those extra oxygenmolecules, makes it more likely that the sup-posedly inert nitrogen will combine withsome oxygen, worsening the NOx emis-SIons.

Optimizing an Engine's DietWhile the generalizations above are

broadlyapplicableto all engines,establish-ing the correctair:fueldietfor anyparticularengineover a full rangeof speedsand loadscan only be achievedby a long and tedious

stoichiometry

A mixture "fish hook,"showing the points ofmaximum power and ofmaximum economy. Fuelflow is expressed in termsof the quantity used perhour per horsepower pro-duced.

7

2,000 6,000

100 200 300

II

400

Horsepower

A series of fish hooks established at various engine speeds produce, when linked together, a curved band that represents the practicalrange of mixture strengths for any given engine. The line forming the upper boundary of this band represents the mixture curve for max-imum power; the lower one for maximum economy.

8

process that involves dyno testing. Theengine is run at some fixed throttle opening,and the load is adjusted to keep the rpm con-stant. Starting with a very rich air:fuel ratio,the mixture is adjusted leaner in small steps,and the fuel flow is measured at each settingin, say, pounds per hour. As the mixture isgradually leaned out, power initiallyincreases until some maximum is reached.

Further leaning results in a reduction inpower but, initially at least, the quantity offuel burned for each horsepower producedactually grows smaller.

This relationship between fuel consump-tion and power production is described by

the expression brake specific fuel consump-tion, or BSFC. The first word, "brake," sim-ply refers to the fact that the engine is beingrun under load on a dynamometer, or"brake." (The original dynamometer wassimply a friction brake; modem hydraulicand electric dynos differ in construction, butthe principle of using a retarding device toproduce an artificial load remains the same.)"Fuel consumption" is pretty obvious; it issimply the number of pounds per hour(lb/hr) of fuel being consumed at some par-ticular throttle opening and speed."Specific" is actually a contraction of"power-specific," meaning that the fuel con-

0.,....,.0.9...f-!

S I 0.8U2 ,.-....

0 I 0.7u

ill I 0.6

'? I 0.5.--1u '-'.,....,

I 0.4'+--i.,....,UQ)

U)

sumption at any given setting is divided bythe horsepower produced at that setting. Theresults are expressed in units of pounds offuel burned per horsepower per hour-Ib/hp/hr. As a rough rule of thumb, we canfigure that an unsupercharged engine burn-ing gasoline will have a BSFC of about 0.5Ibs/hp/hr at peak power. That is, an enginemaking 300 hp will need to bum about 150Ibs. of fuel per hour to do it.

At some mixture strength, the engine willproduce a maximum value of BSFC-amaximum amount of power from eachpound of fuel, but this will not actually givethe maximum possible power. At some rich-er setting, the power will likely be some-what higher, but disproportionately morefuel will have to be burned to achieve that

slightly higher peak power. As the mixture isleaned out past the point of peak BSFC, thepower falls off markedly, as you mightexpect. What is particularly interesting isthat the amount of fuel burned, in relation to

the power (the BSFC, in other words) actu-ally increases-although engines can bemade to run on such extremely lean mix-tures, it is actually wasteful of fuel to do so.The results of such a test are then plotted ona graph, with BSFC on the vertical axis andpower output on the horizontal axis. Theresulting curves resemble, and are called,"fish hooks."

Once one such test has been completed,the whole thing is repeated allover again atsome other engine speed, until the entireoperating speed range has been covered in,say, 500rpm steps. Then the entire set oftests is repeated at different throttle open-ings. As we said, the operation is tedious.

Special DietsOne special circumstancethat requires a

much richer than stoichiometricmixture iscold starting.It may come as no surprisetolearn that the various hydrocarbons thatmakeup gasolinehavewidelydifferentboil-ing points and so evaporate at differing

rates. At very low temperatures, some ofthem may not evaporate at all, so the onlyway to ensure there is enough of the onesthat do vaporize to make a burnable mixturein air is to provide a lot of gasoline overall.Typical cold start fuel-to-air ratios arebetween 2-to-l and I-to-l.

Historically, there are two other situationsthat are (or were) thought to demand a richmixture-idling and acceleration. Certainlyidle enrichment is needed on typical carbu-reted engines, and to a lesser extent on thosewith throttle body injection (TBI) systems,but this is mainly a matter of the problemsthat arise from trying to distribute from acentral point all the air:fuel mixture neededby a multi-cylinder engine. Proof that verylittle enrichment at idle is necessary in prin-ciple comes from current emissions-certi-fied production engines, which get by withidle mixtures very close to stoichiometric.

The other situation conventionally thoughtto demand significant enrichment is acceler-ation. Every successful carburetor evermade had either an accelerator pump thatshot in an extra squirt of fuel every time thethrottle was opened, or (more rarely) someother means to temporarily richen the mix-ture under sudden throttle opening. It seemsthat much (although not all) of this "need,"too, turns out to be due to secondary fac-tors - in this case the nature of carbure-tors - rather than a characteristic of the

needs of engines themselves. In view of that,this seems as good a time as any to veer offfrom our consideration of the kinds of food

that engines prefer and to look at the variousmethods for bringing that food to the table.

The Drawbacks of CentralDistribution

For satisfactory engine operation, whatev-er mixes the fuel and air has to closelymatch, in terms of air:fuel ratios, the various

feeding requirements of the engine underdiffering conditions, and must be able tomove smoothly and continuously between

9

Ideally, all fuel ingested by the engine would be in the form of vapor. In practice, some droplets remain.Squirting the fuel through a small orifice under considerable pressure improves atomization, which is onereason why fuel injection is superior to carburetors. (Robert Bosch Corporation).

them as the situation requires. There is moreto it, however, than just keeping the propor-tions right. Large blobs of fuel haphazardlydistributed throughout the air just will notdo, even if the overall proportions are cor-rect.

To begin to understand the reasons forthis, imagine setting fire to a tablespoonmeasure full of gasoline. Yes, it will bum offfairly fast, but consider that an engine mak-ing 225hp goes through about that much gasevery second. Allowing for the fact that eachpower stroke occupies at most half of a com-plete revolution of the crankshaft, and that it

takes two full revolutions for a completeengine cycle, the combustion event in theengine obviously occupies, at most, onequarter of that time. You cannot bum a table-spoon full of gasoline in one quarter of asecond.

Vaporization-If you were to divide thesame amount of fuel into, say, three tea-

10

spoon measures and set them all alightsimultaneously, then the gas will bum morequickly. If you further divide it into largedroplets, it will bum quicker still. The morefinely you divide the fuel, the more surfacearea each particle has in contact with theoxygen in the air, in relation to the volumeof fuel within the droplet, so the faster theenergy gets released. The ideal would be todivide the fuel into the smallest possibleunits-individual molecules. In that case we

will not see any liquid fuel at all; it will allexist as a true vapor. In fact, we cannot usu-ally get quite that close to perfection, so theintake charge will consist of a mixture of air,gasoline vapor and fine droplets. One of theinherent advantages of fuel injection overcarburetors is that the fuel is introduced into

the intake air under comparatively highpressure. In the same way that a showerhead produces a fine spray when the taps arecranked all the way open, but gives forth

large drops when the taps are nearly closed,the pressurized mist issuing from a fuelinjector helps this process of vaporization.. Intake PortAirflow-But thereis moretoit than that. Consider one cylinder of a 320cubic inch (ci) V-8, turning 6000 rpm. Thatindividual cylinder displaces 40 ci and soinhales that much air every second enginerevolution (again, assuming it is a four-cycleengine), for a total of 120,000 ci of air perminute. That air flows through the intakeports in the cylinder head and intake mani-fold which may have a cross-sectional areaof somewhere around 3 square inches (sqin). The average rate of flow through thathole is simply the volume divided by thearea of the hole it flows through, so thespeed of flow is:

120,000/3 = 40,000 inches per second, orabout37.8mph. .

Now 38 mph doesn't sound like a particu-larly high speed, but the air on its way to thecylinder usually has to negotiate some turns,and those turns can be mighty sharp-maybe something like 3" radius. If you both-er to do the arithmetic, you will discover thatthe airflow negotiating a turn with a radiusof 3" experiences an acceleration equivalentto 382 times the force of gravity (382 g).

Now, if all that is flowing through theintake ports is air and fuel in vapor form,those 382 g turns won't bother the gasses atall. But with any arrangement that mixes thefuel and air at a central location, those sharp,high-speed turns really disturb the move-ment of any fuel droplets that are mixed inwith those gasses. What will happen, in fact,is that they will get centrifuged to the out-side of the bend and form puddles of liquidon the inside surfaces of the ports.

At first it may seem that this does not mat-ter much; the fuel will get carried along bythe air rushing past and will eventually makeits way into the cylinder and the correct fuel-to-air ratio will be maintained, at least on

average. But "on average" is not goodenough; the mixture strength in the cylinder

will vary from moment-to-moment, accord-ing to the whims of the puddles. Of course,if there is only one cylinder, there is lessneed for bends in the intake plumbing, butthings turn really ugly when we are dealingwith more than one cylinder eating from thesame trough, so to speak.

When multiple cylinders are fed from onecommon source, as is the case with TBI andcarburetor induction systems, there mustinevitably be bends, and probably lots ofthem. Unavoidably, this fuel-drop-out effectwill favor some cylinders and short-changeothers. In the days before concern aboutemissions (which means in the days of car-buretors), a variation of four numbers inmixture strength between cylinders in thesame engine was not uncommon-somecylinders might be working on 16:1, otherson 12:1.To keep the engine lit at idle, it wasnecessary to provide a surplus of fuel over-all in order to ensure that the leanest runningcylinder got a burnable mixture. Withpainstaking development of the manifolddesign, it is possible to reduce this cylinder-to-cylinder variation, and modem manifolddesigns for engines fitted with carburetors orTBI systems do much better than in the badold days. Still, the desirability of ensuring asnearly complete vaporization as possibleshould be obvious.

At idle speeds, the speed of the flow ofgasses through the ports is obviously vastlyreduced, so the tendency of droplets to sep-arate out from the gas flow because of cen-trifugal forces will be dramaticallydecreased. At the same time, the high vacu-um condition existing in the intake manifoldof an idling engine encourages fuel dropletsto vaporize. In the same way that water boilsat a lower temperature (that is, evaporatesmore readily) on a mountaintop than it doesat sea level, gasoline evaporates more readi-ly in thin air than when it is more dense.Ironically, the problem of incomplete atom-ization remains - at least for carbureted

engines-simply because the lower rate of11

Single Point Fuel Injection (TBI)!.Fuel2.Air3.ThrollieValve

4. IntakeManifold

5. Injector

6.Engine

1===>

{J/0W---- 5

~13

6

Introducing fuel at a single point-whether from a carburetor or a throttle body injector-"wets" a largearea of manifold surface, and gives unevaporated droplets many chances to drop out, forming puddles offuel. (Robert Bosch Corporation)

flow meansreducedturbulencewhichmightotherwise break large droplets up into small-er ones.

Quite apart from the matter of the rate offlow, there are other factors at work that

oblige engines with a "central mixer" -andespecially those fitted with a carburetor-tooperate with rich air:fuel ratios at idle. Twoof these are charge dilution and reversion,terms we shall soon define.

Valve Overlap-To explain, in the kind ofidealized engine that appears in beginner-level explanations of how piston engineswork, the intake valve opens at top deadcenter (TDC) of the intake stroke and closesat bottom dead center (BDC). Following thecompression and power strokes, the exhaustvalve opens at BDC, and closes again atTDC. Yet surely anyone reading this bookknows that on all real world engines the cam

12

timing is arranged to open the valves earlierand close them later than this.

The reason for this is the inertia of the

intake and exhaust gasses. Although gassesare very light, they are not weightless-acubic foot of air, for example, weighs about0.08 lbs. Thus, as the valves open and close,and the columns of gas passing through theports on their way to and from the combus-tion chamber start and stop, the inertia ofthose gas columns makes them lag behindpiston movement. To give sufficient time foremptying and filling the cylinder, especiallyat high engine speeds, the intake valveopens before TDC and closes after BDC,while the exhaust valve opens before BDCof the power stroke and does not close untilafter TDC on the exhaust stroke. As a result,there is a period toward the end of theexhaust stroke when the intake and exhaust

~I~m-v n!Il

1-'.

rrJAi

r r f ,,A, i' 4, !

Multi-Point Fuel Injection (MFI)!.Fuel2. Air

3. ThrottleValve

4. Intake Manifold

5. Injectors

6. Engine

4

20

",P' I3

5

6

Multi-point injection ensures equal mixture strength at each cylinder. (Robert Bosch Corporation)

valves are both open together.This valve overlap helps an engine to pro-

duce useful power at high speeds, as the"lead" in the valve timing gets to be in synchwith the "lag" in the movement of thegasses. But it also leads to problems at verylow speeds, because there is then obviouslyan opportunity both for some of the freshintake charge to zip right out the exhaustand/or for some of the exhaust gasses tomake their way backward, upstream, intothe intake manifold.

Charge Dilution-This intermingling ofexhaust gasses with the fresh intake mixturethat occurs at idle because of valve overlapis termed charge dilution, and for years itwas argued that this demanded a rich idlemixture, because the dilution tends to keepthe hydrocarbon and oxygen molecules sep-

arate. A hydrocarbon molecule at one side ofthe combustion chamber, it was argued,might be desperately seeking an oxygenmolecule at the other side but they would beunable to meet, because of the crowd ofexhaust gas molecules in between.

While the above argument appears plausi-ble, it is apparently at least partly wrong, asconfirmed by the near stoichiometric idlemixtures of modem fuel injected engines.The difference between these motors and the

typical rich-idle engines of a few years agoseems mainly to be the difference betweenthe degree of variation in mixture strengthbetween one cylinder and another that existwhen the mixture for all an engine's cylin-ders is delivered at one central point, such asin a carbureted or TBI engine, compared toone with multi-point fuel injection, in which

13

each injector is located near the one individ-ual cylinder that it serves.

TBI engines, at least, are not afflicted byanother complication that affects carburetedengines at idle, that of charge reversion. Theviolent pulsations that occur in the intakeports and manifold of an engine at idle speedmeans that a portion of the flow on its wayto the cylinders sometimes actually reversesdirection- some of the intake air, alreadymixed with fuel, actually pops back outsidebriefly! This can sometimes be seen as apulsating cloud of air:fuel mist hoveringover the intake. Apart from being a potentialfire hazard, a consequence of this in-out-in-again motion is that some air passes throughthe carburetor three times. Because a carbu-

retor adds fuel to gas flowing through it, andneither knows nor cares which direction the

flow is going, fuel gets added on every trip.On the face of it, it should be possible toadjust the basic idle mixture setting to takeaccount of this triple-dosing. Indeed, a cor-rectly adjusted carburetor will do so, but thereversion phenomenon is unpredictable, sothe mixture has to be set somewhat rich totake account of the instants when it is inef-fective.

Another area where fuel injectiori - atleast multi-point Fl, with each injector locat-ed very near the intake valve it serves-hasan inherent advantage over carburetors is theability of Fl systems to cope with suddenincreases in engine speed and load. There iscertainly a need to shift from the leaner-than-stoichiometric, best-BSFC air:fuelratio, to the richer-than-stoichiometric best-

power ratio if the throttle is suddenly openedwhile cruising along at light load, but whyshould the actual fact of acceleration itself -

that the engine speed is increasing - intro-duce any greater need for enrichment at anyinstant during the acceleration than does anyother situation demanding that same poweroutput at the same rpm?

The answer to this is that, as previouslynoted, some of the mixture supplied by any

14

"central mixer" will wind up as liquid on thewalls and floors of the ports and manifold.This fuel will eventually flow to the cylin-ders, but note that the density and viscosityof this liquid fuel is much greater than thatof the air, gasoline vapor, and very tinydroplets that make up the remainder of themixture, so its motion is much slower than

those gaseous and near gaseous compo-nents.

Under steady state conditions, this "pool"of fuel is being consumed and replacedmore or less equally all the time, so underthose conditions there will always be somemore-or-less constant quantity of liquid inthe manifold at any moment. That quantity,however, will vary according to just whatsteady state conditions prevail. Under largethrottle opening/high load conditions, thedensity of the air in the manifold and ports ishigher than otherwise and, as noted above,gasoline evaporates less readily in air that isdense than when it is thinner.

An engine operating under a large loadwill thus require a larger quantity of liquidfuel in the manifold in order to maintain

equilibrium between what gets consumedand what gets supplied. On sudden throttleopening, additional fuel thus has to be sup-plied to rapidly build this pool up to the larg-er size of "store" needed to maintain equi-librium under the new, higher load condi-tions.

It is noteworthy, too, that to encouragevaporization under these and other condi-tions, carbureted engines for street use arealmost always provided with some means toheat the mixture in the manifold, whether bythe heat of the exhaust gasses (the familiar"heat-riser") or of the engine coolant. Whilethis reduces the problems of having compar-atively large amounts of liquid fuel washingabout in the manifold, the heating of theentire intake charge lowers its density. Aswe have already seen, hot air is less densethan cooler air, so a cylinder full of such awarm mixture will contain fewer oxygen

Mixtureformationintermittentinjectionontotheengineintakevalve.

Multi-point injection also provides the opportunity to locate each injector very near the valve it serves.This reduces port "wetting," and so reduces the amount of enrichment needed for acceleration. (Robert

Bosch Corporation)

molecules- and thus can make less

power- than the samecylinderfilledwith acool mixture. The vaporization problem ismost severe when the engine is cold, somany engines provided with such a "hot-spot" have some means to turn down theheat once the engine is fully warmed up.Usually, however, the heat does not get com-pletely turned off.

To some extent, this problem of wetting ofmanifold and port walls also affects multi-point fuel injection systems, to a degree thatdepends on where the injector is located, rel-ative to the intake valve. On some engines,the injector is as close to the back of theintake valve as possible. Here, clearly, thelength of port that is wetted by fuel isextremely short, so the additional amount offuel needed to fatten up the "store" is mini-mal. However, on other engines - perhapsfor reasons of packaging, or access for ser-vicing- the manufacturer decides to locatethe injectors a considerable distance fromthe valves. The more remote the injectors,and so the greater the length of the wettedport and manifold, the more the additional

fuel requirements on acceleration resemblethose of a central mixer arrangement, suchas a carburetor.

Carburetors have an accelerator pump totide them over this transition, but the amount

of additional fuel supplied by the acceleratorpump of a carburetor considerably exceedsthe amount that should be needed to main-

tain the "pool."To seewhy this is so, and tobetter understand some of the other short-

comingsof carburetors,we shouldend thischapter by taking a brief look at their nature,before moving on to consider the fuel injec-tion systems that have now almost com-pletely displaced them.

The Carburetor

The earliest attempts to feed a gasolineengine hinged on wacky arrangements like adrip-feed of fuel into the air intake pipe, oran arrangement of cotton wicks with theirbottom ends immersed in a pool of fuel andtheir upper ends exposed in the intake pipe.These hit-and-miss methods at controllingmixture strength came to an end soon afterthe invention of a basic carburetor, in 1863,

15

DISCHARGETUBE

VENTURI

A rudimentary carburetor.The essential features are a"pool" of gasoline main-tained at a constant heightby a float-operated valve; arestriction (the main jet) toregulate the flow; a venturi,with a discharge tube con-necting the low pressurearea in the venturi to thefuel supply; and a throttleplate, to control the totalamount of mixture flowingthrough. There is also a cal-ibrated air bleed, thatmeters extra air into thefuel before the dischargetube, usually through a per-forated "emulsion tube,"that helps to atomize thefuel.

16

,'0°0',°0°,

,gogo°0°0°,,°0°0°0'°0°0°0'

***g,'°0°0°0°0°,gogogogog,°0°0°0°0°00

;ggogogo:o:,

Cigggg ggg~g0'0°0 °0°0 ,00<5;0 °0'00,

,0gO;? gogo,0000 00000

,gogo ogog,'00000 °oooc'00000 00000'00000 00000,'00000 00000,;ogogo °ogogc,gogog ogogog'00000 000000'00000 000000''00000 00000''00000 00000''000000 oOoooc'000000 000000'000000 000000

ogogog 020202,000000 000000'000000 °00000(°o~o~o 0 0 Q

attributed to a Frenchman by the name ofLenoir. By 1894, the German engineerWilhelm Maybach had advanced the devel-opment of this device by including the now-familiar float-and-needle-valve arrange-ment, to maintain a reliable supply of fuel ata constant height. Arguably, no small part ofthe success of Karl Benz's "Patentwagen" of1897 was attributable to its wearing one ofMaybach's float-equipped carburetors.

The adjacent illustration shows the essen-tial workings of a simple carburetor. Fuel issupplied by a pump to the float chamber (orbowl), where a float and needle-valveassembly maintains the contents at a fixedlevel, very much like the float and valvearrangement in your toilet tank. The floatchamber connects to a smaller reservoir-the main well- so the fuel level in the well

will be at the same height as in the floatchamber. A pipe, usually termed the dis-

FLOATCHAMBER

-----------

MAINWELL

charge tube, leads slightly uphill from thewell to the carburetor's throat-the main air

passage through it. With the engine stopped,at least, the level of the fuel will reach partway up the discharge tube, close to- but notquite-spilling out the end. At the pointwhere the discharge tube enters it, the car-buretor throat is fitted with a narrowingpiece, called the venturi. Note, too, that boththe well and the float chamber are open tothe atmosphere at the top.

As the engine draws air through the carbthroat, the restriction created by the venturiobliges the air to speed up. It is no coinci-dence that the venturi is named after an

Italian physicist of that name; it was he,G.B. Venturi, who established, a couple ofhundred years ago, that when a fluid passingthrough a pipe or channel is forced to speedup, its pressure drops. As a result of this phe-nomenon,the pressurewithinthe venturiis

DOING OCTANE NUMBERS I

"Octane" is a measure of a fuel's detonation resistance, compared to two referencefuels, both of them components found in pump gasoline. One, a particular hydrocarboncalled iso-octane, strongly resists knocking. It defines one end of the scale-lOa octane.The other reference hydrocarbon, n-heptane, detonates like crazy, so it defines the zeroend of the scale. Mix the two together and you get a tendency to knock that varies fromzero to 100 according to the proportions of the mix. A fuel under test that behaves like a90:10 mix of iso-octane and n-heptane, then, would be rated at 90 octane.

The actual measurement is performed using a standardized single cylinder test enginewhich is specially designed to allow its compression ratio to be varied while the engineis running. During the test, the compression is raised until the engine just starts to knock.From knowing exactly what blend of iso-octane and n-heptane would also just barelyknock at that CR, it is possible to establish the octane number.

Just to keep things baffling, though, there are two octane numbers, "Research" and"Motor," established under different test conditions. From the chart below it appears thatthe Motor method provides a more severe test. In fact, the Research method tallies clos-er to real world experience in ordinary cars on the road. To add to the confusion, the num-ber printed on a gas pump is the average of the Research and Motor numbers.

TEST CONDITIONS FOR RESEARCHAND MOTOR METHOD OCTANE RATINGS

Research Method(RON)25.6212

600 rpm13 degrees BTC

Inlet air temp. 1

Coolant temp.Engine speedSpark advance

Air/fuel ratio Adjusted formaximum knock

lower than that of the surrounding atmos-phere, which thus pushes the fuel out of thedischarge tube and into the carb throat.

The rush of air past the end of the dis-charge tube tends to tear the fuel intodroplets, thus going some way to achievingatomization. An air bleed at the top of themain well allows a certain amount of air to

leak into the flow of fuel, via a perforatedsleeve called the emulsion tube, before thefuel reaches the discharge tube, thus assist-ing atomization. This is, in fact, almost a

Motor Method(MON)300.2212

900 rpm19-26 degrees BTC(varies with CR)Adjusted formaximum knock

secondary function of the air bleed. Its pri-mary reason for existence is to deal with anodd phenomenon that would otherwisecause the mixture to become ever richer as

the rate of airflow through the venturispeeds up, as it does with any increase ofengine speed.

Because the pressure drop within the ven-turi is a function of the rate of flow throughit, it might seem that the fuel flow wouldincrease in proportion, maintaining a con-stant air/fuel ratio over a wide range of flow

17

rates. In fact, the ever-increasing pressuredrop at the venturi as the airflow increasesmeans that the air within it becomes increas-

ingly rarified. Meanwhile, the fuel flow, pro-pelled by the pressure difference betweenthe venturi and the atmosphere, increases asthe volume of air increases, but the densityof the air is dropping at the same time. Theresult is that the mixture grows ever richer.A carburetor venturi, it turns out, is a volumemeasuring device, rather than one that mea-sures mass flow. The air bleed provides acorresponding "leak" of air that compen-sates for that phenomenon (indeed, it issometimes called a compensating circuit.)

A carburetor as simple as the one justdescribed would in fact work, and wouldsupply the engine to which it was fitted witha reasonably constant mixture strength overa reasonable range of speeds and loads, butonly if those speeds are constant and moder-ately high. Under two sets of conditions, itwould fall flat on its face.

At very low flow rates, such as at idle, thepressure drop through the venturi is so lowthat the fuel cannot make it out of the end of

the discharge tube. In these circumstancesthe carburetor cannot function at all. To dealwith this, most real world carburetors havea separate idle circuit, in fact a miniaturecarburetor-within-a-carburetor, that com-

18

pletely bypasses the main throat and ven-turi. But we needn't bother ourselves withthe details of these.

The other situation that hobbles our simplecarburetor is changing from one speed toanother. (Actually, slowing down probablywould not present much of a problem;speeding up, however, would.) We haveexplained that air weighs about 0.08 lbs. percubic foot. While not weightless, this is vast-ly less dense than gasoline, at about 43 lbs.per cubic foot. Thus, if an engine shouldsuddenly begin to draw much more airthrough the venturi, the airflow will speedup almost instantly, but the fuel, because ofits greater inertia, will lag behind. As aresult, the mixture will lean out, to the pointwhere the engine will, in all probability,quit.

To deal with this, most carburetors areequipped with an accelerator pump, thatmechanically squirts an extra shot of fuelinto the venturi whenever the throttle is sud-

denly opened. Note, again, that this tenden-cy to run lean on sudden throttle opening isa peculiarity of carburetors. The acceleratorpump exists as much to make up for thisdeficiency as it does to build up the largerquantity of liquid fuel "store" in the mani-fold, as described above.

Despite the fact that there were athousand and one other problemsto be dealt with in the early days

of the automobile, auto pioneers were quickto focus attention on the shortcomings ofcarburetors. The drawbacks included the

possibility of ice forming within the venturiin cold, damp air, and a tendency to vaporlock in hot weather when volatile fuels were

used. Conversely, carburetors were unsuit-able for use with fuels that did not readilyvaporize (remember, the automobile camefirst- gasoline followed, almost anythingthat would bum was used as a fuel at one

time or another in the earliest days). Also,the venturi restricts airflow, and thus power.Because fuel delivery depends on the levelin the float bowl, a carburetor is sensitive tothe inclination of the car, such as whenclimbing hills. In addition, there was somedawning awareness of the problems of feed-ing multiple cylinders from a single "centralmixer," as described in the previous chapter.Accordingly, some of the critics set aboutdesigning an alternative, specifically whatwe now call fuel injection.

Early Production FI SystemsAlthough there were even earlier experi-

ments, perhaps the first production applica-tion of PI was by the German company,Deutz, who built about 300 fuel-injected sta-tionary engines around the turn of the centu-ry. Hampered by a lack of precision manu-facturing facilities, Deutz abandoned thescheme soon after. (Deutz is still in the busi-ness of manufacturing industrial and marinediesel engines.) Perhaps aware of Deutz'swork, Robert Bosch began experiments withfuel injection in 1912. In 1927, Boschbought the Acro company, and the patentrights of an Acro employee, Franz Lang,

who had successfully developed the equip-ment needed for high-pressure fuel injec-tion. Shortly afterward, Bosch began manu-facturing diesel injection equipment of apattern that has changed little in the 70-oddyears since (see sidebar "The Real Jerk").

The adaptation of this system to gasoline-fuelled engines began in the 1930s, initiallyfor use in aircraft. In this application, threeadvantages of fuel injection over carburetorsstand out: freedom from problems of vaporlock, no risk of icing, and a complete indif-ference to the attitude of the vehicle. Vaporlock is a significant problem at high alti-tudes, because the reduced atmosphericpressure there makes it so easy for the fuelto evaporate that it is inclined to boil, form-ing bubbles of vapor in all sorts of awkwardplaces that can effectively prevent any fuelfrom getting to the engine. Carburetor icing,while merely an annoyance in an automo-bile, can be potentially fatal in an aircraft.And a fuel system that can keep the enginerunning no matter what the attitude of theaircraft becomes a matter of first importanceduring extreme maneuvers, such as invertedflight. Most, if not all, Daimler Benz enginesfitted to German aircraft during World WarII enjoyed this advantage over their Alliedopponents.

Early Racing FI SystemsIf we substitute severe cornering for

The straight eight engine ofMercedes Benz's first post-war FormulaOne race car-this streamlined W196-injected fuel directly intoeach cylinder, a systemdeveloped from Diesel injec-tion principles. (MercedesBenz).

19

THE REAL JERKIn a diesel, the cylinder at the end of the compression stroke is filled with nothing but air; combustion cannot

begin until the fuel is injected. Thus, the moment of introduction of the fuel is equivalent to the ignition timing ona spark-ignition gasoline engine. For that reason, control is needed over not just the quantity of fuel injected, butalso the timingof that injection. .

Because the fuel has to be kept separate from the air until the correct instant for ignition, the fuel is injected direct-ly into the combustion chamber- the injector nozzle lies inside the cylinder. As a result of that, the injector isexposed to full cylinder pressure during the power stroke. To prevent combustion pressure from blowing the fuelbackward through the fuel lines, the injector is fitted with a very stiff spring-loaded check valve. This check valvealso performs a number of other important functions. It prevents fuel from dribbling out of the injectors betweentimed squirts, and ensures that the start and end of each injection is clearly defined. Without this valve, the fuelspray would taper off gradually toward the end of each injection. The check valve, in conjunction with similarcheck valves at the pump outlets, also ensures that the fuel lines leading from the injection pump remain filled withfuel at full pressure, to ensure that the pressure pulse from the pump is transmitted immediately to the injector noz-zle. To overcome the resistance of these check valves, the entire system has to operate at a very high pressure-asmuch as 3000psi!

The design of the classic diesel injection pump, and the gasoline injection pumps directly derived from thatdesign, follows from the three requirements: to control the quantity of the fuel delivery, and its timing, and to pro-vide a very high pressure output. In construction, it consists of an approximately rectangular body, bored with anumber of small cylinders-one for each engine cylinder-each of which is lined with a cylindrical sleeve. Thewhole thing thus roughly resembles a miniature engine block, a similarity that extends to the "bottom end" of thepump, which has a shaft running lengthwise through it, analogous to an engine's crankshaft. Instead of crankthrows, however, this shaft has a number of individual carns, one for each cylinder. Each cylinder in the pump con-tains a piston-like plunger that is driven up its sleeve by the corresponding cam lobe as the pump shaft rotates, drivenby the engine at half crank speed, like an ordinary ignition distributor.

The mechanical drive thus provides the timing and the necessary pressure, leaving the problem of control of fuelquantity. To achieve this, each sleeve has a "spill"port drilled through its side, while the upper edge of each plungerhas its side cut away to form a sort of spiral ramp. The sleeve fits in its bore with a very slight clearance, and so isfree to rotate. Rotation of the sleeve in its bore thus covers or exposes the spill port, according to the angular posi-tion of the sleeve relative to the spiral cutaway on the plunger. With the sleeve rotated so that the spill port is cov-ered even when the plunger is at the bottom of its stroke, the only route out of the cylinder is through a hole at thetop, where the fuel lines to the injectors connect. A single stroke of the piston will thus expel the entire volume offuel held in the cylinder; this represents the maximum capacity of the pump and thus corresponds to full power.

At anything less than wide-open throttle, a lesser amount of fuel per revolution is obviously needed, and thisreduction is achieved by rotating the sleeve around, relative to the plunger, so that the spill port lies some distanceabove the plunger's spirally formed top edge. Upward movement of the plunger thus initially causes fuel to be dis-charged from this port, from where it is fed back to the tank, until the plunger rises far enough to close off the port,and injection commences.

The rotation of the sleeve(s) is accomplished by a toothed rack that meshes with gear teeth cut onto the outsideof each sleeve. The rack is connected directly to the throttle linkage, so moving the pedal rotates the sleeves with-in the pump, thus controlling the quantity of fuel delivered with each stroke of the plunger.

Although such a mechanical injection pump is, in fact, quite a simple device, and contains few parts -just twoper engine cylinder served, plus the camshaft and rack- the quantity of fuel delivered per stroke is very small, soall those parts have to be made with extreme precision, involving much hardening, grinding and precision gaug-

20

ing. Accordingly, mechanical injection pumps-known widely and for fairly obvious reasons as "jerk" pumps-are mighty expensive. When adapted for use with gasoline, rather than the kerosene used in diesels, there is theadditional problem that gasoline is a "dry" fuel, lacking the lubricating properties of diesel fuel, thus demandingthe use of extremely hard, wear-resistant alloys, which further adds to the machining difficulty and expense.

The famous Mercedes 300SL "gull-wing" coupe introduced fuel injection to the street. Like the W1.96, a system of direct injection wasused. (Mercedes Benz).

inverted flight, most of the advantages listedare also highly applicable to race cars.Accordingly, by 1937 Mercedes Benz hadtested a single cylinder mockup for a racingengine, using high-pressure injection of fueldirectly into the combustion chamber. In themeantime, a handful of European makers ofsmall, two-stroke engined passenger carshad introduced this "direct" fuel injection, inan attempt to tame the notorious thirst oftwo-strokes that results from something likeone quarter of the air/fuel mixture whistlingstraight out the exhaust ports during theintake/exhaust event, when both inlet("transfer") and exhaust ports are open

together. Because these engines were two-strokes, the injection pump had to run atcrank speed, rather than half-speed, as for afour-stroke. This provided Bosch with valu-able lessons in running their injection equip-ment at high pump rpm. Mercedes's firstpost-war Formula One race engine madedirect use of the experience so gained, boththeir own and that of Bosch.

In both the M196 Formula One enginethat appeared in 1952 and in the engine forthe famous 300SL "gull-wing" sports car,first exhibited in New York in 1954,Mercedes retained the direct (into the com-bustion chamber) injection scheme used in

21

.L~~-..~.

r~~~~.

The location of the injector-screwed right into the cylinder head-is apparent in this cross-section of the 300SL engine. (Daimler-Chrysler Archive)

diesels and in the earlier experiments withgasoline fuel. Combined with suitable injec-tion timing, this allowed radical valve tim-ing and experiments with "tuned" intakepipes, without introducing the problem ofpoor fuel economy from portions of theintake charge being lost out the exhaust.Until injection occurred, after the exhaustvalve had closed, the engine was inhalingonly air.

Diesel Jerk PumpGiventhe existenceof anestablishedtech-

nologyfor injectingfuel,it is hardlysurpris-22

ing that the diesel "jerk" pump was adaptedin this way for gasoline injection. The adap-tation, however, required hurdling a majordifficulty - the problem of regulating thequantity of fuel delivered throughout the fullrange of engine operating speeds and loads.

Unlike gasoline engines where power iscontrolled by closing off ("throttling") theair supply, yet where a very narrow range ofair/fuel ratios must be maintained, on adiesel there is no "throttle" as such. At anygiven speed, the engine inhales the same fullload of fresh air no matter what the positionof the gas pedal; varying the power is sim-

The resemblance to Diesel engine practice is clear in this exterior shot of the 300SL engine. Gasoline's lower lubricity, compared to dieselfuel, compelled the use of expensive, wear-resistant alloys in the injection pump. (Daimler-Chrysler Archive)

ply a mater of injecting more or less fuel.Power is regulated, in other words, byadjusting the mixture strength. In the sim-pler diesel systems, at least, this is directlycontrolled by the driver's right foot-theengine runs extremely lean at light throttle,and rich to the point that it visibly smokeswhen maximum power is demanded. In fact,it is the smoky exhaust that sets the limit forthe rated power of a diesel engine; morepower would be available simply by inject-ing yet more fuel- if we were prepared toput up with the smoke. (Even at full power,the amount of fuel injected falls short of a

stoichiometric mixture-diesels are alwaysrunning "lean," which is one reason they cannever produce as much power as a gasolineengine of the same displacement.)

The inefficient breathing of any pistonengine, gas or diesel, at speeds well awayfrom the rpm at which torque peaks meansthat the mass of air inhaled per revolution, atany given throttle setting, will most definite-ly not be constant across the speed range.Because of the gasoline engine's finickyappetite, the amount of fuel mixed in withthat air also has to be varied on the basis of

the quantity of air inhaled, and not just

23

6 5

7

8

11 12 13 15 1614

4

3 I

d1.Sensoron contoure cam

2 12.Controlrockhead3. Enrichmentsolenoid

1

4. Thermostat

1 5. Barometriccell

6. Checkvalve

7. Plungerunit

8.Toothedsegment9. Controlrack

10. Rollertappet11.Camshaft

12.Governorcontrollever

13. Contouredcam

14. Centrifugalgovernor

15. Idleadjustingscrew16.Shut-offsolenoid

The complexity of the mechanical controls for the Bosch system used on the 300SL is mind-boggling. Fuel delivery of the pump (only thecamshaft is shown, 1.1.)is controlled by the rack (9). The position of the rack is governed by engine speed, via the centrifugal weights(1.4) acting on a contoured cam (1.3), and by throttle position. Other factors are accounted for by a barometric capsule (5) and a tem-perature sensitive device (4). (Robert Bosch Corporation)

according to speed and throttle position.Adapting to Gas Engines-To adapt a

diesel jerk pump to gasoline operation, then,means that the rack (see sidebar, page 20)cannot simply be directly hooked up to theloud pedal; the relationship between the twohas to be modulated by some other con-troles). The usual way this is done is byadding a set of centrifugal weights, some-what like those in a traditional ignition dis-tributor, to provide a signal proportional toengine speed, plus a diaphragm that "reads"

24

manifold vacuum, a fairly close approxima-tion of engine load. These additional controldevices are connected to the linkage control-ling the rack via a system of cams and links,with the shape of the cams tailored accord-ing to the idiosyncrasies of the engine'sbreathing.

The monkey motion of these additionallevels of control added even more to the

substantial cost of the pump itself, demand-ed meticulous adjustment, and its complexi-ty mocked the essential simplicity of the

basic pump. Indeed, one observer at the timemused aloud that if all engines had wornfuel injection all along, then someoneinvented the carburetor, he would have been

hailed as a genius.Injector Location-Apart from the

mechanical complexity of the "add-on"speed- and load-sensing mechanisms at thepump, another problem confronted thisadaptation from diesel to gasoline - that ofshielding the injector nozzles from the heatand fury of combustion. Surprising as it mayseem, gasoline combustion chambers seehigher peak temperatures than diesels. Ontheir Ml96 Formula One engine, Mercedesdealt with this by fitting the injector into theside of the cylinder, where it was masked bythe piston at TDC and thus shielded from theworst of the inferno.

Certainly, timed direct injection deals withthe potential issue of fuel consumption,especially with radical valve timing, and thehigh pressure spray of these systems isfavorable for atomization, but against theseadvantages were set the problems notedabove, plus the issue of the power requiredto drive a pump that has enough grunt toblow open the stiff check valves in the injec-tors and at the pump outlets, used respec-tively to prevent combustion pressure fromblowing back through the injection lines andto keep the lines fully charged.Nevertheless, the fact that engines operatequite happily on carburetors made it clearfrom the outset that there was no absolute

need for precision timing of the fuel deliveryin a spark-ignition engine.

Granted, a carbureted engine would prob-ably make a little less power than a similarone with direct injection, but some part ofthis might be attributed to the fact that a sep-arate injector for each cylinder eliminatesthe problems of a central mixer, rather thanto the timing. Of course, the same thing canbe achieved with carburetors if a separatecarb is provided for each cylinder. That,however, would be heavier, probably more

expensive and arguably even more complex,depending on the number of cylinders. Thus,while a handful of other Formula One teams

and upmarket European sports car manufac-turers also used Bosch gasoline injection, tothe best of this writer's knowledge, all theseothers left the injector just outside the com-bustion chamber, aimed at, and sprayingagainst, the head of the intake valve.

But if you decide to locate the injectorsoutside the combustion chamber, then youdon't need to fight combustion pressure, soyou don't need a very high pressure pump.Also, if the injector lies outside the cylinder,then you don't need to time the injection soit occurs with both valves closed. In that

case, why bother to time it at all? And if thetiming of discrete squirts is not crucial, thenwhy does the fuel have to be delivered inindividual pulses?

Early ContinuousInjection Systems

These questions had occurred to a good

many Eeople, and an alternate form of fuelinjection came into being. In these, fuel isinjected continuously, rather than in pulses,and at very much lower pressure than a"jerk" pump provides, though still muchhigher than the small pressure differencethat exists between a carburetor venturi and

the atmosphere.The Wright brothers used a primitive form

of this continuous injection, as the arrange-ment is called, on the engine of their firstsuccessful aircraft. An engine-driven pumpof the "positive displacement" type-onewhich supplies a fixed quantity of fuel ateach revolution - was used, to continuouslyspray fuel directly into the engine air intake.It should be clear that while a satisfactorymixture strength might be arranged at fullthrottle-presumably after some tinkeringwith the size and/or drive ratio for the

pump - with no means to reduce the pumpoutput the mixture must have become hope-lessly rich at anything less than full throttle.

25

Although Stu Hilborn had proved, years earlier, that such a system could work on the track, Rochester were the first to demonstrate thepracticality of the simpler, continuous flow fuel injection system, as on this early "fuelie" Corvette. (Dave Emanuel)

But then the Wrights' had no immediateinterest in operation at part throttle!

Winfield's F] System-About the sametime that Bosch and Mercedes were begin-ning their experiments with timed, high-pressure, direct gasoline injection in themid-l930s, Ed Winfield developed a lowpressure, continuous injection system,intended for Indy cars. Winfield's design hadprovision for varying the fuel flow withthrottle position, but a lack of interest frompotential customers forced him to eventual-ly abandon the scheme, and he allowed hispatents to lapse. It was not until fifty yearsafter the Wrights that racers had access to acontinuous injection system that dealt withthe part-throttle problem.

Hilborn ]njection-Stu Hilborn achievedthis by incorporating a controllable spillvalve, connected to the throttle linkage, that

26

tapped off (and returned to the tank) someportion of the pump's output, according tohow wide the throttle was opened.

On the face of it, this might appear to fillthe bill, but the assumption built-in here isthat an engine, at any given throttle setting,needs a fixed quantity of fuel per crank rev-olution, and we have already seen that this ismistaken, because of the way an engine'sability to take a full breath varies across itsspeed range. As a result, gasoline PI systemsthat use just engine speed and throttle posi-tion to determine fuel quantity can bearranged to provide a suitable mixturestrength only under very limited circum-stances, and must inevitably miss the markby a large margin under others.

Still, schemes such as Hilborn's have along and honorable history when used withmethanol fuel. As discussed in more detail

in Chapter 8, methanol will tolerate widevariations in mixture strength with littleeffect on power output, which helps maskthe inherent shortcomings of such a simplesystem. In this case, rpm and throttle posi-tion alone provide a sufficiently accuratemeasure of fuel flow requirements, so it isnot necessary to measure the rate of airflowat all. Since there is no need to measure it,there is no need for the restriction of a ven-

turi or any other airflow sensor stuck in theairstream. This improves engine breathingand so promises more power just for thatreason alone.

RochesterFISysternFor a street engine running on gasoline,

however, any purely mechanical system,whether injecting continuously or in a seriesof pulses, needs to somehow measure orcompute the rate of airflow into the engine,and to use that information to modulate the

delivery of fuel so as to keep the air/fuelratio appropriate to the circumstances. Wehave discussed how this can be achieved

with a jerk pump, but what about a continu-ous gasoline injection system?

There have been many such systemsthroughout automotive history; one of themore recent and more widely used of theseis the Bosch K-Jetronic system, described inChapter 6. Nevertheless, the example mostof us remember best is likely the Rochestersystem used on "fuelie" Corvettes from1957 to 1965. The heart of the Rochester

system is a spill valve that serves as a regu-lator of fuel quantity- indeed, it is the onlymixture control element in the whole sys-tem. Fuel is fed to the spill valve by anengine-driven gear-type pump that simplyserves as a source of fuel at medium-highpressure (200psi at maximum), with thepressure tending to rise with increasing rpm.The spill valve consists of a sleeve with anumber of ports arranged radially in its side,plus a plunger that covers or exposes theseports, according to its position. The position

of the plunger, in turn, depends on a balancebetween the fuel pressure and a forceapplied to the top of the plunger by a mix-ture control mechanism.

Increasing fuel pressure tends to raise theplunger, thus uncovering the sleeve portsand so bleeding off some of the pressure.The mixture control device, shown schemat-ically in the following illustration, amountsto a simple mechanical computer thatapplies a spring force to the top of theplunger, modified by the forces acting ontwo diaphragms - one connected to theintake manifold, and one to a venturi

through which all the air entering the enginehas to pass. The eight individual port injec-tors downstream of the spill valve thus see afuel pressure that depends on a balancebetween the supply pressure and the forcesacting on the diaphragms, and their outputvaries accordingly.

The Rochester system is strikingly simple,especially when compared to a jerk pumphaving the controls needed for street use ongasoline, and none of its parts require excep-tional precision or expensive materials fortheir manufacture. Although the injectorsthemselves had to be individually calibratedand installed as a matched set, given suffi-ciently large production numbers it seemslikely that the system overall would havebeen little (if any) more expensive to makethan a pair of four barrel carburetors.Nevertheless, it remained a somewhat rare

and costly performance option, and justifiedits price premium by offering more power.Partly this was because the near-eliminationof cylinder-to-cylinder variations in mixturestrength permitted a higher compressionratio (CR) than was safe with carburetors,but an advantage still showed even when theCR was unchanged. We might attribute thisto three sources: nearly identical air/fuelratios at each cylinder, elimination of mani-fold heating, and a reduction in the breathingrestriction imposed by the venturi, com-pared with a carburetor. Although a venturi

27

IUUI

<::~~\ \\\I \ \ \

,--.---.----.

-----.--

MANIFOLD VACUUM

VENTURIVACUUM

OUTPUT TOINJECTORS

The heart of the Rochester system's control unit was a spill valve that returned to the tank a certain fraction of the fuel delivered by thegear-type pump. The position of the piston in the spill valve was determined by two diaphragms-one hooked to a manifold vacuum, oneto a single large metering venturi that fed the air box. A system of links and rollers linked movement of the diaphragms to the spill valvepiston.

was needed to measure air quantity, theamount of pressure drop needed to provide auseful signal to the mixture control unit wasmuch less than that required to lift gasolineout of a carburetor's discharge tube.

In its day, the Rochester FI was regardedby the auto industry and the public alike asvery much a high-tech piece, and possiblythe final word in fuel delivery. Ironically, atabout the same time that OM was introduc-

ing this system, another completely differentform of fuel injection was being unveiled, tomuch less public acclaim, yet this one would

28

eventually sweep away not only the carbu-retor, but most other forms of fuel injectionas well.

The First Use of Solenoid Valves

The basis of this little-hailed breakthroughwas the use of an electrically operated sole-noid valve as a fuel injector. Supplied withfuel at a modest constant pressure, it squirtsat a constant rate as long as an electric cur-rent is applied to the solenoid windings, andstops when the current is turned off. Thus,the quantity of fuel injected depends simply

on how long the juice is turned on. If oneelectrical pulse is delivered every secondrevolution of the crankshaft (for a four-stroke engine), then the air/fuel ratio of themixture delivered to the engine can be con-trolled simply by varying the length of eachpulse.

The idea of using a solenoid valve as a fuelinjector is a surprisingly old one. Such ascheme had been developed by an engineernamed Kennedy at the Atlas Imperial dieselEngine Co, in 1932, and an installationappeared on a marine engine exhibited at theNew York Motor Boat Show in 1933,around the same time that Bosch and

Mercedes in Germany, and Winfield in theU.S. were beginning their experiments.Given the primitive state of the electricalsciences back then, it is mildly astonishingthat the thing worked at all, yet in 1934a .

similar but smaller engine was installed in atruck and driven successfully from LosAngeles to New York and back.

The "Electrojector"-A quarter centurylater, A.H. Winkler and RW. Sutton of theEclipse Machine Division, Bendix AviationCorporation, announced the fruits of fouryears' work at Bendix to develop ann FI sys-tem for gasoline-fuelled automobiles usingsuch a solenoid-controlled valve as an injec-tor, in conjunction with an electronic controlunit, or (as they termed it) "brain box." Theydescribed this combination of componentsas "electronic fuel injection"-surely thefirst time this phrase was ever used. Theresulting "Bendix Electrojector" compriseda separate port injector for each cylinder,each supplied with fuel from a common railat a regulated 20 psi. Sensors responsive tomanifold pressure, engine speed, atmos-pheric pressure, and air and coolant temper-ature fed their various outputs to the "brainbox" - the electronic control unit (ECU)-which then calculated the instantaneous fuel

requirements of the engine. On the basis ofthat calculation, the ECU delivered a pulseof current to a rotary distributor which fed

that current pulse to each solenoid valve inturn, causing each successively to open andsupply fuel to the cylinder it served. Thequantity of fuel delivered by each injectionwas simply a matter of the duration of thepulse-as long as it was "on," the injectorwould deliver fuel at a constant rate. At idle,the pulses were of very short duration; atmaximum load, the injectors were open foras much as 150 crankshaft degrees.

At the beginning of the development ofthe Electrojector, about 1951 or 1952, thetriggering signal for the ECU was providedby a mechanical commutator-a simplewiper making intermittent contact with asegmented brass ring, as the engine-drivenring swept past. That soon gave way to trig-gering by an extra set of contact points in ahousing sandwiched under the ignition dis-tributor. In at least the three earliest versions,the working bits in the ECU were. . . vacu-um tubes! One such unit was installed on the

V8 engine of a 1953 Buick, and used todemonstrate the system to the auto industry.

By 1957, when it became available to thepublic in very limited numbers, the electron-ics were, thankfully, all transistorized. Thiseliminated the need to wait for the tubes to

warm up before the system became func-tional, shrank the size of the ECU and sub-

stantially reduced the electrical powerrequired to run it, and doubtless greatlyimproved the system reliability - vacuumtubes do not take kindly to be rattled around!The Electrojector was listed by AmericanMotors as an option for the 1957 RamblerRebel, but very few cars so equipped wereever sold. A year later, Chrysler offered it onthe 300D and DeSoto Adventurer, and sev-eral hundred with the option were manufac-tured in 1958 and 1959.

Despite occasional problems from exter-nal electrical fields causing the brain box togo haywire, and a degree of unreliability ofthe injectors themselves, on the whole thesystem worked satisfactorily and matchedBendix's claims of modest (5 percent)

29

improvements in both fuel economy andpower, compared to a similar engine fittedwith a pair of four-barrel carburetors. Theteething troubles were eventually workedout, and by 1965 Bendix had a fully devel-oped production version ready to go.

At this point, rather than try to produceand market the Electrojector themselves,Bendix decided instead to enter into an

agreement with the Robert Bosch Co.,which gave Bosch access to the Bendixpatents. From a business point of view, thismade a good deal of sense. In view of theirreputation in the auto industry as an estab-lished OEM supplier-one with wide expe-rience both in fuel injection and in electricaland electronic equipment-Bosch wasclearly in a better position to exploit thepotential market. And fuel prices in Europewere many times what they were in the U.S.of the 1950s,justifying a much higher initialcost there, in exchange for the fuel economybenefits alone.

Bosch Jetronic FIIn most respects, the Bosch "Jetronic" FI

that first appeared on the 1967 Volkswagen1600, was functionally identical to theElectrojector system described by Winklerand Sutton ten years earlier. Apart from tak-ing advantage of advances in electronics toproduce a more compact ECU, this sys-tem-since identified as D-Jetronic, to dis-tinguish it from later Jetronic versions-dif-fered from the Bendix prototype in onerespect. In the Electrojector, the housing thatwas piggybacked onto the ignition distribu-tor and which contained the triggering con-tact points also contained a rotor, just like anignition rotor; this distributed the timedpulse to each injector sequentially. In the D-Jetronic as applied to the four cylinder VW,there was no rotor, but there were two sets of

triggering contacts, one for each pair ofcylinders. Each set of contact points drovetwo injectors simultaneously, timed so thatone cylinder received its injection while its

30

intake valve was open; for the other cylin-der, the injection occurred with valvesclosed. Fuel thus accumulated in the port,awaiting that cylinder's next intake stroke.

We have mentioned earlier that the mere

fact that a carbureted engine will function ina satisfactory way should make it obviousthat injection timing is, to say the least, notcritical, and the power gains realized with afew continuous injection systems tended toconfirm this view. The matter was prettymuch clinched in 1959, when Mercedes suc-cessfully used a Bosch jerk pump with justtwo plungers, each feeding three enginecylinders, on its six cylinder 220SE. Thismust all have improved the comfort level ofthe Bosch and VW engineers who adoptedthe hop-skip timing used on the D-Jetronic,an arrangement that allowed eliminating therotor used in the Electrojector.

The Electrojector-based, intermittentinjection "Jetronic" systems from Boschhave undergone successive improvementsand refinements over the years since the D-Jetronic's introduction. Apart from the elec-tronic innards of the control unit gainingcomputing power while shrinking ever fur-ther in size and weight, and the use of all-electronic triggering (rather than contactpoints), the major differences between onevariant and another have been in the way thequantity of air supplied to the engine is mea-sured. These basic differences between sys-tems are reflected in the nomenclature used:

L-Jetronic, LH-Jetronic, etc. In addition,most later systems accept inputs from addi-tional sensors, most notably an oxygen sen-sor, also called an °2 sensor or lambda sen-sor. As mentioned in the previous chapter,this is a device fitted to the exhaust systemthat" sniffs" the exhaust gas to gauge itsoxygen content, thereby providing a directmeasure of what the air/fuel ratio actually isat any given moment, rather than obligingthe system to depend totally on a set of pre-programmed "maps" or "look-up tables"that predict what the ratio should be for a

With the rear-mounted "pancake" engine nestled under the apparent floor of the trunk, earlier carburetted versions of the VW 1500/1600were notorious for vapor lock problems. This is the later 1600 engine with Bosch D-Jetronic injection. The ECU is visible at the far left.(VW Canada).

given set of operating conditions. (See alsothe sidebar "The Oxygen Battery.")

In many cases, control over ignition tim-ing has been integrated into the ECU ofthese systems, in which case they are termed"Motronic," again with variants according tothe method of air metering. Over time, anincreasing number of functions have beencombined into the Motronic system, includ-ing integrated control over automatic trans-mission shift points and torque converterlockup, the waste-gate on some turbocharg-er-equipped cars, selection among differentlength intake ports on some others, andauthority over the variable valve lift and/ortiming found on a few vehicles. We discussthese Motronic systems in some detail inChapter 4.

K-]etronic-At the start of the new mille-

nium, the principle of intermittent injectionusing electronically controlled, electricallypowered solenoid injectors is now universalon domestic passenger cars, and near-uni-versal on models imported from Europe andAsia. Yet not everyone was an immediateconvert to this plan. It seems ironic now thateven after the pulsed (intermittent) Jetronicsystem was commonplace,Bosch felt obligedto introducea continuousinjectionsystem,theK-Jetronic (The "K" stands for kontinuier-lich, German for continuous). Early versionsof this were purely mechanical; in responseto ever-tightening emissions standards, laterones-such as KE-Jetronic and correspond-ing Motronicversions- added a degree ofelectronic control. In part this apparent pref-erence for an all-mechanical system may

have been due to a powerful conservative

31

Intermittent electronic fuel injection allows modern cars to combine high powerwith driveability, economy, and low emissions. (VW Canada).

32

streakamongsomemanufacturersand theirengineers.In part, too, it may have been amatter of economics-the intermittentsys-tems may have been more expensiveat theOEM level. Although there were a few hold-outs until the early 1990s, most of the vari-ous K-Jetronic systems were history by themid 1980s. These now-superseded (not tosay obsolete) systems are dealt with inChapter 6.

We will not be dealing with a handful ofsub-variants of Jetronic and Motronic PI that

are used on vehicles not imported into NorthAmerica.

To summarize, then, the essential differ-

ence between carburetors and fuel injectionsystems is that a carburetor uses the pressuredrop within its venturi both to measure theairflow and as the means to propel the fuelinto the intake air stream, whereas PImechanically forces the fuel into the intakeair in an amount usually determined by aseparate airflow or manifold pressure sen-sor, acting on an electronic or mechanicalsystem that actually controls the quantity offuel injected.

Basic Types of Fuel InjectionAll fuel injectionsystemscan be divided

up intotwobasictypes:thosethatinjectfuel

continuously and those that inject intermit-tently, in a series of pulses. Again, in eithercase control over how much fuel is injectedcan be achieved either mechanically or elec-tronically. In the present context, only theold, diesel-type jerk pump uses mechanicalcontrol over intermittent injection; all Boschintermittent systems use electronic control.Bosch K-Jetronic continuous injection sys-tems-sometimes called CIS-may usepurely mechanical control, as in the originalK-Jetronic, or may include some degree ofelectronic fine-tuning on top of the basicmechanical computation.

Many FI applications, especially in newercars and in older high-performance ones,use a separate injector for each cylinder.This greatly reduces the quantity of liquidfuel laying on the walls of the ports. If theinjectors are located very close to the intakevalves and the intermittent injections aresequential, the possibility of maldistributionof the mixture between one cylinder andanother is essentially eliminated.

Many others, though, are of the "throttlebody" type, in which the fuel is sprayed intoand mixed with the air in one centralized

location, then conducted to the cylindersthrough a thus "wet" manifold, just like thatof a carbureted engine. This would seem toput it on the same footing as a single carb interms of mixture distribution, but the fact

that the fuel is injected under pressuremeans that it emerges from the injector noz-zles as a very fine spray, with droplets so

small that they have a much better chance ofevaporating than is often the case with thelumpy soup coming from a carburetor.While a TBI system is thus likely to be bet-ter in this regard than a carb, an individualinjector at each intake port can do even bet-ter.

One of the further advantages sometimesargued for FI is that it achieves the maxi-mum possible engine power because it doesaway with the venturi. That is not quite true,because it is a basic principle of science thatyou cannot measure something withoutaffecting the thing you are measuring.Certainly a venturi measures airflow, and inthe process it also restricts that flow, butelectronic fuel injection (EFI) systems thatuse a spring-loaded flap valve or a "hotwireIIairflow sensor (the current needed tokeep the wire at a constant temperature isused as the air-quantity signal) both offer atleast some restriction to flow, too, althougha hot wire (or hot film) sensor will impedeairflow much less than any carburetor ven-turi.

The ultimate sophistication is to providean oxygen sensor (°2 sensor, lambda sen-sor) in the exhaust pipe and to correct themixture strength from moment to moment,based on that information; an excess of °2calls for a richer mixture, and vice versa. Forall these reasons, EFI with an injector foreach cylinder is unarguably the state-of-the-art for fuel-air mixing and delivery.

33

34

Rich mixtures, we explained in Chapter 1, mean that some of the hydrocarbon molecules in the fuel enteringthe engine pass out into the exhaust unreacted - there simply aren't enough oxygen molecules available for themto combine with. These unburned hydrocarbons are a source of air pollution, and legislation severely limits thequantity of them that is allowed to escape into the atmosphere. The shortage of oxygen also means that the reac-tion with the hydrocarbons that do bum is chemically incomplete-instead of the hydrocarbon-oxygen pairingsforming CO2 and H2O, many of them wind up as CO-carbon monoxide. Emissions of carbon monoxide arealso strictly controlled by law.

Lean mixtures would keep down the CO levels, and as long as they were just a little lean, the HC levels, too.Ironically, excessively lean mixtures actually raise HC levels above those found with a stoichiometric mixture,becauseas the mixtureapproachesthe "leanlimit"- the air/fuelratio atwhichthe mixturesimplywon'tignite-some cylinders misfire, so all of the HC in that particular cylinderful goes straight out the exhaust unburned.

But even mixtures just lean enough to minimize CO and HC create high combustion temperatures. Too-hotcombustion encourages some of the nitrogen (N) in the air, normally thought of as inert, to combine with someof the excess oxygen, to form various oxides of nitrogen-NOx. NOx is also a pollutant-it is the principlesource of photochemical" smog" - so the amount of it, too, is limited by law. (Damned if you do; damned if youdon't!) A compromise settingwon't work, either. Anythinglean enough to meet HC and COlimits would bust the NOx limit,and anything rich enough toavoid excessive NOx would

lead to illegal levels of HC andCO. (Damned even if you sit onthe fence!)

The solution arrived at by automakers, beginning about 1975,has been the "three-way catalyt-ic converter." In essence, thisdevice splits the oxygen fromthe NOx and helps it combinewith the CO to make CO2, andwith the HC to make H2O, andmore CO2' This works wellenough to satisfy the G-men. . .provided that the mixture wasneither too rich nor too lean in

the first place. And the three-way catalytic converter ismighty fussy indeed about just how Constructionof a three-waycatalyticconvertor.Convertorsareessential.

h . .. to meet currentemissions limits,but can onlydo theirjObif the air/fuel mixtureclose to the stOIC lOmetnc ldeal the fedto the engineis heldveryclose to the chemicallycorrectratio.Andto achieveingoing mixture has to be; mixtures that, a lambda(oxygen)sensor is essential. (RobertBosch Corporation).that are less than one percent either

1

2

3

1.Ceramicmaterialwith catalyticallyactivesubstances2.Steelwaolretainer

3.Housing

1. Probehousing2. Ceramicshield

3. Electricalconnections

4. Shieldtubew/slits

5. Activeceramicsensorlayer6.Contact7.Shield

1 2 3

4 5 6 7 9 10

Lambda sensors have a voltage output that rises from near zero in the absence of oxygen in the engine exhaust toa bit less than one volt when small traces of it are present. This characteristic enables an electronic fuel injection systemto control fuel delivery on the basis of the sensor's reading of the exhaust gas. The sensor only works when hot, however.To reduce the time between engine start-up from cold and the sensor reaching its operating temperature, some versions areelectrically heated. (Robert Bosch Corporation).

side of the chemical ideal render the converter next to useless.

No fuel injection system yet developed (and certainly no carburetor!) can hope to control the mixture strengththat closely, based simply on a set of values derived from the characteristics of some similar engine in a test lab.The device that enables EFI to "know" just what the mixture strength needs to be at any instant for any particu-lar engine out in the real world is the "lambda sensor."

In construction, the lambda sensor consists of a hollow ceramic bulb, with both inner and outer surfaces cov-ered with a microthin layer of platinum. For protection, the outer surface is surrounded by a shield. This bulb isinserted into the exhaust system so that its exterior surface is washed by the exhaust gas stream, while its hollowinterior is exposed to the atmosphere. In operation, the lambda sensor acts like an oxygen-powered battery: a dif-ference in the oxygen level of the gasses surrounding the two surfaces of the sensor causes it to generate a smallvoltage.

As you might expect, there is very little oxygen in the exhaust of an engine running on a stoichiometric mix-ture, yet there is some-about 0.5 percent. Surprisingly, even with a mixture that is about five percent rich, thereis still about 0.2 percent oxygen. Equally surprising, a mixture that is five percent lean will only contain aboutone percent oxygen. A particularly useful feature of the lambda sensor is that its output voltage changes very dra-matically right around the point of stoichiometry; very small changes (a few tenths of one percent) in oxygen

content lead to large changes in the sensor output. While the sensor will put out about 900 millivolts (0.9v)

when the exhaust oxygen content corresponds to a lambda of about 1.2 (that is, the mixture is about twenty per-cent lean-an air/fuel ratio of a bit less than 12:1), and while its output drops to near zero at about lambda=0.8(an air/fuel ratio of near 18:1), almost all of that voltage change occurs within one percent of stoichiometry. Thevoltage output of the sensor is fed back to the ECD, which slightly reduces or increases the duration of the ontime for the injectors, according to whether the sensor reports the mixture is richer or leaner than stoichiometric.This "closed-loop" control maintains the air/fuel ratio within 0.1 percent of the chemical ideal, thus allowing theconverter to do its job.

In practice, an output from the sensor of 450mv is used by the ECD as the threshold of the lean limit and 500mvas the rich limit. This means that the air/fuel ratio actually oscillates back and forth (though always within 0.1percent of stoichiometry) as the ECD responds to the voltage signal from the sensor. The time required for thesensor output voltage to vary between these limits depends on its temperature. At idle, when the gas temperaturesare low-say 3000 C (about 5700 F)-that time interval might be as much as a couple of seconds; under typicalroad loads, the exhaust temperature is more like 6000 C (11000F), and the "cycle time" is reduced to less than 50milliseconds (1120second).

The sensor is essentially inoperative at temperatures below 3000 C, so during engine startup and initial warm-up the ECD ignores the sensor voltage and falls back on a set of internal "maps," until such time as the signals itis getting from the sensor make sense. Accordingly, emissions are inevitably increased during this warm-up peri-od, so to reduce the wait for the sensor to reach operating temperature, some sensors have an internal heating ele-ment to bring it on-line as quickly as possible.

36

-

A~noted in the previous chapter, thefirst production application of

osch electronic fuel injectionwas the version installed on the Volkswagentype 3 1600, in 1967, for the 1968 fastbackand Variant (station wagon) models.Although the prototype for this system- theBendix Electrojector - was introduced adecade earlier, it was a very rarely encoun-tered option, on just a couple of uncommonand expensive models. By contrast, that firstBosch/VW system was standard equipmenton a low-priced, mass production car, and assuch is rightly regarded as something of amilestone in automotive history.

D- JetronicAt the time of its introduction, this system

was simply identified by Bosch as"ECGI"- Electronically Controlled Gas-oline Injection; it was soon relabeled"Jetronic." To distinguish it from subsequentversions, it has since become known as the"D-Jetronic." Here, the "D" stands fordruck-German for pressure-because thecalculation of the quantity of fuel to beinjected was based on manifold pressure.After the 1968 VW, the D-Jetronic was alsoused on the Porsche 914, and a number ofMercedes, Volvo and Saab models, in somecases until 1975. The last cars to be fitted

with this system were the 1975 Mercedes450 and the Volvo 164E of the same year.Doubtless many thousands remain in usetoday.

Mechanical ComponentsThe mechanical parts of the system

includethe fuel pump, filter,pressureregu-lator,fuel linesand injectors.An electrically

1

4

1.Fueltank

2.Electricfuelpump3.Fuelfilter

4. Fuelrail

5.Fuelpressureregulator

6.Fuelinjectionvalve7. Coldstart valve

driven roller-type rotary pump draws fuel,through a filter, from the fuel tank and deliv-ers it to a "common rail" fuel line supplyingthe injectors. Pressure in the fuel rail ismaintained at a constant 28-32psi by a reg-ulator which functions by bleeding offexcess fuel and allowing it to flow back tothe tank via an essentially unpressurizedreturn line. The regulator is thus located,schematically at least, at the opposite end ofthe fuel rail from the pump.

The need for tight control of fuel pressureshould be apparent from the system descrip-tion in the previous chapter: The systemcontrols fuel quantity by varying the "on-time," or "pulse time" of the injectors, butthe rate of fuel flow through an injector thatis "on" is a direct function of the supply

Schematic layout of the fuelsupply components of a typi-cal Bosch intermittent injec-tion system. (Robert BoschCorporation)

37

pressure. For the on-time to be the control-ling variable, the supply pressure must befixed within narrow limits.

Each injector is installed near the intakevalve it serves. While all D-Jetronic installa-

tions thus employ a number of injectorsequal to the number of engine cylinders, theinjectors do not "fire" individually, butrather in groups of two, as mentioned in theprevious chapter. Thus, on four-cylinderengines there are two groups of two, onsixes two groups of three, and on eightsthere are two groups of four.

Each electromagnetic injector consists of avalve assembly with a needle valve thatopens a single orifice nozzle. The oppositeend of the needle valve is attached to the

core of the solenoid. When no current is sup-plied, the needle is pressed onto its seat by aspring. When current is supplied, the needleis pulled inward very slightly (about 0.006inch), and fuel flows out through the cali-brated nozzle opening, past a protruding tipat the point of the needle that aids atomiza-tion of the fuel. The action is very much likethat of an ordinary paint spray gun.

Electrical ComponentsThe electronic components of the system

include a pair of contact points, a manifoldpressure sensor, an air temperature sensorand an engine temperature sensor. There isalso a "throttle position sensor," although inearly versions this consists simply of twomicroswitches that detect just full throttleand closed (idle) throttle. All these compo-nents feed signals to the electronic controlunit (ECU), which contains a couple of hun-dred discreet transistors, diodes, and capaci-tors (condensors) mounted on a circuitboard, and all contained in a metal box aboutthe size of a large book.

The heart of the ECU is a device known inelectronics as a multivibrator. This is not

actually a single component, but rather a cir-cuit comprising a number of components,and having the characteristic that, once trig-

38

gered, it will turn on an electric current for acertain length of time. (There is one of thesein your windshield wiper "intermittent" con-trol). The actual duration of that time inter-val depends on the value of the componentsin the circuit. Thus by arranging for the elec-trical characteristics of one or more compo-nents to be variable, the on-time followingeach triggering can be altered. (We willpostpone describing the internal logic of theECU until we deal with the L-Jetronic sys-tem, below.)

It seems that the very earliest versions ofD-Jetronic achieved the mixture enrichment

needed for cold starting simply by provid-ing, in response to information from theengine temperature sensor, a longer durationpulse for the injectors. Although the historyhas been somewhat difficult to reconstruct,

it also appears that this proved inadequatefor some areas in North America, shortlyafter cars fitted with this early equipmentstarting arriving on these shores in 1969.What is certain is that a factory "cold startkit" was made available, to be retrofitted tothese early units. The heart of this kit was an

extra injector, fitted onto and feeding intothe intake manifold, that opened and stayedopen continuously while the engine wasbeing cranked. This so-called "fifth injector"(the name stuck, even when there were six

or eight injectors, for equally numerousengine cylinders) subsequently became anintegral part of the system.

Drawbacks

While it was in general an effective andreliable system, the D-Jetronic was not with-out its shortcomings. One of these was themechanical triggering contacts. Now, amechanical device operating in conjunctionwith any electronic device often proves to bea weak spot, but because the triggering con-tacts for the PI only carried a tiny current,arcing of these points was not the problem itis for a traditional ignition distributor. Therenevertheless remained the problem of wear

of the rubbing block on the moveable con-tact. While these contact points were consid-ered good for 100,000 miles-essentiallythe life of the vehicle- they were an occa-sional source of grief.

Early versions of D-Jetronic exhibitedanother occasional source of complaint(more an annoyance than a serious defect): avery brief hesitation on sudden throttleopening at low rpm. We can speculate thatthis resulted from the absence of any directprovision for acceleration enrichment. Laterversions had a more elaborate throttle posi-tion sensor that enabled the ECU to detect

sudden throttle opening and to adjust thepulse time appropriately.

Somewhat surprisingly, the manifold pres-sure sensing principle itself was also eventu-ally regarded by Bosch as a shortcoming.On the face of it, manifold absolute pressure(MAP) appears to be a valid measure ofengine load, and thus of fuel quantityrequirements, and one having inherent cor-rection for variations in air temperature andatmospheric pressure. There are a couple ofpoints about which we might speculate.First, to measure manifold absolute pressureobviously requires that there be a manifold!Manufacturers wishing to make full use ofacoustic ramming (see Chapter 8) mightprefer to have the intake pipe for each cylin-der as completely separate as possible fromall the others. And when multiple intakepipes are joined together in a common man-ifold, the pressure fluctuations within thoseindividual pipes are likely to cause oscilla-tions in the manifold pressure, especially onfour cylinder engines. Apart from thesedesign considerations, time revealed that theD-Jetronic was very sensitive to engine con-dition. An aging engine with poorly sealingvalves could confound the calculations ofthe ECU.

L-JetronicMany of the shortcomings of the

D-J etronic system were answered in

the L-Jetronic, introduced in 1974, and first

appearing on the Porsche 914 of that modelyear. Although there have been numeroussubsequent versions of EFI manufactured byBosch under the "Jetronic" label, the

L-Jetronic established the principle of cal-culating the fuel quantity required- andthus the pulse time of the injectors-on thebasis of airflow, rather than manifold pres-sure as on the D-Jetronic. This fundamental

difference is reflected in the designation-the "L" stands for luft, the German word foraIr.

Components of the L-Jetronicsystem. (Robert BoschCorporation)

39

1.Throfflevalve

2. Airflowsensor

3. Intakeair temperaturesignalto the ECU4. ECU

5. Airflowsensorsignalto the ECU6.Airfilter 4

5L 3

6

(Q)~ I Inductedairquantity

21

Airflow sensor in the intake system. (Robert Bosch Corporation)

Airflow Sensor

The airflow sensor is a box through whichall of the intake air has to pass. Within thebox is a spring-loaded vane or flap, hingedat the top, which is deflected from the down(closed) position by the momentum of theair passing through the sensor. Attached tothe vane's hinge is a potentiometer-a vari-able resistor, like a radio volume control.Movement of the vane causes an electrical

contact in the potentiometer to wipe arounda ring-shaped surface that is wound with aresistance wire, so that the total resistance of

the potentiometer depends on the position ofthe wiper, and thus of the vane. By measur-ing this resistance value, the ECU is thusable to "know" the position of the vane andthus the rate of airflow through the meter.

The internal "floor" of the box is shaped sothat, in combination with the weight of thevane and the calibration of its return spring,the angle of the vane is proportional to the

40

- -

logarithm of the airflow rate. That is, a dou-bling of the flap angle corresponds to a ten-fold increase in airflow. The sensitivity ofthe airflow meter is thus greatest at smallvalues of airflow, when the greatest meter-ing precision is needed.

As mentioned in Chapter 1, the openingand closing of an engine's valves causespressure fluctuations in the intake air stream.To reduce the tendency of these fluctuationsto cause the vane to oscillate or flutter, a sec-ond flap is arranged at ninety degrees to thefIrst. Pressure pulsations act on both vanes atthe same time; a pulse of pressure that wouldswivel one of the vanes in one direction has

the opposite effect on the other vane, so thetwo forces tend to be self-cancelling. Thissecond balance (or damping) vane also hasanother function. As the measuring vaneopens, the balance vane moves into a closelysurrounding space, so the tendency of themeasuring vane to slam open- and perhapsbounce-is damped by the balance vane run-

1Q)1

w;::,.

V6543210

.-- Us

30°

60°

90°

~- <::s

100 150 m3/h50

QL .

Interrelationships between intake air quantity,sensor-flap angle, voltage at the potentiometerand injected fuel quantity. (Robert BoschCorporation)

ning into this dead-end air pocket.Now, the force that pushes on the measur-

ing vane is a result of the momentum of thepassing air, and momentum is simply theproduct (multiple) of mass and velocity. Atfirst thought, then, it might seem that thissensor apparatus directly measures massflow - the weight of air passing per minute,which is what matters. In fact, the only waythat the vane could measure the total

momentum of the passing air would be tobring it all to a complete halt - no air wouldflow! Obviously, the vane must obstruct theairflow as little as possible (the pressuredrop caused by the vane is just 0.017 psi), soit only measures a small fraction of the totalmomentum, and we have no direct way ofknowing what that fraction is. As a result,the meter is actually measuring volumeflow. The relationship between the volumeof a certain quantity of air and its mass is afunction of its temperature-cold air isdenser than hot air. Consequently, an air

1.Ringgearfor springpreloading

2.Returnspring

3.Wipertrack4. Ceramicsubstratewith resistors

andconductorstraps

5.Wipertap6.Wiper7. Pump contact

7

65

1 2!

iI

4 3

The vane-type air meter of an L-Jetronic system, viewed from the electrical con-nection side. (Robert Bosch Corporation)

2

5 4 3

1.Compensationflap2.Dampingvolume

3.Bypass

4.Sensorflap5.Idlemixtureadjustingscrew(bypass)

The vane-type air meter of an L-Jetronic system, viewed from the other side. Notethe compensating flap which serves both as a counterweight and as a pulsationdamper. (Robert Bosch Corporation)

temperature sensor is provided, to allow theECU to correct the volume flow informationfrom the airflow sensor on the basis of the

air's temperature.

Idle Air BypassTo permit a degree of manual adjustment

of idle mixture strength (mainly to allowadjustment of idle speed), a small amount of

41

. Signals and controlled variables at the ECUQL Intake air quantity, 1'}LAir temperature, II Engine speed, P Engine load range,1'}MEngine temperature, VEInjection quantity, QLZAuxiliaryair,VEsExcess fuel for starting,VB Vehicle-system voltage.

Input variables Control unitand supply

QL

~

11M

==8>BOSCH

~M~

Signals and controlled variables at the ECU. (Robert Bosch Corporation)

air is allowed to bypass around the measur-ing section of the airflow meter through aseparate small passageway. This extra idleair is thus unmetered; the ECU is "unaware"of it. The amount of air that is bypassed inthis way can be adjusted by a small needlevalve arranged at the end of the bypass pas-sage nearest the engine. Turning this screwin makes the opening smaller, reducingbypass airflow and so richening the mixture.

Other Sensors

Apart from the main air quantity signalfrom the airflow sensor and the density cor-rection signal from the air temperature sen-sor, there are additional sensors continuous-ly providing information to the ECU. Likethe D-Jetronic, there is an engine tempera-

42

Output variables

VE==8>

~

~

ture sensor,the outputof which is used bythe ECU to enrich the mixture during coldstarts and during warm-up.Also like theearly D-Jetronic,there is a throttlepositionsensorthat reportsjust closed (idle)throttleandfull throttle.This informationis usedbythe ECUto providemixturestrengthcorrec-tionsfor those two operatingconditions.

Even though some domestic auto mak-ers retain manifold absolute pressure asthe principle load signal for their EFI sys-tems, Bosch regards the L-Jetronic's air-flow meteringas an improvementover themanifold pressure metering used on theD-Jetronic. Certainly it makes the systemless sensitiveto changesin engineconditionwith age, such as deteriorationof valveandpiston ring sealing, carbon build-up, etc.,

1

2

3

1. Eleclricalconnection

2. Housing3. NTCresistor

Engine temperature sensor. (Robert BoschCorporation)

but it has some potential faults of its own.For one thing, early versions could bedestroyed by a backfire, although laterones incorporate an "anti-backfire valve"that opens when exposed to a high pres-sure on the engine side of the vane. Andlike the D-Jetronic, any air leaks down-stream of the meter allow unmetered air

(sometimes called "false air") into theengine, causing the mixture to become lean.The airflow meter is also more expensive tomanufacture than the simple aneroid bel-lows of the D-Jetronic's pressure measuringapparatus.

Mechanical ComponentsThe mechanical components of the

L-Jetronic are similar to those of the

D-Jetronic. As there, an electricallydriven roller-type rotary pump draws fuel

2

3

4

1. Fullloadcontacl

2. Contouredswitchingguide3. Throttlevalveshaft

4.Idlecontact5. Electricalconnection

Throttle valve switch. (Robert Bosch Corporation)

from the tank. The L-Jetronic's pump isequipped with a check valve that keeps thefuel lines full and pressurized even when theengine is stopped. This permits quickerrestarts, and helps to prevent vapor lock. Thepump also has a relief valve at its intake end,which returns fuel to the tank if the internal

pressure rises too high.Because roller-cell pumps are much better

at pushing than at pulling, the pump is locat-ed low down, near the tank; indeed, in somecars it is actually inside the tank, immersedin fuel! Even in those cases where the pumpis located outside the tank, its internalelectrics are totally submerged in the fuelcarried within the pump; this helps to coolthe pump motor. Although this sounds dan-gerous, it is a practice that has been used inaircraft for years, and three decades ofuneventful experience with Bosch systemsusing this scheme should dispel any anxiety.(Although the pump can be damaged by

43

1.Intake

2.Rotorplate3.Roller

4. Rollerraceplate

5.Outletport

2 3 4

2 3 4 5

1. Intake

2. Pressurelimiter

3. Rollercellpump4. Motorarmature

5. Checkvalve

6.Outletport

The roller cell fuel pump. The rollers fit loosely in their lands; centrifugal forcepushes them against the contoured interior of the pump body. (Robert BoschCorporation)

overheatingif it is run dry for long; bearthisin mind if you ever run completelyout offuel.)A paperelementfilter is provided,butunlike theD-Jetronicthis is locatedafterthepump ratherthanbeforeit.

As with the D-Jetronic, the regulator isschematically at the far end of the fuel rail,maintaining pressure within the rail for theinjectors, and bleeding excess fuel back tothe tank. (A notable side benefit of this con-stant circulation of fuel is that it helps keepthe fuel cool, thus further reducing the riskof vapor lock.) The L-Jetronic's regulatormaintains either 36psi or 44psi fuel pressure,according to the vehicle model, somewhat

44

higher than the D-Jetronic's 28-32psi.Generally, the higher pressure (about 44psi)is used with more powerful engines; higherpressure means a greater fuel flow for agiven injector pulse time. Of course, the rateof flow at a given fuel pressure depends onthe local pressure where the fuel is beinginjected. Because manifold pressure variesaccording to engine load, the pressureagainst which the fuel is working has to betaken into account. Accordingly, the spring-loaded diaphragm in the L-Jetronic's regu-lator has a connection from its back surface

to the manifold. As manifold pressurevaries, the load against the diaphragmchanges, so the regulator maintains a con-stant pressure difference between the fuel inthe rail and the air in the manifold. The D-Jetronic did not need this feature because the

ECU took the manifold pressure intoaccount in the calculation of the injectorpulse time.

Although all later L-Jetronic systemsused digital electronics, the first versionwas analog (see the sidebar "Much orMany?"). Nevertheless, the principles ofoperation that were established with thefirst L-Jetronic system have enough incommon with all subsequent versions that itis now worthwhile to look at the way theinputs from the various sensors are dealtwithby the intemallogic of the ECU.

Inside the ECUIn contrast to the mechanical contact

points used in the D-Jetronic, L-Jetronic sys-tems use the firing of the ignition system asa trigger signal. In a four-cylinder, four-stroke engine, there will be a spark plug fir-ing twice per engine revolution. Thus, theECU (electronic control unit) receives twotriggering pulses per revolution. If you haveever seen an oscilloscope trace of an operat-ing ignition system, you will understand thatthese pulses, as received by the ECU, are"ragged"- the electricaloscillationsin thecoil windings produce a voltage trace with a

succession of "wiggles" that rapidly dampout.

The first operation within the ECU is toturn these ragged, spikey voltage pulses intoa simple square pulse of fixed duration-upon triggering by the ignition" spike," thevoltage at the output of these stages risesnearly instantaneously from zero to somesmall positive value, dwells there for a brieftime, then drops back to zero. (In electron-ics, this "pulse-shaping" circuitry, known asa "Schmidt trigger," is commonplace. Theyare used, for example, to eliminate theeffects of the bouncing that occurs with ordi-nary mechanical switch contacts, etc.) Thus,there are now two square-wave pulsesdeparting from the pulse-shaping circuitwith each revolution of the crankshaft.

These pulses then pass to a frequencydivider, which, in effect, ignores every otherone and passes on just one of them. The nextstage in the internal electronics thus receivesjust one pulse per revolution.

Pulse Time-That next stage is termed byBosch the "division control multivibrator."

Here a basic pulse time is generated, on thebasis of engine speed and air mass flow. Therpm is known from the frequency of thepulses arriving from the frequency divider;the air mass flow is known from the positionof the vane in the airflow meter, plus a cor-rection introduced on the basis of informa-

tion from the air temperature sensor. If alloperating conditions are "normal"-enginefully warmed up, running at a moderatespeed and load, and battery voltage nearnominal value-that pulse time would givean appropriate injector pulse time. Thesepulses of "basic" duration are then passed onto the next batch of circuitry- the "multipli-er stage."

Multiplier Stage-When circumstancesvary from the optimum that calls for a pulseof "basic" duration, further internal calcula-tions are performed by the multiplier stage.Low battery voltage, for example, will slowdown the opening of the injector valves. The

46

basic pulse time is calibrated on the assump-tion that the injector spends a mere one mil-lisecond opening, stays open for the dura-tion of the pulse, then closes almost instant-ly when the pulse ceases. Of course, almostinstantly means that there is some smallamount of fuel injected after the pulse ceas-es, while the spring within the injectorreturns the needle valve fully onto its seat.As long as the one millisecond delay inopening can be counted on, this "incidental"fuel on closing is offset by the openingdelay, so the basic pulse time will deliver anappropriate amount of fuel. Low voltagedoes not affect the closing spring, so theincidental delivery remains pretty muchconstant, but low voltage does extend theopening delay, so less fuel will be deliveredunder these conditions. A connection to the

battery "hot" terminal provides the multipli-er stage of the ECU with information aboutbattery voltage, allowing the calculation ofan extension to the basic pulse time, to makeup for the short-changing of fuel that resultsfrom the slow injector opening accompany-ing a low system voltage, such as occursafter an extended period of cranking. Thisvoltage connection is usually made as directas possible (no fuse, for example) to reducethe risk that corrosion at terminals may givethe ECU a false reading of battery voltage.

Inputs from the air temperature and enginetemperature sensors are used by the multi-plier stage to add more time to the basicpulse duration to account for cold startingand warm-up phases. Similarly, the pair ofmicroswitches that signal idle and full throt-tle also feed information to the multiplierstage, which adds the appropriate amount ofpulse duration to provide further enrichmentunder these two conditions.

Acceleration Enrichment

Apart from the microswitch that signalsfull throttle, there appears to be no provisionfor acceleration enrichment on many ver-sions of L-Jetronic. In fact, this is handled

0 1.

4 I-

~

+

BOSCH

16

Schematic diagram of an L-Jetronic system with lambda (oxygen sensor) control. (Robert BoschCorporation)

mechanically, rather than through the elec-tronics. When the throttle is opened sudden-ly, the rapid increase in airflow through thesensorcausesthe vane to overshoot momen-

tarily. Because the BCD is using the vaneposition as a measure of airflow, this exag-gerated movement of the vane influencesthe BCD to briefly increase the pulse time,thereby providing the enrichment needed toavoid a "lean stumble" when the throttle is

banged open.Some versions of L-Jetronic have afurther

refinement in this regard, in addition to theenrichment from vane" overshoot." The

same signal from the vane position sensor

that provides the BCD with its basic infor-mation about air quantity is also analyzedfor the rate-of-change of the vane's position,

so a rapidly moving vane will prompt theBCD to provide more enrichment thanwould a more slowly opening one. In eithercase, the BCD will add acceleration enrich-ment whenever the microswitch that signalsfull throttle is closed.

On cars with a lambda sensor-generally,those built after 1980-the output from thatsensor are also fed to the BCD. The multi-

plier stage processes information from thelambda sensor, increasing the pulse time(that is, richening the mixture) when theoxygenlevel detectedby the sensoris high,reducing the pulse time (leaning the mixtureout) when the oxygen level drops. This cou-pling between the lambda sensor and theBCD is called "closed-loop" control- infor-mation from the sensor induces the BCD to

1.Fueltank

2. Electric fuel pump

3.Fuelfilter4.ECU

5.Fuelinjector6.Fuelrail andfuelpressureregula-tor7. Intakemanifold8.Cold-startvalve

9.Throllievalveswitch10.Airflowsensor

11.lambdasensor

12.Thermo-timeswitch

13.Enginetemperaturesensor

14. Ignitiondistributor15.Auxiliaryair device

16.Ballery

17. Ignitionandstartingswitch

47

make changes, which are detected by thesensor which then provides a different signalto the ECU, which adapts again. . . and soon. (See also the sidebar "The OxygenBattery" in Chapter 2.) Whenever the fullthrottle microswitch is closed, the ECUreverts to "open-loop" control-it ignoresdata from the lambda sensor, and dependson its internal "maps."

After all these calculations, the pulse thatemerges from the multiplier stage has the"basic" on time span, plus some added dura-tion for enrichment, according to inputsfrom all the various sensors. That pulse,however, is extremely weak; the internalelectronics all operate on just tiny amountsof voltage and current. Heftier electroniccomponents are needed to deal with the 12volts and comparatively large currents thatactuate the injectors. That task is handled inthe final" amplifier stage" of the ECU.

There are two points to note here. First, thepulses are coming once per crank revolu-tion-on a four-cylinder, four-stroke enginethat means two pulses per cycle-and all theinjectors fire at once. Second, while theabove description of pulses leaving the ECUto drive the injectors is a fair schematic rep-resentation of what is going on, the injectorsare actually hot all the time-the ECU infact grounds the injectors for the duration ofeach pulse.

Cold StartingAs might be expected, cold starting is a

special case that may lie outside the normalrange of calculation of the multiplier stage,and the total fuel requirement may exceedthe capacity of the injectors, even when theyare "on" almost all the time. Different ver-

sions of L-Jetronics adopt different strategiesfor cold starting.A few get by simply by vastincreases (a doubling or tripling) of pulsetime; most use a separate "cold start" injec-tor, just like the mature version of the D-Jetronic; a few others use both approaches.

The two temperature sensors on

48

L-J etronic systems - one for intake air,the other for engine temperature - are bothbased on what is called a "negative temper-ature coefficient" (NTC) resistor. This refersto the fact that, unlike many resistors whoseresistance value increases positively withtemperature, the NTC has the opposite char-acteristic; when the NTC is cold, it has ahigh resistance, when hot the resistancefalls. NTC I (or Temp I), in Bosch's termi-nology, is the air temperature sensor; NTC IIsenses engine temperature. So, while infor-mation from NTC I is used all the time bythe ECU to correct the airflow information

from the airflow sensor vane, in effect to cal-

culate air mass flow, NTC II is the key com-ponent in achieving the wholesale enrich-ment needed for cold start. On water-cooled

engines, NTC II will be found in the samesort of places - such as the thermostat hous-ing- where you would expect to find anordinary "temperature sender" for the dashtemperature gauge. . . and may be confusedwith it! On air cooled engines, NTC II willbe screwed into a cylinder head.

In the comparatively new L-Jetronics thatuse just an increased fuel delivery throughthe standard injectors for cold starting, ahigh resistance value (say 3000 Ohms ormore) detected at NTC II will simply influ-ence the ECU to increase the pulse time. Inthese systems, the required fuel flow for a"cold-cold" start (engine temperatures muchless than room temperature) may require theECU to "double-up" the firing of the injec-tors as long as the starter is cranking, to pro-vide two injection cycles per crank revolu-tion, instead of the usual one cycle. TheECU makes this decision based on both the

temperature reported by NTC II and by thecranking rpm. The ECU also keeps track ofthe time elapsed during cold cranking, andwill reduce the quantity of fuel injectedunder circumstances that seem likely to leadto flooding.

However, many L-Jetronics (and certainlymost early ones) use a "cold start valve," in

2

1. Electricalconnection

2.Housing3.Bimetal

4. Heatingwinding5. Electricalcontact

3

4

5

The thermo-time switch. The contacts are closedwhen the switch is cold, and open when it is hot,either from a warm engine or because the electricheating element has had time to warm it. (RobertBosch Corporation)

conjunction with a "thermo-time switch."This cold-cold start circuit, which operatesonly as long as the starter is cranking, isquite independent of the ECU. Power is fedto the cold start injector (which Bosch termsa "cold start valve") via the starter terminalon the ignition switch, but whether or not theinjector is actually energized depends onwhether or not it is grounded. And thatdepends on the thermo-time switch.

The thermo-time switch comprises a pairof electrical contacts, one fixed and one atthe end of a bimetallic strip, which consistsof two layers of different metals bondedtogether. The two metals chosen have wide-ly different thermal expansion characteris-tics, with the result that the strip bends to agreater or lesser extent, according to its tem-perature. When "cold" (less than about 95°F), the strip is nominally straight, and thecontact on its end touches the fixed contact,

closing the switch so current will flow. The

1

6

1. Electricalconnection

2.Fuelinlet

3.Valve(solenoidarmature)

4.Solenoidwinding5. Swirlnozzle

6.Valveseat

4 .

3

5

A cold start injector (cold start valve), shown in the "on" state. (Robert BoschCorporation)

cold start injector is wired in series with thisswitch, so when the starter is energized withthe thermo-time switch closed, current flows

through the windings of the cold start injec-tor, so it opens and passes fuel into theintake manifold.

To prevent engine flooding, an electricalheating element surrounds the bimetallicstrip. This heater is fed by the same currentthat is passing through the contacts, so aftera brief time (never more than about ten sec-onds, even in the coldest weather) the stripwill heat up, bend, and so open the switch,cutting off juice to the cold start injector.Because its working end is immersed inengine coolant, or at least attached to theengine block, the thermo-time switch willremain open if the engine is much aboveroom temperature.

Auxiliary Air BypassBecause Bosch EFI systems have no

mechanism corresponding to the fast-idlecam on a carburetor equipped with an auto-

49

,,,,

\,,,,,,,,'\," ~ut-off

\,~ lImit\ ""

'-..

"------

~ -Cut-inlimit

0

-30 -10

To improve economy and

reduce emissions, the fuelisshut offwhen the throttleis

closed, as signalled by the

throttle switch. To prevent

"hunting," the cut-off and

cut-in limits are separated by

a few hundred rpm. The rpm

at which cut-off occurs is

lowered proportionally as the

engine warms up, to prevent

stalling with a cold engine.

(Robert Bosch Corporation)

50

10 50 110 Deg C70 9030

Coolant temperature

matic choke, the extra internal friction of a

cold engine could pull the idle speed downso low that the engine may not continue torun. To provide the equivalent of a slightthrottle opening so as to maintain a reason-able idle speed, an "auxiliary air bypass" isprovided. This bypass amounts to a smallrotary disc valve; depending on the positionof the disc, more or less air (or none) isallowed to bleed around the nearly closedthrottle. The position of the disc, in turn, isgoverned by a bimetallic coil. When the coilis warm, the valve is completely closed;when very cold, it is fully open; at interme-diate temperatures, it adopts a positionsomewhere in between. A smooth and grad-ual transition from open to closed as theengine warms up is assured by the profilingof the valve disc and the calibration of the

bi-metallic coil and of a return spring thatpulls the valve to the closed position. A fewearly versions of this valve are heated byengine coolant; most recent ones are electri-cally heated, just like the thermo-time

switch.

Note that unlike the idle air bypass, theauxiliary air valve merely skirts around thethrottle plate; all the air passing throughthe auxiliary air valve has already passedthrough the airflow meter, and so is"counted" by the ECU. Note too that,unlike the functioning of the thermo-timeswitch/cold start injector which is never inoperation for more than a few seconds, invery cold weather the auxiliary air bypassvalve can take several minutes (thoughnever more than eight or ten) to move fromfully open to fully closed.

Warm-UpThe first 30 seconds after a cold start

represent perhaps the toughest challenge toauto emissions engineers. On the onehand, considerable enrichment is needed

to keep a cold engine running and to havegood "driveability;" on the other hand, allthe fuel in excess of stoichiometry just flush-es out the exhaust, adding to HC emissions.Once a cold engine has fired, the ECUimmediately cuts back on the wholesaleenrichment that has been provided for start-ing. A much richer than nominally stoichio-metric mixture is still needed, however, untilthe engine is fully warmed up, so instead oftwo to three times normal pulse time (pluswhatever has been added through the coldstart valve), the enrichment drops to, say, 1-1/2 times the "basic" pulse time. The ECUgradually tapers off the "post-start" andwarm-up enrichment, on the basis of bothtime elapsed and the information from NTCII. The initial rate of this progressive leaningof the mixture toward a stoichiometric value

is quite rapid, and is determined mostly onthe basis of time; after about 30 seconds, thecontinuing reduction of fuel quantity isessentially determined by engine tempera-ture. Even in the coldest weather, the enginewill have warmed up enough for the mixtureto be dialled back to near nominal values

within a couple or three minutes.

RPM

3800

3400

'""d 3000Q)Q)P-< 2600[f)

Q)2200

"'-;

0.0

1800

1400

1000

Idle/Coasting "Cutoff" SwitchWe have mentioned that the microswitch

that signals closed (idle) throttle signalsthe ECU to provide a slight enrichment foridle. There are a couple of other functionsfor this switch on some vehicles equippedwith L-Jetronic. To avoid a "stumble" dur-

ing the off-idle transition, the opening of thisswitch- indicating that the throttle has justbeen opened - is used to trigger a brief addi-tional enrichment. Also, on later vehicles,

the closing of this switch when engine speedis well above an idle causes the fuel flow to

be shut off entirely, to eliminate emissions(and fuel consumption) during this "over-run" situation. To prevent the engine fromstalling, the fuel is turned back on again asthe idle speed is approached, at an rpm thatdepends on engine temperature.

LH -JetronicFor all that it was an advance on

D-Jetronic, the L-Jetronic system wasnot without its own shortcomings. One ofthese was the need to make combined use of

two separate signals-one from the airflowmeter and one from NTC I- in order to

compute the air mass flow. The design of theairflow meter itself might also be regardedas imperfect, depending as it does on a com-bination of electronic and mechanical com-

ponents. While modem electronics are aboutas reliable as man-made systems get to be,anytime a mechanical device is intercon-nected with an electronic one, the mechani-cal bits are the most likely source of futureproblems. Apart from reliability issues, cer-tain types of road disturbances can poten-tially "jostle" the metering vane, leading tomomentarily inaccurate information comingfrom it. The vane-type airflow meter is alsoa somewhat expensive device to make.

To produce a light, compact, simple, all-electronic sensor with no moving parts thatdirectly measures air mass, with no need forcalculations relating volume flow, tempera-

1.Temperaturesensor

2.Sensorringw/hot wire3.Precisionresistor

Qm =Massairflow

1 2 3

QM

.~/

Components of the hot-wire air mass meter. (Robert Bosch Corporation)

ture, and density, Bosch embraced a devicelong used in fluid-flow laboratory work-the hot-wire sensor. Other advantages of thisform of airflow sensor are its more rapidresponse (measured in milliseconds), and aneven further reduction of the already near-trivial resistance to airflow of the vane typesensor. The adoption of the hot-wire sensorgave rise to a next-generation form of inter-mittent EFI called LH-Jetronic, first intro-duced in 1982 and used on the Volvo 2.3

liter GL model of that year. The "H" stands

Hot-wire air mass meter. Thewire is so fine (less than0.003 inch) it may only bevisible when it glows red hotduring its "burn-off' cycle.(Robert Bosch Corporation)

51

t

0

15

1.Fueltank

2. Electricfuel pump3.Fuelfilter4. ECU

5.Fuelinjector6.Fuelrail

7. Fuelpressureregulator8. Intakemanifold

9. Throttlevalveswitch

10. Hot-wireair masssensor

11. lambdasensor

12. Enginetemperaturesensor

4

10

+

BOSCH

13.Ignitiondistributor14.Rotaryidleactuator

15.Battery16.Ignitionandstartingswitch

Schematic diagram of an LH-system. The "H" stands for heiss-the German word for hot. Apart from the air meter itself, this is almostidentical to the L-Jetronic. (Robert Bosch Corporation)

for heiss, the Gennan wordfor hot.Other than the nature of the sensor itself,

many of the operating principles (and a fewof the components) of L-Jetronic and LH-Jetronic are held in common. The mechani-

cal components-fuel pump, regulator andinjectors-differ only in detail, if at all, andthe computation underlying the fuel quanti-ty/pulse time calculation is logically identi-cal to that of the L-Jetronic. Note, however,that while early versions of L-Jetronic useanalog electronics, all LH-Jetronic systemsare purely digital. Further, all LH-Jetronicsystems include lambda feedback control.

52

Hot-WIre Sensor

The sensing element is a very fine wire ofplatinum alloy, just 0.7 millimeters (lessthan 0.003 inches) in diameter, strung acrossthe interior of a housing through which allthe incoming air stream has to pass. Thewire is heated by an electric current, butcooled by the passing air stream. An electri-cal circuit adjusts the current passingthrough the wire so as to maintain it at atemperature that is consistently 180° Fabove that of the entering air. Thus, if the airtemperature is 80° F, the wire will be run-

ning at 260° F; if the air is at 30° F, the wiretemperature will be at 210°. The currentrequired to maintain this equilibrium is ameasure of the mass of air passing the wire,and is used as the principal signal to theECU as to air quantity.

You many wonder how the electronics"know" that the wire is 180° hotter than the

incoming air. The electrical circuit referredto above is a "Wheatstone bridge," an ele-gantly simple form of electrical circuit, andwhile we will not delve into the actual elec-

tronics, the principles are not particularlyhard to understand, even for the "electroni-

cally challenged." First, note that the electri-cal resistance of platinum varies verystrongly with its temperature (it is some-times used as a temperature detecting/con-trolling device, just for this reason). Second,note that, like the L-Jetronic, the LH isequipped with an intake air temperature sen-sor, which is itself a temperature-sensitiveresistor - a thermistor - but one whose

resistance varies much less with temperaturethan the platinum wire.

If an electrical path is divided into twoparallel branches, and each branch containsa resistor, the voltage difference between thetwo branches, measured after the resistors,will depend on the relative values of the tworesistors. Because one of those resistors-

the wire- has a resistance that varies strong-ly with temperature, the voltage differencebetween the two branches will similarlydepend strongly on the temperature of thewire. If the current through both resistors ischanged, the wire will either heat up or cooldown.At some level of current- that is, atsome temperature of the wire-the tworesistors will have equal values, so the volt-age between them will be zero. So, by intro-ducing a detector and amplifier that con-stantly adjusts the current so as to maintainthe voltage between the two branches of thecircuit at zero, we can be sure that the wireis some fixed temperature (in this case 180degrees) higher than the temperature at the

1.Hybridcircuit12 2. Cover

3.Metalinsert

4. Venturi w/hot wire

5. Housing6. Screen

7. Retaining ring

!3

4

5

Another view of the hot-wire air mass meter, with integrated electronics. (RobertBosch Corporation)

other resistor - the air temperature sensor. Inthis case, the current flowing through thecircuit will be in direct proportion to the heatlost by the hot wire, which is proportional tothe mass flow of air. The ECU, which sup-plies the current, thus "knows" what mass ofair per minute is passing through the airmass sensor.

Like the L-Jetronic, early LH systems hadprovision for adjustment of the idle mixturestrength. Unlike the L-Jetronic, however, inwhich the adjuster actually varies the size ofthe air "leak" around the flow meter, theadjustment on the LH-Jetronic is electricalin nature. The small adjuster (located undera sealed cover on the meter) is actually apotentiometer that modifies the current sig-nal received by the ECU. Later LH-Jetronics

53

a.Housingb.Hotfilmsensor(filledinmiddleofhousing)1.Heatsink2. Intermediatemodule

3. Powermodule

4. Hybridcircuit5. Sensorelement

a

5

b

5

13

42

The hot-film air mass meter, as used on most later versions of LH-Jetronic. This isless prone to damage than the fine wire in the earlier meter. (Robert BoschCorporation)

54

omitted this adjustment.

Drawbacks

One potential problem with a hot wire as ametering element is that if its surfacebecomes contaminated, the rate of heat flowout of it will be reduced, in which case it

would send the BCD a false signal, underes-timating the amount of air passing. To dealwith this, the BCD's program includes a"bum-off" cycle that feeds a much largerthan normal current to the wire for about one

second immediately after the engine shutsoff. This brief, large surge of current heatsthe wire to about 1800° F, which turns it red-

hot and vaporizes any dirt depositson the

wire. (The whole concept is a bit like a self-cleaning oven!)

Safety Issues-Of course, there are a cou-ple of safety issues here-anything red-hotin close proximity to a device that metersgasoline is a potential hazard. Accordingly,Bosch provides a couple of safeguards.First, if the last reported rpm before the igni-tion was turned off was less than 200 rpm,then the engine was not idling, it stoppeddead for some reason - and that reason

might be an accident. In that case, the burn-off cycle is suppressed. Second, if the enginehas not exceeded 3000 rpm since the lasttime the ignition was shut off, then maybethe driver just started the engine, thenchanged his mind. . .and maybe he willchange it back again and refire the engine.Again, you don't want any chance of theengine turning over with the wire red-hot, soagain the bum-off is cancelled.

In more recent versions of LH-Jetronic

(and LH-Motronic-see next chapter), thehot-wire is replaced by a hot-film air masssensor. Apart from the rearrangement ofthe material that forms the heated surface,

the operating principle is identical to thatof the hot-wire sensor. The hot-film sensoris used on most modern Audis andMercedes, as well as some VW and BMWmodels.

Idle Speed StabilizationWhile early versions of the LH sys-

tem used the same auxiliary air valveas the L-J etronic to help maintain a con-stant idle speed, irrespective of engine tem-perature, most LH systems use a differentdevice for the same purpose. Like the auxil-iary air valve on the L-Jetronic, this "idlespeed stabilizer valve" functions by bypass-ing a certain amount of air around the closed(idle) throttle. (Again note that this air haspassed through the hot-wire air sensor, so itis "measured" air-the functioning of theidle speed stabilizer does not affect the mix-ture strength. Functionally, this is exactly the

1

1.Fueltank

2. Electricfuelpump3.Fuelfilter

4. Fuelinjector5.Fuelrail

6. Fuelpressureregulator

t

911_11<>L:J

+0

BOSCH15

7. Intakemanifold

8. Throttlevalveswitch

9. Airflowsensor

10. ECU

11. lambdasensor

12. Enginetemperaturesensor

13.Ignitiondistributor14.Auxiliaryairdevice15.Battery16.Ignitionandstartingswitch

Schematic diagram of an L3-Jetronic, having the additional control of a lambda (oxygen) sensor. (Robert Bosch Corporation)

same thing as cracking the throttle openslightly.)

While the L-Jetronic's auxiliary air valveis a free-standing piece of equipment, withno connection to the ECD, that determinesfor itself the amount of air that should be

allowed to pass based on its own tempera-ture, the idle speed stabilizer of the later LHsystems is connected to the ECD, and theamount of air that is bypassed through it isdetermined by the ECD.

Instead of the auxiliary air valve's bimetal

strip and electric heating element, the LHsystem's idle speed stabilizer is opened andclosed by a rotary actuator whose position isa function of the "on time" to "off time" of a

series of digital pulses coming from theECD. The greater the on/off ratio, the furtherthe valve opens, passing proportionallymore air. The ECD, of course, "knows" the

engine rpm from the frequency of the igni-tion triggering pulses it receives. If idle rpmdrops-say, because the driver turned onsomething that produces a heavy electrical

55

load, such as a rear window defroster-the ECU increases

the on/off ratio of the pulses it is sending out to the idlespeed stabilizer so as to bleed more air past the closed

. throttleto maintainan appropriateidle speed.As with the connection with the lambda sensor, this cou-

pling of the actions of the ECU and the idle speed stabiliz-er produces a closed-loop control. Because this providesdirect, feedback-driven control over idle speed, the idlerpm can be set lower than when the amount of air bypassedis based on a fixed, "best guess" program. During coldstarting, the ECU breaks the feedback loop, and directs theidle speed stabilizer to open a preprogrammed amount,according to temperature.

An ingenious additional feature of the idle speed stabi-lizer is found on some LH installations. Arguably thelargest load change that an idling engine experiences is theengagement of the air conditioner compressor. This load issufficiently large, and can commence with such sudden-ness, that there is a risk of the engine stalling before thestabilizer has a chance to respond. To conquer this tenden-cy, the air conditioner controls are fed through the ECU,and when the alc thermostat "tells" the compressor to cutin, the ECU briefly delays the actual clutching-in of thecompressor, to give the idle speed stabilizer a chance toanticipate the additional load. Clever!

MUCH, ORMANY?ANALOGVS. DIGITALA dozen eggs, a quart of milk; a hundred bricks, a bag of cement. We gauge the quantity of some things by count-

ing, but others we measure by weight or volume. The difference is subtle but basic-the first case involves a dig-ital operation, the second is an analog process. Now, while one major aspect of the electronic revolution was theinvention of the transistor, the shift from electronic devices that operated on an analog basis to ones that work dig-itally is at least equally significant.

At its simplest, a transistor can be thought of as an electronic control valve. There are three wires attached; two"in" and one "out." You feed a fixed, comparatively large amount of power to one of the inputs, and a much small-er, variable amount of juice to the second. What comes out of the third wire may vary from zero up to the full sup-ply voltage, depending on the size of the signal (the voltage) coming in on wire number two. The special-purposecomputers that comprised the "brain box" of early electronic fuel injection systems used transistors in this way-as proportional devices. A sensor-say for engine coolant temperature-was hooked (in principle, at least) to theinput of a transistor. Variations in engine temperature would thus directly alter the output, which in turn controlledthe "on time" (pulse time) for the injector(s), adjusting the mixture strength according to whether the engine wascold, cool, warm, or hot.

But if you fix the "signal" voltage at some comparatively high value, then a transistor will operate more like asolenoid relay - if the input at the signal terminal is "on," the transistor puts out; if the signal voltage is "off," so isthe output. That makes the transistor function as a digital device. On the face of it, it seems a shame to waste theability of a transistor to work proportionally, but this digital way of doing things has the enormous advantage thatnow you can use transistors to create an electronic memory. String a few of them - say sixteen- together, and byhaving some of them "on" and others "off," you have 256 (16 x 16) possible arrangements. Each arrangementmight then correspond to one of 256 different values along the cold-hot continuum.

Part of the attraction of operating digitally is the unambiguous result you get when you count things. One plusone always equals exactly two, but when you add "some" and "some," you surely get "more," but you cannot besure you have twice as much. Even slight variations in components-whether from production tolerances, or fromaging,or heavenknowswhat- wouldmeancorrespondingvariationsin the electricaloutputs.Air/fuelratioscon-trolled by those electronics would then be slightly (or perhaps wildly) in error. Doing things digitally means that agiven sensor input (coolant temperature equals 86° F, say) would correspond to "coolant temperature value #47,"

56

for example, and "#47" stored in the digital memory would correspond to an injector pulse lasting so many mil-liseconds.

The opportunity to make use of a memory also means that many more combinations of factors can be taken intoaccount in the process of determining the pulse time of the injectors, and thus mixture strength. Thus, the appro-priate amount of fuel for a certain engine speed and load when the coolant is at 86 degrees and the air temperatureis 7SOF is likely to be different from the same situation but with an air temperature of minus 10°F.And when igni-tion control gets added in-as with the Motronic systems-the complication becomes unmanageable without adigital memory.

The air meter of an L3-Jetronic. Use of digital electronics and advances in component miniaturization permit inte-

grating the ECU into the body of the air meter itself. (Robert Bosch Corporation)

In many of the very earliest internalcombustion engines ignition was bymeans of a "hot tube" igniter. The tube,

usually made of porcelain, was locatedimmediately outside the combustion cham-ber, and was continuously heated by a smallgas flame that made it glow red hot. As thepiston approached top center on the com-pression stroke, a timing valve opened,exposing the contents of the combustionchamber to the hot tube, which lit (some-times) the combustible mixture in the cylin-der. Motoring accounts written around theturn of the century make frequent referenceto stoppages resulting from the flame blow-ing out!

By the early 1900s the days of hot tubeignition were over, and the automotiveworld embraced ignition by an electricspark. As is common with emerging tech-nologies, there was a bewildering prolifera-tion of designs at first, but there were reallyonly two general categories: magnetos ofvarious types, and systems based on aninduction coil. Magnetos eliminated theneed for a battery- in those days a fragiledevice, and even heavier than today's batter-ies-and so became almost universal on

race cars, motorcycles, and aircraft.Induction coil systems offered the advantageover magnetos of improved ignition at lowspeeds, but required an on-board source oflow voltage-a battery. As electric lighting(and, later, electric starting) became wide-spread, the battery was seen as less of a lia-bility, and design rapidly converged towardthe coil and breaker system designed in1911 by Charles Kettering. This "Deleo"system- named after the Dayton ElectricCompany, founded by Kettering - rapidly

became almost universal after its introduc-tion on GM cars in 1922, and remained sofor more than half a century, through theearly 1970s.

Coil and Breaker IgnitionAn induction coil consists of a central core

of soft iron surrounded by two separatewindings of copper wire. One of the wind-ings, called the "primary," consists of asmall number of turns (typically a couple ofhundred) of comparatively thick wire; theother, called the" secondary," consists of amuch larger number of turns of much finerwire. At one end, the primary winding isconnected to the "hot" lead of the electrical

source, while the other end is grounded, atleast most of the time. The secondary wind-ing shares the primary's connection to the"hot" lead of the battery; the other end isconnected to the spark plug's center elec-trode, via the distributor cap and rotor.

When current flows through the primarywinding, a magnetic field is created in thecore-the coil is said to be saturated-and

this field is maintained as long as current isflowing. But if this flow then suddenly stopsfor any reason, the magnetic field in the corecollapses, and the stored magnetic energy"induces" a burst of electricity in the sec-ondary winding, hence the name "inductioncoil." The interruption of the current flow isaccomplished by the breaker points. Whilethe primary circuit nominally operates atjust twelve volts, the induced secondaryvoltage is a thousand or more times higherthan that- sufficientto jump the gap in thespark plug. Notably, the collapse of thecore's field also induces a high voltage in theprimary windings, which thus momentarily

59

experiencea surge of perhaps a couple ofhundredvolts.

Transistorized IgnitionFor passenger cars, the "Kettering" system

worked adequately well-though only just.(At the very least, it was an advance on hotbulb ignition!) The major catch was the needfor regular maintenance, but there wereother drawbacks as well.

One limitation of the coil and breaker sys-tem is the current-switching capacity of thepoints-a primary current of more than 5 or6 amps will tend to cause burning and arcingat the points, shortening their life. This, inturn, places a restriction on the strength ofthe magnetic field that can be stored in thecoil, which limits the amount of poweravailable to fire the plug.

Dwell

Another problem is the business of dwell.With the traditional cam and breaker

arrangement, the primary has current flow-ing through it for a certain fixed number ofdegrees of crank rotation, but the corre-sponding length of time that the current isflowing thus varies with engine speed. Thisleads to a dilemma: If you keep the dwellangle small, the coil will have insufficienttime to become fully saturated at highengine speeds, leading to misfiring at highrpm; but if you arrange for a longer dwelltime, say by using a distributor cam with adifferent lobe shape, so the points spendmore time closed and less time open, then atlow speeds the primary becomes saturatedearly in the cycle, but then the current keepsflowing because the points are still closed,and the coil will tend to overheat.

Finally, the fiber rubbing block on themoveable half of the point set slowly wearsdown, which causes it to contact the cam

ever closer to the peak of the lobe, so thepoints open progressively later, and the igni-tion timing gets more and more retarded. Wewill soon discuss the importance of this

60

issue of spark timing.These first began to emerge as serious

problems during the 1960s, when manufac-turers were heavily engaged in a horsepow-er race. Compression ratios over 12:1, com-bined with engine speeds over 6000 rpm,tested the upper limits of coil and breakerignition, and it became clear that the days ofthe classic KetteringlDeko system werenumbered. Ironically, what finally ensuredthe emergence and universal adoption ofmodem electronic ignition was not streethemi's and other monster motors, but ratherthe introduction of emissions legislation, inthe 1970s. Though compression ratios tum-bled and revs dwindled, the need to fire verylean mixtures, and to do so reliably through-out the life of the car, became the new fac-

tors that finally eliminated contact points.When you are scraping to meet standards forthe emission of hydrocarbons you cannotafford even one misfire in ten thousand, andwhen you have to certify the engine's emis-sions for 50,000 miles without a tune-up,you need a system that does not depend onbreaker points, and one that can still fire aspark plug when its gap has eroded open to50 thou or more.

The solution that was eventually adoptedretained the familiar induction coil, buteliminated the problems of contact points byeliminating the points themselves. Instead ofa mechanical switch to turn on and off the

current flow in the primary windings of thecoil, the switching is done electronically.The distributor cam is replaced by a rotatingmagnet, having a number of teeth or "poles"corresponding to the number of cylinders.As each pole passes a fixed pickup head, asmall pulse of electricity is produced. Thatpulse is then used to trigger a transistor-anelectronic relay - that briefly interrupts,then quickly restores, the current in the pri-mary winding of the coil.

Compared to a conventional Ketteringsystem, there are two major advantages tothis scheme. First, the constant wear on the

rubbing block is eliminated, so timingremains accurate and no adjustment orreplacement is ever needed. Second, whenyou remove the concern over the amount ofjuice you can put through the points, the cur-rent in the primary circuit can be dramati-cally increased, so the coil can be redesignedto provide a greater secondary voltage out-put that can fire a plug with a gap that islarger, whether by design or as a result ofwear, without the risk of frying the points.As long as there is enough juice to jump thegap, a wider plug gap provides a biggerspark that is more likely to light the mixturein the cylinder. (The beliefthat "the mixturedoesn't care if it's lit by a match or by ablowtorch" is mistaken; the amount of ener-

gy in the spark does matter.) The sameincrease in potential output from the coilalso means you can fire a plug that has someadditional insulation, beyond the air gap,between its electrodes, such as one that is

fouled. The potentially higher energy outputof a modem electronically switched coil willlight the fire in such an engine when a pointstype system would not.

The "brains" of an electronic ignition sys-tem makes this possible without riskingoverheating of the coil because it can, ineffect, vary the dwell angle. At low enginespeeds, a timing circuit in the electronicsdelays restoring the primary current until itjudges there is just enough time remainingto saturate the core. At higher speeds, it triesto keep the dwell time nearly constant,allowing primary current to flow during alarger number of crank degrees. This fea-ture, together with freedom from the currentlimitations of breaker points, is what permitsthe redesign of the coil to give potentiallymore secondary output, without cooking itthrough an excessive primary current. Butbeyond the potential for greater spark ener-gy, and near bulletproof reliability, electron-ic triggering of ignition also makes it possi-ble to use electronic control for the timing ofthe spark.

Ignition TimingWe explained in Chapter 1 that the com-

bustion event, while extremely fast, is not aninstantaneous explosion that takes place themoment the plug fires; rather, it is an occur-rence that takes some time. Indeed, as a veryrough first approximation, we can say that atany given load, the time taken by the com-bustion event is pretty much constant, over aconsiderable range of engine speeds. Whatthat means, however, is that the combustionevent occurs over a variable range of crank

angle, the variation depending on the rpm.Now, if the spark was arranged to occur

exactly at TDC, then the engine might runsatisfactorily at very low speeds, but as therpm's increase, the combustion would lagfarther and farther behind the piston move-ment, reaching the point where the fire wasjust barely getting started at the moment theexhaust valve opened, so most of the energywould just get flushed down the pipe. This,clearly, is no good for power output, fueleconomy or emissions. Thus, real worldengines incorporate a degree of ignitionadvance- the spark occurs before the pistonreaches TDC. And because of the fixed-

com bus tion -timelv ariab le-engine-s peedrelationship, that advance is usuallyarranged to be variable-less advance atlow speeds; more at high speeds.

Limits and ComplicationsOf course, there are limits, and there are

complications. As to limits, if the sparkoccurs too soon, the pressure will have risenso high by the time the piston reaches TDCthat the pressure opposing the rising pistonnegates much of the power delivered later,when the piston is on its way back down onthe power stroke. Also, an advanced sparkmeans that the pressure of the early stages ofcombustion add to the pressure rise causedas the ascending piston squeezes the com-bustion chamber contents ever smaller, and

we saw in Chapter 1 that the uncontrolled61

combustion called detonation is initiated byexcessively high temperatures and pressuresduring these early phases of combustion.Thus, excessive spark advance leads to det-onation. Finally, the assumption that com-bustion takes a fixed length of time, even atconstant load, is only valid at moderateengine speeds. In fact, at high speeds theincreased degree of turbulence in the com-bustion chamber, both before ignition andduring the early stages of combustion,increases the speed of combustion and so therequirement for an ever increasing advancewith rising rpm levels out at some point,beyond which the appropriate amount of"spark lead," as it is called, is more or lessfixed.

As to complications, all of this discussionrefers to an engine at constant load, and realengines don't work under those simplifiedconditions. A larger load at a given speedmeans the throttle is open wider, so a greatermass of air/fuel mixture will be inhaled, sothe pressure in the cylinder near the end ofthe compression stroke will be higher. Thusto avoid detonation, increased load calls forless ignition advance.

In the old days of "flintlock" ignitions, thevariation of advance with rpm was handledby a centrifugal advance mechanism. Thiscomprised a set of weights spinning withinthe distributor body that were flung furtheroutward as speed increased and so, througha system of links, rotated the plate on whichthe points were mounted, relative to theposition of the cam that opened them.

Vacuum Advance Mechanism

The reduction in advance called for by anincrease in load was provided by a vacuumadvance mechanism. (The name is poten-tially confusing; while it did in fact increaseadvance in response to high manifold vacu-um, its function was really to retard thespark when vacuum was low, indicating ahigh load.) The vacuum advance generallytook the form of a chamber containing a

62

diaphragm that had one side exposed tomanifold vacuum and a link connected to

the other surface that, like the centrifugalweights, swiveled the distributor plate(although this motion opposed that causedby the weights).

Asking mechanical devices like these tobalance all the variables that determine opti-mum spark advance is, of course, asking toomuch. Accordingly, soon after the introduc-tion of electronically triggered ignition-generallycalledtransistorizedignition- thetask of timing the spark was sometimes alsoentrusted to the "black box." Because the

centrifugal and vacuum advance mecha-nisms were eliminated from the distributor,leaving simply the rotor and cap to direct thespark to the appropriate plug, the distribu-tors on engines equipped with such elec-tronic ignition have a stumpy, squat appear-ance. We are terming this as electronic igni-tion, to distinguish it from transistorizedignition, described above. Note that this isnot quite the same thing as distributorlessignition, in which the rotor and cap are elim-inated and a separate coil provided for eachplug, individually triggered by the ignition"black box." Bosch introduced a true distrib-

utorless system like this in 1990, with theMotronic M3.1. BMW was the first to use

this on its in-line, 4 valve per cylinder in-linesix, and on 2.0 and 2.5 liter four cylinderengines, as used in both 3- and 5-series cars.

Apart from eliminating the potentiallytroublesome mechanical components, andbeyond the fact that electronics also respondmuch faster than the old fly-weights andvacuum diaphragm arrangement, electroniccontrol over spark timing meant that theseignition systems could (at least potentially)calculate the appropriate amount of sparkadvance based on many more factors thansimply rpm and manifold vacuum. Theseadditional factors include the rate of changeof load and/or speed, coolant temperature,battery voltage, and speed of starter crank-ing, among others. Some sophisticated dis-

IgnitionAngle

Motronic ignition advance (top) comparedwith map of a conventional mechanicalsystem (bottom). Electronic control overignition timing, based on numerous inputs,permits finer and much more detailed con-trol over spark advance, maximizing per-formance without risking detonation.(Robert Bosch Corporation).

IgnitionAngle

\.\\i\\\~ 'S'V~~~

63

6

1

1.Additionalprogrammemory

2. Analogdigitalconverter 23

3.Microcomputerforstandardpro-

gramanddata

4. Integratedcircuitfor enginespeed

andreference-marksignalprocess-

ing

4

5. Ignitionoutputstage

6. Fuelinjectionoutputstage

5

Because the Ideal ignition advance depends on mixture strength, and vice versa, optimum results can only be achieved with integratedcontrol over both ignition timing and fuel injection. This is what the inside of the "black-box" that does it looks like. (Robert BoschCorporation)

64

tributorless ignitions do, in fact, use theseinputs to tailor the spark advance, and fur-ther accept signals from a knock sensor, toretard the spark when detonation is detect-ed-see the sidebar "Knock, Knock?"

Motronic Engine ManagementIt may already have occurred to you that

most or all of these sensors already form

parts of the D/L/LH-Jetronic EFI systems.By'integrating the ignition and fuel injection"black boxes," we should expect, at the veryleast, to wind up with a lighter, simpler andmore compact arrangement than if each sys-tem were to operate completely indepen-dently. Merged systems like this are thebasis of Bosch "Motronic" engine manage-ment. The expected packaging advantagesare realized, but there is much more to itthan that, because the ideal amount of sparkadvance also depends on the air/fuel mixturestrength, and vice versa. Only when the igni-tion and EFI electronics are enabled to talk

to each other can both spark timing and mix-ture strength be optimized together toachieve the best compromise amongsttorque, driveability, economy, and emissionsunder all engine operating conditions.

Engines operating at light loads and highspeeds can tolerate a lot of ignition advancewithout detonating; indeed they work mostefficiently that way, as the existence of thevacuum advance unit on traditional distribu-tors confirms. The reason for this is that the

rate of pressure rise in the combustionchamber after ignition depends, amongmany other things, on the density of theair/fuel charge. With the small throttle open-ings that correspond to light load, the massof mixture trapped in the cylinder is small-its density is low. If, at that same high enginespeed, the throttle is opened wider, the man-ifold pressure will rise (that is, the vacuumwill drop), the density of the charge in thecylinder will increase, the flame will spreadmore rapidly once ignited, so less advance iscalled for.

Experiments have shown that the rate ofthe chemical reaction during the very earli-est stages of combustion is fastest when theair/fuel mixture is near stoichiometric, and

that it slows down quite distinctly witheither leaner or richer mixtures. Because

detonation is very much a time-sensitivephenomenon, optimum spark advance thusalso depends strongly on mixture strength.With most pump gasolines, a mixture slight-ly richer than stoichiometric (around theair/fuel ratio that gives maximum power)will generally increase the tendency to deto-nate, and so would call for less timingadvance. Paradoxically, the same engine islikely to tolerate more advance without det-onation if run either very rich or with mix-tures leaner than stoichiometric.

When a three-way catalytic converter isused, however, the mixture must be keptstoichiometric within very close limits forproper functioning of the converter, and thistight control is provided by an oxygen sen-sor (lambda sensor) feeding informationback to the EFI computer in closed-loopcontrol. All Motronic systems have a lamb-da sensor, so if this ensures that the mixtureis always stoichiometric within a few tenthsof a percent, there might seem to be no pointin taking mixture strength into account incalculatingsparktiming. This would be true ifthe system were always operating in closed-loop mode, but note that both L-Jetronic andLH-Jetronic revert to open loop operationduring cold starts, warm-up, full throttleoperation and sometimes, transiently, duringacceleration.

Cold Start and Warm-UpIgnition Timing

While the optimum ignition timing for anidling engine might be several degreesadvanced from TDC, in a stone-cold enginebeing cranked at very low rpm, any advanceat all may cause the piston to be driven back-ward, possibly damaging the starter drive,and certainly preventing the engine from

65

INPUT

-TDC

Throttle

Boost

rpm

MotrlonicI

con IrolI

OUTPUT

Timing

Boost control

In early versions of Motronic, the knock sensor/control function was physically separated from the main ignition/injection ECU. The prin-cipal remains the same.

66

Air temp

I

Kn<pckcon trol

I

Knock

starting. The very bottom end of the engine'sspeed range-cranking and idling-thuspresented an insoluble dilemma to enginedesigners in the days of mechanical distrib-utors, because the speed was so low that thecentrifugal advance mechanism could notrespond to the comparatively slight differ-ence in rpm between cranking speed andidle speed.

Faced with these same conditions,Motronic will retard the spark to approxi-mately TDC during slow cranking, thenimmediately dial in a few degrees ofadvance as soon as the engine fires (which

the ECU "knows" because the rpm rapidlybuilds, and the starter becomes disconnect-ed). As with L- and LH-Jetronics, a muchricher than normal mixture will be suppliedto a just-fired-from-stone-cold engine; theMotronic system goes further and providesmore spark advance, which will raise theotherwise too-low idle speed of a cold, rich-running engine. On the other hand, anengine being started at a higher temperaturewill be given a small amount of initialadvance during cranking, as this helps start-ing. Motronic makes this "judgement" basedon both temperature and cranking rpm.

Idle Speed AdjustmentsAlthough all Motronics are equipped with

some form of idle speed stabilization,whether the thermostatically controlled aux-iliary air bypass of the L-Jetronic or theECU controlled idle speed stabilizer valveof the LH-Jetronic, these devices are some-

what slow in response. The Motronic systemcontinuously and almost instantly "finetunes" variations in idle speed away fromthe ideal by slight adjustments in sparkadvance-a little more advance will speedup the idle; a little less will slow it down.This ability to tightly control spark advanceat idle reduces the amount of mixture rich-ness needed.

Mter startup and during the initial warm-up phase, until the lambda sensor becomeshot enough to provide meaningful signals tothe ECU, the system will operate open loopfor a certain length of time, according to thetemperature and time elapsed since starting.During this time, the spark advance and fuelinjector pulse time will be gradually adjust-ed toward the "nominal" values stored in the

ECU's memory, but always with each takingaccount of the other. An added subtlety hereis that the ECU will retard the spark, relativeto the nominal value for the circumstances,

in order to raise the temperature of theexhaust gasses and so aid a rapid warm-upof both the lambda sensor and the catalyticconverter.

Once fully warmed up, mixture strengthfor most driving is thereafter maintained atstoichiometry by the ECU, in concert withthe lambda sensor. Ignition advance will beconstantly adjusted according to airflow(and thus load), rpm, injected fuel quantity,coolant temperature, air temperature, theoutput of the lambda sensor, and, on enginesso equipped, signals from a knock sensor.The ECU also takes into account how quick-ly load and speed are changing. When therpm and airflow signals from the flow meterindicate part throttle acceleration, for exam-

pIe, the spark will be retarded slightly, toavoid the knocking that would otherwiseoccur during this transient phase if the tim-ing advance is just below the detonationlimit for the immediately previous (andimmediately following) steady-state condi-tions. But note that sudden changes inengine operating conditions might call forequally rapid changes in ignition timing, andthe ECU is quite capable of that. Suchabrupt changes in spark advance, however,would sometimes lead to a jerky responseby the engine, so the ECU, in fact, smoothesthe transition by spreading the adjustmentover a few engine cycles. At the same time,ignition dwell- the length of time that pri-mary current flows in the coil- is also con-tinuously optimized.

The way all this is achieved gives someidea of the blinding speed at which all theseelectronic calculations are carried out within

the ECU. Based on the firing of the immedi-ately previous cylinder and the rpm, the timeavailable until the next cylinder reachesTDC is calculated, and the appropriateamount of spark advance for the prevailingconditions is looked up in the internallystored maps. A suitable time interval for coilsaturation is computed next (with a correc-tion for the battery voltage-low voltagewill start the process earlier), and the currentto the primary side of the coil is turned on ata moment that anticipates its interruption aninstant later, when the ECU will open theprimary circuit, causing the spark. It is worthbearing in mind, here, that in an eight cylin-der engine turning 6000rpm, the intervalbetween two consecutive sparks will be just0.025 seconds! As an added feature to pre-vent overheating the coil, the primary cur-rent is shut off if the engine is stopped withthe ignition on.

Knock Sensor FunctioningUse of knock sensing allows the basic,

open-loop spark advance map in the ECU tobe biased toward a bit too much advance.

67

IgnitionAngledegreescrankshaftbeforeTDC

...:~_:'....

'---" ,~'c-.~

>'. \ r'::"""..,...",-__,_'-T--r---'1 r-~;~e )\leen ----

-,.

0

IgnitionAngledegreescrankshaftbeforeTDC

. --n./ .~-:~/---<.<...

"~~i~~i~;>,~~;;-~ ~/~. ',... ---~-r t"\),,,e,-. ---'1

.-" -__,_,-T--r

0

Some Motronic systems have two sets of ignition timing "maps." A typical one for premium fuel is at the top; below is one for regularfuel. If the knock sensor still reports detonation even though the ignition is retarded as far as "premium" map says it should be, then thesystem switches to the map for "regular" fuel. Only performance suffers, not the engine. (Robert Bosch Corporation)

When knock is detected, the spark is retard-ed a little while the circuitry "listens" again.If knock is still detected, the timing is retard-ed a little further. If not, the advance is

cranked up a little, until knock is againheard. Thus, the engine is always operatingon the brink of detonation, but the knockingis always held at a point where it is justdetectable to the electronics, but inaudible to

the human ear, and perfectly safe for the

68

engme.In some cases, two sets of internal map are

provided. One is programmed for regularfuel, the other for premium. If the ECUdetermines that the reduction of advance

needed lies well off the premium fuel map,it switches to the other one. Thus, the ownerof a car that calls for premium fuel can buya tankful of the nastiest, cheapest sort of reg-ular and suffer nothing worse than a reduc-

1 2 3

, , , , .

~

1. Permanentmagnet

2. Housing

3. Enginehousing4. Soft-ironcore

5.Winding6. Ringgearwith reference

point

Engine speed sensor-essential for the kind of precision control offered by Motronic systems. (Robert Bosch Corporation)

tion in perfonnance, resulting from theknock sensor/ECU backing off the advanceto keep the engine from making buckets-of-bolts noises.

Again, the speed with which all thisadjustment of spark lead happens is mind-boggling. There is always one cylinder thatknocks first, but because the ECU "knows"the crank angle at any instant, it "knows"which one it is, and can retard the spark forthat cylinder alone, while dialing up moreadvance for the next one! And note that even

though many other changes in spark timingare phased in comparatively gradually bythe ECU, the onset of knocking will be actedon instantly.

TriggeringHigh precision control of ignition timing,

as just described, demands at least equal pre-cision in detecting the crank angle. For thatreason, Motronic systems detect crankshaftangle directly at the crank, rather than froma camshaft or other half speed shaft, such asthe distributor drive, where gear (or chain orbelt) "backlash" can introduce errors, bothfixed and variable. In many cases, it is themovement of the flywheel itself that isgauged, with a magnetic sensor that "reads"the ring gear teeth going past. This providesa finely detailed rpm signal- there are a lotof teeth, and each one provides a little"blip"-so detailed, in fact, that the ECU isable to detect a change in speed after just afew degrees of crank rotation.

69

1 3 52 4

a.fullb.empty1.ventto atmosphere.2. accumulatorhousing3.spring4. rubberdiaphragm5. volumeofstoredfuel.

....

a

.-b

The fuel accumulator that helps damp out system pressure fluctuations and maintains some residualpressure when the engine is stopped to ease warm restarts.

To establishTDC on thenumber 1 cylin-der, a separatemagnetic sensor is usedwhich produces just a single pulse perenginerotation by reading,for example,asinglebolt head on the flywheelperimeter.On installations where this location is

impractical, a separate toothed wheel is pro-vided, often at the nose of the crank. In these

cases, a single sensor is used for both rpmand #1 TDC, the TDC signal being provid-edby a missingtooth on the rotatingwheel.

In either case, it is of course necessary todistinguish between TDC on the compres-sion stroke and TDC at the end of theexhaust stroke. This is achieved with a third

sensor that reads a timing mark on thecamshaft. Since the carn rotates at half crank

speed, this simply establishes which strokethe engine is on; the precision timing is leftto the crank sensor.

Fuel Sub-SystemIt is the integration of ignition and EFI

"black boxes" that characterizes Bosch's

Motronic systems. Since we earlierexplained that the fuel delivery aspect ofthese systems is essentially either L-Jetronicor LH-Jetronic, our attention so far has beenfocussed on the ignition aspects.Nevertheless, there are some points of inter-

70

RPM

6000 ::t80rpm

0

Fuel off

Time .Another feature of Motronic systems is a built-in rev limiter. Although engine rev limiting can be achieved by interrupting the ignition, thiscan cause rough running and backfires and be hard on the machinery. Cutting off fuel is smoother, and reduces emissions, too. Once thepreset rpm limit is reached, the cut-off rapidly oscillates between "fuel on" and "fuel off," with about +j- 80 rpm between the two states.

est in the fuel side of things, too.For one thing, many later versions of

Motronic inject sequentially, rather thanusing the "ganged" firing of the injectors inD-, L- and LH-Jetronics. Rather than injec-tors firing in pairs or all together, each cylin-der receives its dose of fuel at exactly thesame point in the intake stroke. The exactphasing of the start of injection, however,varies from one family of engines to anoth-er-the injection does not necessarily takeplace only while the intake valve is open.

While it requires more, (and more rapid)computations, and thus demands a morecomplex ECU, this approach has some

potential advantages. For one thing, theproblem of port wetting, discussed in earlierchapters, can be minimized or even elimi-nated. That reduces the amount of accelera-

tion enrichment needed, and so improvesboth economy and emissions. For anotherthing, the ganged firing of earlier systemscauses both a brief, transient drop in fuelpressure, and a large momentary electricalload. Even though the fuel rail on D-, L- andLH-Jetronics has an internal volume equalto many times the volume of fuel of oneinjection, and so the fuel pressure pulsationdoes not drive the mixture strength out ofwhack - it could be accommodated in the

71

'\j 4500Q)Q)p.,lfl

Q)3000

.,...,

QJ)

1500

internal "maps," anyway - it can cause dis-agreeable noise. (Motronic systems that doinject on the "ganged" basis fit a surge sup-pressor-a kind of hydraulic accumulator-at the upstream end of the fuel rail, to damp-en these pulsations.) Reducing the size ofthe electrical load "spike" reduces the peakelectrical consumption of the system andallows some components to be made small-er and lighter. Further, the system's responseto changes in engine speed and load can bemade more rapid - acceleration enrichment,for example, can begin with the next cylin-der, rather than having to wait for a full tworevolutions of the crank before fattening upthe mixture. Finally, each injector canremain open for much longer than if all werefired together. Because ganged injectorsopen only every other revolution of thecrank, the maximum number of crankdegrees they are open is something less than360 degrees. Individual, sequential injectorscan stay open for a bit less than 720 degrees.That permits injectors with a smaller flowrate to be used which, in turn, simplifies theproblem of metering very small amounts offuel at idle.

Injector PulseThe nature of the pulse that opens the

injectors is also different from that in the"fuel-only" systems. Rather than a single"on" pulse that persists for the duration ofthe injection, the injectors in LH-Motronicsystems are driven by a "stream-of-blips."The injector valve is first opened by a com-paratively large voltage "spike" from theECU, then held open by a sequence of fur-ther on-off blips that cycle rapidly. Theinjector never gets a chance to close, how-ever-it is held open continuously. The volt-age doesn't drop below 8V, and it will stayopenif the voltageis above6Y.Whileit hasbeen suggested that this is merely anapproach that makes the design of the elec-tronics more convenient, another reasonmay be that it reduces the total electrical

72

power consumed by an injector during acycle. Apart from a miniscule savings infuel- the engine doesn't have to work ashard turning the alternator-this may alsopermit smaller and lighter windings in theinjector solenoids, without the risk of over-heating them, in the same way that dwellcontrol avoids overheating the ignition coil.Note that this issue of having to limit the"duty cycle" (the ratio of on-time to off-time) is intensified if sequential firing isused and long periods of time are spent withthe engine at high power and thus demand-ing that the injectors remain open for longperiods.

Adaptive ControlA final refinement is "adaptive" control, in

which the ECU "learns" changes in theengine's condition and responds appropri-ately. Recall that in closed loop mode, theECU is continually adjusting the mixturestrength on the basis of signals receivedfrom the lambda sensor. Thus, the tendencyof an aging engine with numerous air leaksadmitting unmetered air to run lean will becorrected by the ECU detecting this on thebasis of the signals from the lambda sensor,and make appropriate corrections for it. Inopen-loop mode, however, such as duringcold starts, cold engine idling and full throt-tle acceleration, the ECU reverts to its inter-nal maps. If these maps were fixed for alltime, the engine would consistently run leanwhenever it operated in open loop. Given asufficiently large and subtle memory, theECU can "remember" that the enginealways needs a richer mixture than its mapscall for. With adaptive control, the ECUadjusts the values in the maps, for use inopen loop mode, on the basis of the amountof enrichment beyond the map values whenrunning closed loop. So much for regulartune-ups!

KNOCK,KNOCK?

In broad tenns, an engine running on a stoichiometric mixture will produce the most power for the least fuelwhen the spark is advanced to a point just short of where detonation or "knocking" occurs. Working at that opti-mum value of spark advance, however, is fraught with dangers.

When combustion goes haywire and the entire contents of the cylinder" go off with a bang," power is lost andfuel wasted because the violent turbulence that accompanies detonation scrubs the hot gasses against the interi-or surfaces of the combustion chamber, so much of the heat energy winds up in the exhaust or cooling system..Far worse, the rapid "spike" in cylinder pressure from this explosive combustion is potentially crippling to anengine. If knocking persists for long, piston crowns can-quite literally-have holes punched through them,spark plugs can have their side electrodes knocked clean off the shell and/or have their porcelain bodies cracked,rod and main bearings can be damaged, head gaskets can be eroded away.

In the days of mechanical distributors, the static setting of ignition timing had to be chosen conservatively,because the additional advance arrived at by the combination of mechanical and vacuum advance controls wasapproximate at best, and changed with wear of the mechanical parts. (Ever installed a new set of points in an oldervehicle, had it tend to knock slightly, but then marveled that it seemed to "heal up" after a time? That's becausethe rubbing block on the moveable half of the points wore down enough to retard the timing a few degrees.)

The higher precision of electronic triggering allowed engineers to work closer to the danger mark, but vari-ability in climate, fuel quality, driving habits, and engine condition still meant that a safety margin had to be heldin hand.Whatwasneededwas somedevicethat coulddetectthe incipientonsetof detonation- plus,of course,the ability to continuously and rapidly adjust the timing so that it was always at the ragged edge. The device isthe knock sensor.

While the internal working principle of knock sensors can have a variety of possible fonns, the one used byBosch (and many other manufacturers) depends on the piezoelectric principle. Some materials - in this case aspecial ceramic-produce an electrical voltage when they are strained, or defonned. Such devices have been usedfor microphones and in inexpensive cartridges for LP record players. As applied to the knock sensor, a rigid, mas-sive housing contains a lump of the piezoelectric ceramic, one face of which is attached to the housing, while theother face is attached to a smaller weight. When the housing is shaken, the "free weight" tries to dance aroundand so strains the ceramic, which accordingly produces a small electrical signal. We can arrange to shake thehousing by attaching it to the engine block.

Shaking of the outer housing, of course, occurs all the time the engine is running- not just because the engineis moving around on its mountings (that movement is too slow to excite any measurable signal, anyway), butbecause the engine block is vibrating from the fury going on within it. The nature of the shake, however, changessignificantly when knocking occurs, so the electrical signal put out by the sensor changes too. If that signal is fedto a device that can recognize the characteristic "sound" of knocking, then we have captured the infonnation weneed to control ignition timing so as to avoid the knock. (While Bosch categorizes its knock sensor as anaccelerometer, it is helpful, and not completely wrongful, to think of it as a microphone.)

The device that does the recognizing may be a separate electronic "black box," as on some early Motronics, orcan be integrated into the ECU, as on models ML3 and later. The characteristic frequencies the electronics arelooking for are in the range of 10,000-15,000 cycles per second-exactly the range offrequencies ofthe soundswe hear as engine knock.

Preventative MaintenanceBeforeattributingsomeoperatingfaultto

the fuel injection system, be sure the remain-der of the engine is in sound order. It is quitepointless to start to troubleshoot the injec-tion system if the spark plugs are years old,the rubber intake ducting is cracked, or anexhaust valve is burned.

Spark PlugsAs in the days of carburetors and point-

and-breaker ignition systems, the fIrst diag-nostic test should be to remove the plugs andexamine them - the removed plugs canreveal a great deal about engine condition.The insulator should be a light gray-to-tancolor. An insulator that is bone-white-orworse, blistered - indicates excessive lean-

ness, or perhaps a plug of the wrong heatrange; a blackened insulator may be theresult of a too-cold plug, or excessively richrunning, or may be a product of oil fouling,because of worn rings or valve stems, aplugged PCV valve, or even simply an over-filled oil pan. To distinguish between carbon(fuel) fouling and oil fouling, rub the plugagainst the heel of your hand-oil foulingwill leave a greasy smudge; carbon foulingwill not.

Any mechanical damage to the plug - acracked insulator, a broken side electrode-

implies detonation, which may have dam-aged much more than the plug. A plug that istruly wet with gasoline implies a non-firingcylinder that has continued to receive fuel.

Check Gap-Also, check the gap on

removed plugs. If the plugs have seen anysubstantial amount of service, the gap is sureto be larger than on a new, correctly gappedone. This widening of the gap results fromerosion by the hot gasses within the cylinder.An excessive plug gap can make unsustain-able demands on the ignition system - thevoltage required to jump the exaggeratedgap may cause the secondary (high-tension)voltage to rise so high that the spark seeksanother path to ground, perhaps punching amicroscopic hole right through the insula-tion on the plug or coil wires, leaving a leakpath to ground that remains even after theplugs are replaced.

Replacing Plugs-The replacement inter-val for spark plugs suggested by the factoryis likely to be highly optimistic; except forplatinum tipped plugs, they should bereplaced annually, as should the air and fuelfilters. Check the gap on new plugs beforeinstalling them even if they come pre-gapped, and take care when replacing thefuel filter that the act of removal and

replacement does not allow dirt to enter thesystem. High-tension wires should bereplaced every couple of years, likewise thedistributor cap and rotor, if applicable.

Routine Checks

Other routine service inspections thatshould be taken care of before going furtherare checking ignition timing and, on engineswithout hydraulic lifters, the valve clear-ances. Inspect all rubber air trunking forcracks and other sources of air leaks.

75

Assuming all is in order so far, the next stepis a compression check. This will revealworn rings or leaking valves. Compressionthat is uniformly down by even as much as20-30psi relative to the factory figure is notmuch to worry about, but variations betweencylinders of that much is cause for concern.

Gas Pains

Many fuel injection troubles can be avoid-ed, or at least long postponed, simply bypaying attention to the quality of fuel used,and where and when it is bought. While theowner's manual may make clear that regulargas of about 87 pump octane is suitable, andwhile the engine may not be able to takeadvantage of the higher octane (about 92) ofpremium fuel, the premium fuel from mostnational gasoline brands contains a moreaggressive detergent additive package thandoes their regular fuel. Clogged injectors areone of the more common causes of griefwith fuel injection systems-all systems,not just Bosch. Higher detergency of the fuelhelps prevent these faults. Even supplement-ing a standard diet of regular with a tankfulof premium every few weeks helps.Aftermarket detergent additives may also beeffective.

Running out of gas is a pain, argumentenough for following the advice implicit inthe old saw that "it costs no more to drive

around with a full tank than a near emptyone," but running most of the time with thegauge showing 1/4 tank or less is especiallypoor practice with fuel injection systems.The larger the air space above the fuel in thetank, the greater the amount of water thatcondenses out of the air, and the greater thesusceptibility of steel and iron componentsin the system to rusting. Apart from thedirect consequences of this corrosion, tinyflakes of rust can play havoc if they get intothe system.

Note, too, that it is a good idea to keep aneye on when your habitual gas station getsits deliveries. The replenishment of the sta-

76

tion's underground storage tanks stirs up rustand sediment in their tanks that may wind upin yours. Better to fill up the day before orthe day after.

"No User Serviceable Parts Inside»

No matter what the results of the diagnos-tic tests described below, bear in mind thatin most cases the only remedy for somethingout of spec is component replacement. Partsthat cannot be repaired or adjusted include,where applicable: pressure regulator; accu-mulator; injectors; cold start injector; auxil-iary air bypass (idle speed stabilizer); fuelpump (except for the check valve, which isreplaceable); and all sensors.

Bear in mind, too, that without the highlyspecialized equipment to which servicetechnicians in dealerships have access, thereis absolutely nothing you can do with anyelectronic control modules; you cannot eventest them. All you can do if an electroniccontroller is suspect is to systematicallyeliminate all other components as potentialculprits and, as a final resort, replace thecontroller.

Troubleshooting TestsDespite the dismayingly long list of things

that cannot be done, there are neverthelesssome troubleshooting procedures that can behelpful. These should be carried out in a log-ical way, and with an understanding of howthe system is supposed to work and whatdifferent components do in various circum-stances. For example, if an engine startsreadily from cold and runs well while firststarting to warm up, but then runs progres-sively richer as it reaches operating temper-ature, guzzling gas and spewing blacksmoke, one plausible culprit is a cold startinjector stuck open. In the same way, if theidle speed is way too high when warm, ortoo low when cold, one of the first places tolook might be the idle speed stabilizer/auxil-iary air bypass. In general, if only one part ofthe engine's operating regime is affected,

WARNING!Gasolineis highlyflammableand potentiallyexplosive.It can be ignitedby an

electricalsparkor by contactwithhot engineparts.Use extremecare whenwork-ing on anyengine'sfuel system.Workonly in a wellventilatedarea,ban smokingor anyopenflamefromtheworkarea,andensurea fullychargedlargecapacityfIreextinguisheris close to hand. Many fuel injectionsystem circuitsremain underpressureevenwhenthe engineis stoppedand the fuelpumpis deactivated;beforelooseninganyfuel fItting,wrap a rag aroundit.

look first at components whose function isdirectly related to operation in that regime.

Air Meter (vane type)As noted, the airflow meter is not service-

able; all you can do is check for smoothmechanical action and test the electrical out-

put of its potentiometer. Test for free,smooth movement simply by moving thevane with your fInger. After unplugging theelectrical connector, apply one probe of anOhmmeter (or multimeter set to "Ohms" or"resistance" scale) to a clean chassis groundand contact each terminal on the meter in

turn with the other probe. In every case, themeter should read an open circuit (infIniteOhms).

The "voltage-in" and "voltage-out" pinsdiffer from one model to another; the manu-al for your vehicle may identify them. Inevery case, however, there will be one pairof pins that show a varying resistance as thevane is moved. The exact numbers don't

matter much; what is important is that theresistance should vary smoothly as the vaneis slowly deflected by hand.

Air Meter (Hot-WIre Type)There is nothing serviceable or adjustable

in the hot-wire air mass sensor; all you cando is confirm its functioning. If it doesn'twork, it must be replaced.

The wire in the sensor is so fIne that, with-out ideal lighting, it is easy to mistakenlysuppose that it is broken or missing. Recallthat a "bum-off" cycle is provided, that feeds

a large amount of current to the wire tovaporize any dirt deposits on it. The wireglows red hot during this brief (about onesecond) cycle, which makes it plainly visi-ble.

With the engine running and the meter ori-ented so it is possible to sight through it withits wire harness and trunking to the throttlebody still connected (it is OK to detach thehousing and move it around, as long as youdon't damage the trunking or wiring), rev theengine to at least 3000 rpm and shut it off.(For safety reasons, the bum-off cycle iscancelled if the engine has not exceeded3000 rpm since the last time the ignition wasshut off). After three or four seconds, thewire will glow for about one second. Whilethis doesn't confIrm that the meter is work-

ing correctly, it does prove the wire is OK.There are fIve wires connected to the hot-

wire sensor. One is a ground; one is voltage-in, one is voltage-out, and the other two areinvolved only in the bum-off cycle. Whichis which depends on the specific model, andshould be specifIed in each vehicle's manu-al. Failing that, the above test (you mayhave to repeat it several times, working thisway) should identify those last two by plac-ing the probes of a high-impedance digitalvoltmeter/multimeter on pairs of terminals(peel back the insulating boot on the con-nector) until you fInd the pair that are "live"and "ground" during the bum-off. DO NOTUSE AN ORDINARY ANALOG (NEE-DLE-AND-DIAL) VOLTMETER; there issevere risk of damage to theECU.

77

Boschintermittent

injection

Domesticthrottle-

bodyinj ection

Typicalcar buretor

system

Boschcontinuous

injection

One of the minor reasons why fuel injection is superior to carburetors is because forcing the fuel through small orifices at high pressuredoes a better job of vaporizing it than sucking it through small orifices with a small pressure difference. A potential downside for servic-ing is an increased fire risk. Many circuits contain fuel under residual pressure, even when the engine is stopped.

With one voltmeter probe grounded, thevoltage-in terminal will show battery volt-age (nominally 12 volts) when the ignition isswitched on. WITH THE IGNITION OFF,the ground will show zero Ohms (no resis-tance) to ground. (Never apply anOhmmeter to a powered circuit.) By elimi-nation, the remaining terminal must be volt-age-out.

The voltage between ground and the volt-age-out terminal will typically vary from abit more than 2 volts with the engine idlingto a bit less than 3 volts at about 3000-

4000rpm. Again, the factory specs may be in

78

the manual for your vehicle, but the exactnumbers are less important than seeing asmooth voltage increase pretty much in pro-portion to engine speed.

Fuel Pressure Checks

Unless the engine has been stopped for avery long time (hours), there is likely to beresidual pressure in the system. Beforebreaking into the fuel plumbing to attach apressure gauge, this residual pressure mustbe relieved. The simplest way to do this is topull the fuel pump fuse, thereby disabling it,then run the engine until it stalls for lack of

fuel. For pressure tests, a pressure gaugewith a capacity of at least SOpsiis needed,together with hoses and fittings to connect it.

Some engines have an extra port for con-venient connection of a pressure gauge.Failing that, the fuel line to the cold-startinjector can be pulled and the connectionmade there. A specific figure for fuel systempressure will be listed in the manual for yourparticular vehicle, usually about 3Spsi,although some high powered engines mayrun at about 44psi. Note, however, that this"factory" figure is measured with the enginestopped.

An alternative test procedure that permitschecking the functioning of both the fuelpressure regulator and the pump is to gaugethe pressure with the engine idling. In thatcase, the pressure should be somewhere inthe neighborhood of 30psi, or a bit less.Again, this applies to most engines; highhorsepower ones that call for, say, 44psiwith the engine stopped will usually show abit less than 40psi when checked idling.

Recall from the previous chapter that thefuel pressure regulator has a vacuum hoserunning to the intake manifold, so manifoldvacuum can act on the diaphragm within theregulator. This allows the regulator to main-tain fuel pressure in the rail at some constantamount higher than the pressure in theintake manifold, no matter how manifold

vacuum may vary. If the vacuum hose is dis-connected at the regulator (and plugged),then with the engine idling the gauged pres-sure should have risen to the "factory" fig-ure.

High Pressure-If the system pressure istoo high, the problem is either a defectivepressure regulator or a restricted return linefrom the regulator to the tank. This last canbe checked by disconnecting the return lineat the regulator and attaching a test line lead-ing to a suitable container. If the pressurereturns to normal, the original line isobstructed; if not, then the regulator isdefective. The pressure regulator on inter-

mittent systems-except for now somewhatrare D-Jetronic-is not adjustable.

Low Pressure-If the pressure is too low,on the other hand, there are numerous possi-bilities. First there are the obvious things: Isthere fuel in the tank? Are there any visibleexternal leaks? Is any part of the supply linefrom the tank to the fuel rail dented partlyshut? Less obviously, the tank vent may beobstructed, forcing the pump to attempt to"implode" the tank. This can be checked byremoving the gas cap. The fuel filter may beclogged, as may the (usually fitted) in-tankstrainer-a kind of wire mesh "sock" that

filters out rust flakes and other large piecesof debris.

Other possible causes for low pressure arelowIno voltage at the pump and a defectivepressure regulator. Confirm that the pump isrunning. In a reasonably quiet environment,you should be able to hear it. If the pump isnot running, check the electrical supply tothe pump, starting with its fuse. If the fuse isOK, check for electrical power at the pump.A voltmeter with one probe applied to one ofthe fuel pump terminals and the other to agood chassis ground should show batteryvoltage (near 12 volts) at one pump termi-nal, and zero at the other. (You will have tofold the rubber boot around the connector

out of the way to gain access to the electri-cal terminals.) Low voltage at the pump islikely the result of corroded terminals or abad ground. Check the pump ground.

Pump Does Not Run-If there is ade-quate power to the pump but it does not run,the pump is defective and will have to bereplaced. If the pump seems to be in order,pinch shut the return line from the pressureregulator to the tank. If the pressure rises tostandard values, the regulator is defective; ifnot, and the alternative faults listed above

are excluded, it is likely the pump after all.Residual Pressure-Recall from the sys-

tem description in Chapter 3 that the pumpon intermittent systems is equipped with acheck valve that keeps the fuel lines full and

79

pressurized even when the engine isstopped. This permits quicker restarts, andhelps to prevent vapor lock. With the pres-sure gauge connected as described above,run the engine briefly, then shut it off (orenergize the fuel pump with the enginestopped). The pressure should not dropbelow about 14 psi for at least 20 minutesafter shut-down. A loss of residual pressuremay be caused by an external leak, one ormore leaking injectors, a defective pressureregulator, or the fuel pump check valve.

To check both these last two, repressurizethe system and, immediately after shuttingdown the engine (or disabling the pump),pinch shut the supply line from the pump tothe regulator. IT the residual pressure nowholds, the regulator is defective; if not, thepump check valve is leaking.

ITall this fails to stem the drop in residualpressure, about the only remaining candi-dates are one or more defective injectors,and the cold start valve.

Injector Leakage, Flow, PatternWith system pressure relieved, remove the

cold start injector (cold start valve), butleave its fuel supply connected. Place theinjector in a suitable container and repres-surize the system; the injector should notseep or drip fuel. ITit does, it is defective.

Remove the injectors and temporarilyplug the holes they came out of. Set eachinjector into the mouth of a graduated ves-sel.A reasonable degree of accuracy is need-ed here, so graduated cylinders (available atany laboratory supply outfit, or ask yourdruggist) are preferable to domestic measur-ing cups, etc. It should be pretty obvious thatplastic vessels are preferable to glass ones!

For safety's sake, disable the primary igni-tion circuit by disconnecting the connectionfrom the battery to the coil. Now run thepump; no fuel should flow from the injec-tors. Look also for seepage around theseams in the injector bodies.

Next, remove the spark plugs and crank

80

the engine for about one minute, allowingthe injectors to discharge into the graduatedcylinders. The amount of fuel discharged byeach injector should be the same, within 10percent or less. Also, observe the spray pat-tern from each. It should be cone shaped andsymmetrical. Partial clogging may result ina spray pattern that is lopsided, or a dis-charge that is hardly atomized at all- morelike a stream than a spray-or may simplyreduce the rate of delivery.A slight degree ofasymmetry in the spray pattern is accept-able, as long as the delivery volumes match,but pronounced lopsidedness or a streamrather than a spray requires that the defectiveinjector(s) be replaced.

Basic Mixture Strength/CO AdjustmentAfter about 1987, LH-Jetronic and

Motronic systems make no provision foradjusting the mixture strength/exhaust COlevel- none is needed. On earlier L-models,

those having a moving vane air meter, thebasic mixture strength can be adjusted byturning a screw. This screw is accessiblethrough a small hole on the top of the airmeter housing, usually closed-off with an"anti-tampering" plug. After the anti-tam-pering plug is pried out, turning this screwclockwise richens the mixture; counter-clockwise leans it. On those LH-Jetronics

that have provision for adjustment, theadjuster is located on the side of the airmeter, again under an anti-tampering plugabout 7/16" diameter. Note that while the

adjustment is always carried out at idle, itseffect is felt throughout the engine's operat-ing envelope.

Note, too, that on systems with a lambdasensor, the lambda sensor/electronic controlwill attempt to bring the mixture back to thestoichiometric value no matter what you dowith the COlbasic mixture adjustment. Thefix is simple-before carrying out theadjustment, disconnect the lambda sensor,obliging the electronic control to operateopen loop.

Engine exhalistbefore convertor

Three-way catalytic conver-tors are essential to meetcurrent emissions limits. Butconvertors can only do theirjOb if the air:fuel mixture fedto the engine is held veryclose to stoichiometry-thechemically correct ratio. Andto achieve that, a lambda(oxygen) sensor is essential.Its electrical output variesvery strongly right around thepoint of stoichiometry.

//

/I

II

""-

./'

//

""-//""-

/

NO X""-

"- "\

""-

""-

""

""- HC""-

""

'----'-.

0.90

13.2

0.95

14.0

1.00

14.7

1.05

15.5

1.10

16.2

A

A/F ratio

Lamb dacontrollimits

Tailpipe exhaustafter three-way convertor

NOx......... '-

HC----

--

0.9013.2

0.95

14.0

1.0014.7

1.05

15.5

1.1016.2

A

A/F ratio81

0.98 0.99

Air-fuel

1.00 1.01 1.02

ratio (Lambda)

Most folks don't have a CO meter, but many have a multi-meter. It is tempting to gauge exhaust CO on

the basis of the output of a lambda (oxygen) sensor, but beware that the sensor's output sags graduallywith age, so simply aiming for about 500 millivolts may not do the trick. The shape of the curve is reli-able, however, so by adjusting the mixture both rich and lean, you can establish the upper and lower lim-its of the sensor's output range. Aim for the middle.

An exhaust gas analyzer (CO meter) isalmost essential to perform this adjustment.Before you start, however, set the idle speedusing the idle air bypass screw. Note, too,that the CO meter has to "read" the exhaust

gas upstream of the catalytic converter-remember, the converter is doing its best to

oxidize the CO to CO2' There is usually aport or pipe or fitting on or near the exhaustmanifold for this purpose.

With the engine fully warm and an appro-priate idle speed set, adjust the mixture untilthe meter reports a CO value correspondingto the figure on the EPA placard in theengine compartment. If the placard is miss-ing, aim for around 0.6 percent. Without a

82

CO meter, about all you can do is adjust forthe leanest setting that still provides asmooth idle, then turn the adjuster 1/4 turnclockwise (richer).

If a CO meter is available, then on lamb-

da-equipped engines both correct mixturestrength and correct functioning of the lamb-da sensor and the electronic control is

absolutely confirmed if the CO reading isthe same with the lambda sensor connected

(closed loop) and disconnected (open loop).

Auxiliary Air ValveL-Jetronics and early LH-Jetronics use an

auxiliary air valve to provide the extra air acold engine needs to maintain an idle speed

1000

>- - New sensor

S 800--- Aged sensor

-----------+J --

\"600 \+J

\ I

0 \\

H 400 \0 \UJ \

\\

Q) 200 \UJ "- " '"--

high enough that the engine will not stall.Asdescribed in Chapter 3, this is simply a smallrotary disc valve that allows air to bleed.around the nearly closed throttle (it is some-times called the "auxiliary air bypass"). Theposition of the disc is governed by a bi-metallic coil. When the coil is warm, the

valve is completely closed; when very cold,it is fully open, with a smooth and gradualtransition between the two as the enginewarms up. In very cold weather the auxiliaryair bypass valve can take eight or ten min-utes to move from fully open to fully closed.

The valve is not serviceable, but its condi-tion can be initially diagnosed by symptom:An idle that is appropriate when cold butexcessively high when warm points to thisvalve being stuck partly or fully open.Conversely, if the warm idle seems right butthe engine requires some throttle opening toavoid stalling when cold, then the valve maybe stuck closed.

If the valve is suspect, check, first, bypinching shut the connecting hose. On acold engine, this should drop the idle speed;on a hot one it should make no difference. If

the valve fails this test, sight through thevalve. Because the valve is powered as longas the ignition is on, the heating elementshould completely close the valve within tenminutes. Ensure that 12 volts is getting tothe valve, allow time for it to warm up, thenlook through to confirm the valve is closed.To confirm the valve is opening when cold,pop it into a freezer for ten minutes andagain sight through it- there should be aclear, round passageway through it.

Idle Speed RegulatorLater LH-Jetronics and all subsequent

Bosch intermittent injection systems modu-late the auxiliary idle air bypass in a differ-ent way. It remains in principle a valve thatbypasses more or less air around the throttleplates, and its effect is most significant dur-ing warm up. However, because its moresophisticated control enables it to maintain

an appropriate idle speed, irrespective oftemperature or engine age and condition, itis renamed an "idle speed regulator," orsometimes "rotary idle actuator."

In this device, the moveable valve that

varies the size of the air opening is driven bya component that looks like an electricmotor. Indeed, in construction it essentiallyis, but in action it never turns more than 90degrees, its electrical components receivingpulses from the ECD that cause it to "dither"back and forth slightly around an averageposition that gives the idle speed pro-grammed into the ECD.

Analogous to the function of the electro-magnetic injectors themselves, the positionof the valve, and thus the amount of air thatpasses, depends on the ratio of on-time tooff-time- its "duty cycle," in other words. Along duty cycle-more on-time than off-drives the valve further open; a shorter dutycycle closes it further. This duty cycle can bemeasured with a dwell meter, set on thefour-cylinder scale.

The exact values under various circum-

stances vary considerably among differentvehicles, so check your vehicle's specs in itsmanual. As a general guide, the duty cyclewith a warm, idling engine will read some-where around 30 degrees on the dwell meter.Adding a fair-sized load, such as by turningon an electric rear window defroster, shouldhave little effect on the idle speed, but theduty cycle should increase somewhat.

To simulate cold start operation, discon-nect the engine temperature sensor. Disablethe fuel pump (pull the fuse), and crank thestarter. Dwell should be dramaticallyincreased; expect about twice the idling fig-ure. A defective idle speed regulator cannotbe repaired; it must be replaced.

83

There are a great many reasons whyauto manufacturers may have beenreluctant to wholeheartedly em-

brace intermittent EFI after it first

appeared-in D-Jetronic form-in 1967/68.One of them may have been cost. While itis impossible to know what it costs Boschto manufacture an intermittent EFI system,it is possible to gain some general indica-tion of the OEM cost of the system to automanufacturers. As noted on Chapter 3, theD-Jetronic intermittent EFI system firstappeared as standard equipment on certainmodels of the Volkswagen type 3 1600.Soon after, however, the system becameavailable as an option on some other modelsat a cost of $250-$300. Considering that a1968 VW 1600 was priced at less than$2,500, this is a rather hefty premium.

Another reason may have been simply thatit was an essentially brand-new andunproven technology. Although some autoexecutives and engineers surely were umea-sonably conservative (and some remain so!),it is not necessary to assume they were all ahidebound bunch of stuffed shirts in order to

understand why they might be reluctant togamble with their own reputation in order tocement that of Bosch. The "mere" task of

training thousands of mechanics at dealer-ships around the world in a new and myste-rious technology is a daunting prospect initself, and an expensive one. (A colleagueexperienced in such matters estimates thatthe cost of merely translating a 100-pageservice manual from one language to one

other could amount to $30,000, in today'sdollars, and VW, for example, deals world-wide in more than 50 languages!)

K-JetronicWhatever the reason for reluctance, there

was evidently motivation in some quartersto find an alternative to the carburetor. . .and one that didn't have dozens of wires

coming out of it. Thus, despite the fact thatby eight years after its introduction theJetronic intermittent EFI system was welldeveloped and widely admired, Bosch intro-duced a purely mechanical gasoline injec-tion system, in 1973. First used on the 1973Porsche 9UT, this "K-Jetronic" gained asignificant following, and wound up onsome vehicles such as the 6.9 liter Mercedes

Benz, where the cost of the alternative sys-tem was, if not irrelevant, surely not theoverriding consideration.

Basic System Operation & ComponentsThe "K" in K-Jetronic stands for kon-

tinuierlich, the German word for continu-ous. This sets out the first obvious difference

between K-Jetronic and the systems wehave been describing so far: the injectors onthe K-system spray continuously, rather thanintermittently. The quantity of fuel deliveredper unit of time thus depends solely on therate of fuel delivery to the injectors.

The regulation of fuel delivery to theinjectors takes place in the mixture controlunit, which comprises two parts: an airmeter and a fuel distributor.The general pur-

85

Reduced cost could not have been a major reason that Mercedes Benz chose a K-Jetronic system for their 6.9 liter (425 cu in) V8! Thisinstallation used the comparatively rare downdraft air meter; most K-systems have an updraft meter. (Daimler-Chrysler Archive)

pose of the air meter should be obvious: toprovide a suitable air/fuel ratio, the rate offuel delivery appropriate for the prevailingconditions depends on the rate of airflow, soit is necessary to measure that airflow. Theoutput from the air meter, in turn, acts on thefuel distributor to modify the amount of fuelfed to the injectors. All operations are pure-ly mechanical.

Air Meter-Although the physicalarrangements are different, the air meter-ing on K systems resembles that of the

86

L-Jetronic in that they use a moveable vaneor flap within a housing through which allthe engine's intake air is drawn. The flapamounts to a circular plate on the end of alever and is positioned across a conicalopening -like a funnel- at the entrance tothe housing. According to the rate of airflowthrough the funnel, the flap is deflected to agreater or lesser degree. so the position ofthe plate is a measure of the airflow.

The weight of the plate and the lever isbalanced by either a counterweight or, on

Components of the K-Jetronic system. (Robert Bosch Corporation)

some later versions, a light spring, so verylittle force is required to lift the plate. Asincreasing airflow raises the plate, it takesup a new position within the funnel, so thesize of the gap around the perimeter of theplate- the space through which the intakeairflows-depends on the details of themulti-angled taper of the walls of the funnel.(The "updraft" configuration just describedis most common, but some K-Jetronics, gen-erally those used on V6 and V8 engines,have the metering apparatus upside down, inwhich case the plate is pushed downward bythe air.)

In the design of the K-Jetronic, the quanti-

ty of fuel injected is a direct, one-to-onefunction of the travel of the plate-a dou-bling of its movement from the rest (closed)position will double the rate of fuel flow. Ifthe funnel had a simple, constant angletaper, the quantity of fuel delivered wouldalso be directly proportional to the rate ofairflow, and the mixture strength would thusbe the same at all rates of airflow. We have

already seen, however, that this is not appro-priate for all circumstances. Accordingly,the funnel is shaped with multiple tapers sothat the relationship between the position ofthe flap and the airflow space around it isnonlinear. Usually, this means that the fun-

87

a.Sensorplatein zeropositionb.Sensorplateinoperatingposi-tion1.Airfunnel

2.Sensorplate3.Reliefcross-section

4. Idlemixtureadjustingscrew5.Pivot6.lever

7.leafspring

1 2 3 4 5

a

7 6

b

Updraft airflow sensor. (Robert Bosch Corporation)

Increasing airflow "floats"the sensor flap upward (onupdraft systems), towardthe larger end of the intakefunnel. This opens up thegap between the sensorand the funnel.

88

nel tapers out slowly at fIrst, then growssteeper, then the angle becomes more shal-low again.

The effect of this is that a small change inairflow at the low flow rates encountered at

idle (and just off idle) requires a compara-tively large movement of the vane. Becausethe rate of fuel flow depends directly on thevane position, this provides the slightenrichment needed for idling. As the rate ofairflow speeds up to a point that correspondsto normal, light load driving, the vanemoves into the wider angled portion of thefunnel. Here the same amount of travel of

the vane will give a more rapid increase inairflow area, so the vane moves less for agiven increase in airflow, and the mixture isaccordingly leaned out slightly. At maxi-mum flow rates, the vane is pulled into themouth of the tunnel where the taper dimin-

g

ishes again, so the increase in flow area fora given travel of the vane will be less, forc-ing the vane to move further to provideenough flow area, and so richening the mix-ture for large loads and high speeds.

Acceleration Enrichment-As on

L-Jetronic systems, accelerationen r i ch men tis achieved automaticallyby "overswing" of the metering vane. A sud-den increase in airflow as occurs when the

driver rapidly opens the throttle will cause

1

2

3

the metering vane to quickly lift upward,and assume some new position. The inertiaof the vane and lever assembly, however,assures that it will initially "overshoot" thatfinal position. Because the rate of fuel deliv-ery depends on the position of the vane,there will be a momentary enrichment tomeet the needs of the accelerating engine.

To prevent damage to the metering vane inthe event of a backfire, "wrong way" airflowis allowed to push the vane past its rest

The height (h) of sensorplate lift needed to obtaina certain flow area (g)depends on the rate oftaper of the funnel.

A funnel with multiple tapersallows a comparatively largeamount of sensor plate liftaround idle, which provides aslight idle enrichment, and asimilar richening effect atmaximum flow rate, yet witha taper in between the twothat gives more rapid vane liftfor a given airflow increase-and so a slight leaning-under light load, mediumspeed operation.

89

7

1.Intakeair

2. (antralpressure3.Fuelinlet

4.Meteredquantityoffuel5. (ontrolplunger

6.Barrelw/meteringslits7. Fueldistributor

Within the air meter, the sensor plate pushes upward on the control plunger through an intermediate lever. Plunger movement is alwaysupward for more fuel, so on downdraft sensors the pivots of the intermediate lever are reversed. (Robert Bosch Corporation)

(engine stopped) position, to a widening inthe funnel that provides enough flow area tovent the backfire without hann. The finallimit of travel in this backwards direction is

established when the metering vane contactsa rubber bumper.

Clearly, the shape of the funnel determinesthe variation in mixture strength under dif-ferent operating conditions. Because of thedifferent breathing characteristics of differ-ent engines, the exact contours of the funnelvary from one installation to another.Indeed, the profile of the funnel is themeans by which the design engineers tailor

90

a K-Jetronic system for any particular instal-lation. To learn just how the movement ofthe vane affects the fuel flow rate we have tolook inside the fuel distributor.

Fuel Distributor-The regulation of fuelquantity takes place in the fuel distributor.The key component here is the controlplunger, which slides in a bore within thecast iron housing of the fuel distributor. Thebottom of this plunger contacts the meteringvane lever, and is pushed upward by thelever in proportion to the extent the airmetering vane is deflected by the intake air-flow. (The control plunger movement is

a

1

2

3

4

..... ... 5........-

6

b c

Barrel with metering slits and control plunger (a) engine stopped (b) part throttle (c) full throttle. (1) control pressure; (2) control plunger;(3) metering slit in barrel; (4) "control" edge of plunger; (5) fuel in from main pump; (6) barrel.

always upward for more fuel, no matterwhether the air meter is an updraft or down-draft type. On downdraft versions, the ful-crum for the air vane lever is moved to the

opposite side of the plunger to reverse theinternal action.)

The plunger looks a bit like the moveablepart of a "spool valve," such as you wouldfind in an automatic transmission - it is

dumbbell-shaped, with a thin section in themiddle and fat ends, with sharp shoulders atthe transitions. The bore that the plungerslides in is lined with a precision-machinedsleeve (Bosch terms this sleeve the barrel)that has a number of ports cut through theupper part of its wall, in the form of verynarrow (about 0.008 inch) vertical slits.With the plunger at the bottom limit of itstravel, corresponding to zero airflow- andthus engine stopped-the "fat" upper por-tion of the plunger completely blocks theseports. As the airflow vane lever pushes theplunger upward, the control edge of theplunger- the sharp shoulder at the meetingof the large and small diameters-progres-sively uncovers an ever greater length of

The position of the control plunger relative to the extremely narrow (about 0.008in) slits in the metering barrel controls the rate of fuel flow in the K-Jetronic sys-tem. (Robert Bosch Corporation)

these slits. With the plunger pushed all theway up, corresponding to maximum lift ofthe metering vane and thus full power oper-ation, the metering slits are fully exposed.

Fuel Flow-Fuel enters the fuel distribu-

tor through a set of ports located so they arealways exposed by the reduced diametercentral section of the plunger. Raising the

91

to control-pressureregulator

t

to fuelinjector

1.

con trol

pressure

fuel in Dfrom pump UJ .

systempressure

differential-pressurevalve

flowrestrictor

Differential-pressure valves-one for each injector-maintain a constant pressure drop across the metering slits, even as the slit areachanges because of control plunger movement. The thin metal diaphragm separating the upper and lower halves of each valve "bulges"in response to changes in the pressure difference between supply-side and delivery-side, tending to block or open up the open end of thenearby tube feeding the injector. This does NOTvary fuel quantity; it just maintains a constant pressure drop.

plungerthus allowsfuel to flowout throughthe meteringslitsand on to the injectors,ina quantitythat is proportionalto the heightof theplunger.(Insomerespects,thisvague-ly resemblesthefuelmeteringarrangementsof theRochestercontinuousPI mentionedinChapter 2, but note that fuel escapingthrough the Rochester's spill ports isreturnedto the tank; it is the remainderofthe pump output that goes to the injectors,oppositeto the K-Jetronic'ssystem.)

At fIrst glance it might seem that thisarrangementis all that is neededto provideaccurate fuel metering- the area of slitexposed is linearly proportionalto controlplungerlift,whichis linearlyproportionaltometeringvane lift,which,in turn, is propor-tional to airflowrate, with mixturestrength

92

corrections for different operating regimesprovided by the varying taper of the airintake funnel. Alas, it is not that simple.

Differential Pressure Valves-Fuel pass-ing through the metering slits is propelled bythe pressure difference between the supplyside-below the control edge of theplunger-and the delivery side, outboard ofthe slits. For a given exposed area of slit, thefuel flow rate will be proportional to thispressure difference. At the same time, for agiven pressure difference, the fuel flow ratewill be proportional to the exposed slit area.

The awkward complication is that, with-out some other mechanism to perform a cor-rection, the pressure difference across theslits varies according to how far open theslits are! If only a small length of slit is

exposed, there will be a large pressure dif-ference between supply and delivery sides; afully open slit will experience a lower pres-sure difference. Thus, while there is a one-

to-one relationship between vane movementand plunger lift, and between plunger liftand exposed slit area, there will not be aone-to-one relationship between exposedslit area and fuel flow rate. The greater the

exposed slit area, the less the pressure dif-ference, so while an increase in slit area willflow more fuel simply because the hole isbigger, the increase will be greater than pro-portional because there is now less pressuredifference between the inside and the out-side of the barrel to resist the flow, so the

mixture strength will tend to become richerwith every increase in airflow through themeter. To correct for this nonlinearity, thefuel distributor also contains a number of

differential pressure valves, whose functionis to maintain a constant pressure dropacross the slits, no matter what the rate offuel flow through them.

Valve Construction-Each differential

pressure valve comprises two chambers-an upper and a lower-separated by a thin,flexible metal diaphragm, backed by aspring. There is one valve for each injector,and thus one for each engine cylinder. Theupper chamber of each valve communicateswith one slit in the barrel (there is one barrelslit for each injector), while each lowerchamber is connected to the annular spacesurrounding the slim central part of theplunger and thus is exposed to the same,constant supply-side fuel pressure. Fuelemerging from each slit flows through theupper chamber of its dedicated differentialpressure valve on its way to the injector.

The outlet from the upper chamber of eachdifferential pressure valve is located justabove the diaphragm separating the twohalves of the valve, so upward deflection ofthe diaphragm tends to restrict the outflow,while downward deflection provides a

greater passage area. It is important to rec-

ognize that this apparent "throttling" of theflow through the valve does not affect thequantity of fuel delivered - that is estab-lished by the exposed area of the meteringslits. All that the movement of the

diaphragm does is to maintain a constantpressure drop across the metering slit itserves. Fuel flows out from each differential

pressure valve to the injector associated withit at system supply pressure minus, typical-ly, about 1.5psi, the difference being a resultof the force of the spring acting on the valvediaphragm.

Control Pressure and Its RegulationIn reading the description of the air meter,

above, it may have occurred to you thatthere is a component missing. Without aspring, or some other mechanism to return itto its rest position, even the smallest rate ofairflow would immediately drive the meter-ing vane to the limit of its travel, and itwould stay there, with the control plungershoved fully up into the maximum fuel flowposition. The force that tends to drive theplunger - and thus the metering vane - backdown is not a spring, but rather the closelyregulated pressure of fuel acting on the topof the plunger.

This pressure is neither the full systemsupply pressure, nor the very slightly lowerpressure prevailing in the upper chambers ofthe differential pressure valves, but rather astill lower pressure that Bosch terms thecontrol pressure. This control pressure act-ing on the plunger top produces a downwardforce that counteracts two upward actingforces-a minor one is atmospheric pressureacting on the exposed lower end of theplunger; the major one is the force on theplunger exerted by the metering vane, via itslever.

The position of the plunger in its barrel isthus the result of a sort of hydraulic balanc-ing act between the force of the incoming airacting on the metering vane and the controlpressure acting on the plunger. An increase

93

a

a.Withenginecold

b.Withengineat operatingtemperature

1.Valvediaphragm2.Return3. (ontrol pressurefrom mixturecontrolunit

4. Valvespring

5. Bi-metalspring

6. Electricalheating6

b

1

5 4

The control pressure regulator, previously termed the "warm-up" regulator (a) with the engine cold (b) with the engine warm. (RobertBosch Corporation)

in the control pressure will increase the forceacting on the top of the plunger, whichopposes movement of the air metering vane.The vane will not move as far as it otherwise

might, the control plunger will not rise ashigh, and so less of the metering slits will beuncovered. Conversely, a reduction in con-trol pressure will allow the vane and plungerto rise higher, uncovering more of the slits.Thus, a higher control pressure will lean themixture; a lower one will richen it.

94

Regulator-The control pressure, in turn,is regulated by the control pressure regulator(a reasonable name!). The regulator main-tains a constant pressure downstream with aspring loaded valve that opens at the setpressure and allows fuel to recirculate backto the tank. It should be clear from the above

description that, in a given installation, theair/fuel ratio depends entirely on the controlpressure. Use is made of this fact for warm-up enrichment,indeedthe originalnamefor

the control pressure regulator was wann upregulator (WUR).

The control pressure/wann up regulator issupplied with fuel from the fuel distributor;a second fuel line leads from the regulatorback to the tank. Within the regulator, a flex-ible metal diaphragm is held very close tothe port leading to the return line by a pair ofsprings, one a coil, the other a leaf spring. Ifthe control pressure tends to drop for anyreason, the force of the springs forces thediaphragm even closer to the return port,obstructing the flow and so raising the pres-sure again. Similarly, if the control pressureattempts to rise, the diaphragm will bulgeaway from the return port allowing moreflow and thus dropping the pressure.

To achieve the enrichment needed for

wann up, the leaf spring is made as abimetallic strip. When cold, it bends down-ward, which opposes the coil spring and soallows the diaphragm to move further fromthe return port. That allows more fuel toflow back to the tank, thus dropping the con-trol pressure. At Montana-in-February tem-peratures, a typical control pressure mightbe as low as 7-8 psi; on a fully wann engine,the control pressure will be about 50-55psi.The WUR is invariably mounted on theengine block, or somewhere else that runs atengine temperature, so the effect is sup-pressed when the engine is wann.

With an engine started from stone cold,the bimetallic leaf would gradually straight-en out again as the engine wanns up, whichwould raise the control pressure and so leanthe mixture back to the standard operatingair/fuel ratio. Trouble is, this would take too

long-maybe ten minutes or more-and theengine would run excessively rich duringthe later stages of wann up. To avoid this,the bimetallic leaf has a heating coil woundaround it, very much like the thermo-timeswitch that runs the cold start injector on theL-Jetronic, as described in Chapter 3. Thisheater is powered every time the ignitionswitch is turned on; after just a couple of

minutes, even in freezing temperatures, itwill have bent so as to return the leaf to its

basic position, and so restore the controlpressure to the basic setting.

On some versions of K-Jetronic, the con-

trol pressure regulator is given the furtherfunction of modulating control pressureaccording to engine load, increasing mixturestrength under conditions of lowvacuumlhigh pressure in the intake mani-fold. This additional function justified thename of the device being changed from"warm up regulator" to "control pressureregulator." This modified form of controlpressure regulator is most often found onturbocharged engines, but some normallyaspirated models use it, too.

This later, wider purpose control pressureregulator can be recognized from its two-piece construction, contrasted with the one-piece WUR. The extra section contains asecond diaphragm within a separate sealedchamber. One side of the diaphragm is vent-ed to the atmosphere, while the other sidefaces the "original" part of the control pres-sure regulator, which is connected by a vac-uum hose to the manifold. That second

diaphragm serves to support a second coilspring nested within the main coil, like"dual" valve springs. With high manifoldvacuum, corresponding to ordinary lightloads, the manifold vacuum pulls thediaphragm to the "basic" position.

When manifold vacuum drops to nearatmospheric pressure, as during full throttlerunning (or when turbo boost drives mani-fold pressure above atmospheric), thediaphragm is pushed down to the high-loadposition. Because the supplementary coilspring is supported by the diaphragm, it nolonger contributes to shoving on the other,"original" diaphragm that restricts returnflow to the tank, so the return flow increas-

es. Accordingly, the control pressure drops,the control plunger in the fuel distributor isallowed to rise further from the force

applied by the metering vane, more slit area95

1

5

1. Electricalconnection

2. Fuelsupplywithstrainer

3. Valve(electromagneticarma-

ture)

4. Solenoidwinding5. Swirlnozzle

6.Valveseat 4

36

A cold start injector (cold. ...start valve), shown In the IS uncovered, and a Dcher Ill1Xture IS thus"on" state. (Robert Bosch provided for high load conditions.Corporation)

Other Cold Running CorrectionsThe degree of enrichment achievable by

modification of the control pressure is limitedby the maximum flow capacity of the fuelslits in the barrel of the fuel distributor.Since

it would be uneconomical to design a distrib-utor having enough range to accuratelymeterfuel for both the tiny appetiteof a warm idlingengine and the voracious one of an engineattempting a cold start in a Canadian winter,this latter, worst case situation is catered forby a cold start injector, or IIcoldstart valve,"as Bosch terms it, the same arrangement asthat used on the intermittent systems dis-cussed in previous chapters.

Cold Start Injector-The cold start injec-tor is a solenoid valve, similar in design andconstruction to the electromagnetic injectorson the intermittent systems, that injects fuel

96

into the intake manifold. As with the cold

start injectors used on intermittent systems,power is fed to it whenever the starter isenergized, but the circuit is only completedwhen a thermo-time switch is closed. This

switch contains the by-now-familiar electri-cally heated bimetal spring/switch arrange-ment. When cold (below about 95° F), theswitch is closed, grounding the injector andthus energizing it, allowing fuel to flowwhen the starter is energized. When hot, thebimetallic leaf bends, breaking the electricalcontact and thus shutting off the injector. Atsub-zero temperatures, the cold start injec-tor will operate for five to twelve seconds,until the electric heater warms the leaf. To

prevent the switch from actuating the injec-tor on an already warm engine, the thermo-time switch is screwed into the engine waterjacket (or a cylinder head on air cooledengines), where engine heat will keep theswitch open.

Auxiliary Air Bypass-Another cold startfeature reminiscent of the intermittent sys-tems is an auxiliary air bypass. Becausethere is no fast-idle cam, the internal frictionof a cold engine could reduce the idle speedso much that the engine might stall. Theauxiliary air bypass (on the K-Jetronic,Bosch calls it an auxiliary air device) pro-vides a reasonable idle speed by allowingsome air to bleed around the nearly closedthrottle.

This bypass comprises a housing contain-ing a small perforated disc valve that con-trols airflow through an air bypass channel,according to the position of the disc, whichis governed by a another electrically heatedbimetallic striplleaf spring. When the leaf iswarm, the valve is completely closed; whenvery cold, it is fully open; at intermediatetemperatures, it adopts a position some-where in between, providing a smooth andgradual transition as the engine warms up.To prevent it admitting an excess of airwhen the engine is warm, the auxiliary airdevice is mounted somewhere on the engine

2 3 4 5

1. Suctionside

2. Pressurelimiter

3. Rollercellpump4. Motorarmature

5. Checkvalve

6. Pressureside

The roller cell fuel pump. The rollers fit loosely in their lands; centrifugal force pushes them against the contoured interior of the pumpbody. (Robert Bosch Corporation)

where it will be warmed by engine heat, thusit is usually mounted remote from the throt-tle body, with connection by hoses. In thisregard it is just like the thermo-time switchfor the cold start injector. However, unlikethe functioning of the thermo-timeswitch/cold start injector, which is never inoperation for more than a few seconds, invery cold weather the auxiliary air devicecan take several minutes to move from fullyopen to fully closed.

Idle Air Bypass-Do not confuse this aux-iliary air device with the idle air bypass,which provides another parallel routearound the throttle plate(s). To allow adjust-ment of idle speed, a small amount of air isallowed to bypass around the throttlethrough a separate small passageway. Theamount of air that is bypassed in this waycan be adjusted by a small needle valve, andthis is the only means that should be used toadjust the idle, never the throttle stopsetscrew.

Pumps and PlumbingThe K-Jetronic uses the same basic design

of roller cell electric fuel pump as the inter-mittent systems described in previous chap-ters, but regulated at a much higher pres-sure-typically 75 psi, vs the 35 psi or so ofthe intermittent systems. Because thesepumps push much better than they pull, thepump is mounted low down, often withinthe tank itself. If the pump is not physicallywithin the tank, a second lift pump-some-times called the "prepump," or "transferpump" -may be used to supply the mainpump from the tank.

Shut-Off Valve-Usually, the pump isenergized, via a relay, whenever the ignitionis on. However, if a fuel line ruptures, per-haps because the vehicle has becomeinvolved in an accident, there is clearly asevere fire hazard; the fuel pump might con-tinue pumping fuel into a blaze. To reducethe risk of this, a safety shut-off is incorpo-rated. In early versions ofK-Jetronic-up toabout 1977- this takes the form of a simplecontact switch under the air metering vane.If the fuel supply to the engine is cut offbecause of a severed line, the engine will

stop, the airflow will cease, and the vane97

:t b

a. In restpasition

b. In actuatedposition

1. Systempressureentry2.Seal

3.Returntofueltank

4.Plunger5.Spring

2 3 54

Primary (system) pressure regulator, fitted to the fuel distributor. (Robert Bosch Corporation)

will drop to its rest position. This opens thesafety switch, which opens the circuit to therelay, interrupting current to the pump. . .unless of course the vehicle has come to a

stop inverted!To avoid the hazard potential in this

arrangement, later versions of K-Jetronic,and all K-Iambda and KE-systems, use theoperation of the ignition system as the indi-cation that the engine is running. Pulsesfrom the primary circuit of the ignition coilare fed to the relay, where a simple internalcircuit evaluates them. If the ignition ceasesto provide regular electrical pulses, thepump relay shuts off, stopping the pump.Cranking speed is sufficiently high to satis-fy the relay-sometimes called the "safety"relay, sometimes the "control" relay-thatthe engine is running.

Fuel Accumulator-Downstream of the

system pump is a fuel accumulator. This issimply a chamber containing a diaphragmand spring. Assuming the accumulator to beempty, on pump startup the fuel pressurepushes the diaphragm against the springresistance as fuel begins to fill the accumu-lator, until the force of the spring balancesthe pressure of the fuel. This takes about one

98

second, during which time the pressure atthe fuel distributor rises gradually, thus pro-tecting it against the sudden pressure surge itwould otherwise experience. The "shockabsorber" characteristic of the accumulator

also helps to damp fluctuations in fuel pres-sure that might result from, say, a suddenlarge electrical load causing the pump tobriefly slow.

When the pump stops on engine shutdown, the fuel pressure drops rapidly at first,but the spring in the accumulator continuesto push on the diaphragm, holding a residualpressure in the system of about 30psi. Thishelps ensure quick restarts on a hot engine,and also helps reduce the chance of vaporlock as the now noncirculating fuel contin-ues to pick up heat from a hot engine. Areplaceable fuel filter lies downstream of theaccumulator.

Pressure Regulator-Because K-seriessystems depend on pressure balances withinthe system for accurate fuel metering, exactcontrol over system pressure is essential.Fuel pressure within the primary system (asopposed to the control pressure) is governedby a pressure regulator, built in to the fueldistributor. There are two versions of this

device. The first is a simple plunger andspring arrangement, in which the fuel pres-sure pushes on the plunger, against the resis-tance of the spring, until the plunger uncov-ers a relief or "spill" port in the side of theregulator housing, allowing surplus fuel tobe recirculated back to the tank. A later

design, used after about 1978, incorporates asecond check valve that seals off the returnline from the fuel distributor. This second

"push valve" is mechanically connected tothe relief valve that governs primary systempressure. Thus, when the pump stops and thespring in the regulator shoves the plungeragainst its seat (which is the "dead end"against which the accumulator continues topress), the "push valve" is also pulled shut,helping to maintain residual pressure withinthe control circuit, with benefit to hot re-starts, as noted above for the accumulator.

In the operation of the K-Jetronic, a verylarge fraction of the fuel supplied by the sys-tem pump winds up getting returned to thetank. This fuel passes near the hot bitsaround the engine, and picks up more heatby being pumped through numerous ori-fices. To ensure thorough intermixing of thehot returned fuel and the cool fuel in the

tank, the fuel tanks on K-Jetronic systemsare usually quite tall in relation to theirwidth.

An immediately visible feature of anengine equipped with K-Jetronic is the indi-vidual fuel lines leading from the fuel dis-tributor to each injector. Unlike the intermit-tent systems, there is no "ring main" thatconnects all the injectors.

Fuel InjectorsThe injectors themselves are much sim-

pler devices than those used on the intermit-tent systems. They play no part in meteringfuel; their only task is to atomize it. An inter-nal check valve opens when the pressure ofthe supplied fuel exceeds the resistance of asmall spring within the injector, and as longas the system pressure acts, the injector

sprays. While this opening pressure obvi-ously has to be less than full system pres-sure, note that it is more than the residualpressure in the system when the engine isstopped, to prevent the injectors from drib-bling fuel into a stopped engine.

Atomization is aided by the design of thecheck valve, which incorporates a small pinthat oscillates or "chatters" in the flow of

fuel around it, helping to break the liquidfuel into tiny droplets. (The vibrations ofthis pin are quite audible when an injector isworking). When fuel pressure ceases, thespring reseats the pin, blocking fuel flowand maintaining pressure in the lines, help-ing to ensure a rapid restart. There are noelectrical connections, and nothing much togo wrong, unless an injector becomesplugged.

K-Lambda SystemProgressively stricter emissions limits

imposed throughout the latter part of the1970staxedthe mixturecontrolcapabilitiesof the basic K-Jetronic system describedabove. In response,Bosch adapted lambdacontrol to their K-system. Starting in the1980model year, almost all manufacturersusing the continuoussystemchangedto thisimprovedversion.(For more detailson thelambda sensor and its operation, see thesidebar "The Oxygen Battery,"in Chapter2).

Basic System Operation & ComponentsIn the K-Jetronic with lambda- K-Iambda

for short-the lambda sensor in the exhaust

pipe sends information on the oxygen con-tent in the exhaust gasses to an electroniccontrol unit. The control unit, in turn, sends sig-nals to a lambda valve which modifies the mix-

ture strength. If the sensor reports a high oxygenlevel, then the mixture is too lean, so the elec-tronic control commands the lambda valve to

richen the mixture; if the sensor reports low lev-

els of oxygen, the engine is running rich, so thecontrol tells the lambda valve to lean the mix-

99

1

!

1.Fueltank

2.Electricfuelpump3.Fuelaccumulator4.Fuelfilter

5. Warm-upregulator

6. Injectionvalve7. Intakemanifold

Schematic diagram of the K-Jetronic system with closed-loop lambda (oxygen sen-sor) control. (Robert BoschCorporation)

100

:QjJ.

8. Coldstart valve

9. Fueldistributor

10.Airflowsensor

11.Timingvalve12. lambdasensor

13.Thermo-timeswitch

14.1gnitiondistributor

ture out a bit. This closed-loop feedbackcontrol means that the control unit is contin-

uously "chasing the ball," richening inresponse to a "lean" signal, then leaning inresponse to the "rich" signal. As a result, theair/fuel ratio is constantly oscillating aroundthe stoichiometric value that is necessary ifthe catalytic converter is to do its job,though always within some fraction of onepercent.

Tounderstand how the lambda valve mod-

t

! lit !

+0

BOSCH19

15.Auxiliaryair device16.Throttlevalveswitch

17. Controlunit

18. Ignitionandstartingswitch

19.Battery

ifies the air/fuel ratio, we first have to back

up a bit. Recall that we explained that thedifferential pressure valves in the plain-vanilla K-Jetronic are there to maintain a

constant pressure drop across the meteringslits in the fuel distributor barrel, so the rateof fuel flow through the slits is purely afunction of the area of slit exposed. Withoutthis stabilizing influence, increasing theexposed area of the slits would increase fuelflow, but because of the added flow, the

4

pressure drop across the slits woulddecrease, so the flow increase would begreater than might be expected on the basisof the slit area increase alone.

The lambda valve functions by, in effect, fight-

ing the stabilizinginfluenceof the differentialpressurevalves,by adjustingthe pressurein thelowerchambersof the valves.Recallthat in thebasic versionofK-Jetronic,the lowerchambersof the differentialpressure valves are alwaysoperating at primary system pressure, whilethe upper chambers are operating at systempressure, minus the pressure attributable tothe force of the spring. If the pressure in thelower chambers were to drop for any reason,the pressure in the upper chambers will like-wise drop-it will always be equal to lowerchamber pressure minus spring force.

A drop in pressure within the upper cham-

3

10 6 7 8

~

9

t

bers will increase the pressure differencebetween the primary pressure on the supplyside of the slit and the delivery pressureinside the upper chamber, and so increasethe fuel flow for a given amount of exposedslit area. Thus, a drop in pressure in thelower chambers richens the mixture for a

given metering vane/control plunger posi-tion, and an increase in pressure there willlean-out the mixture.

Controlling Pressure-K-Jetronic withlambda allows the lambda valve to control

the pressure in the lower chambers by intro-ducing a new fuel circuit that supplies fuel tothe lower chambers through a restriction,then bleeds fuel off for return to the tank viathe lambda valve. The more the lambda

valve restricts that return flow, the higher thepressure builds in the lower chambers, so

1. lambdasensor

2. lambdaclosedloopcontroller

3.Frequencyvalve(variablerestrietor)4.Fueldistributor

5. lowerchambersofthedifferentialpres-surevalves

6.Meteringslits

7.Decouplingrestrietor(fixed restrictor)8.Fuelinlet

9.Fuelreturnline

Additional components forclosed-loop lambda (oxygensensor control of K-Jetronicsystem. (Robert BoschCorporation)

101

the higher the pressure becomes in the upperchambers, the less pressure drop is experi-enced at the slits, so for a given plungerposition, the mixture is leaned out.Conversely, if fuel flows more readilythrough the lambda valve, the pressure inthe lower chambers will drop, so the pres-sure in the upper chambers will drop, thepressure drop across the slit will increase,more fuel will flow, and the mixture is rich-ened.

The lambda valve "turns the tap" not, asyou might suppose, by varying the size ofsome orifice, but rather by alternately-andvery rapidly-completely opening and com-pletely shutting a passage of fixed size. Ifthis begins to sound familiar, then you haveremembered the operating principle of theelectromagnetic injectors of the L-series ofJetronics, described in earlier chapters; thelambda valve closely resembles one of thoseinjectors, both in function and appearance.

Duty Cycle-The signals sent from theelectronic control to the lambda valve

amount to pulses of current that hold thevalve open as long as the pulse lasts; when apulse ends, the valve closes. The averagerate of flow through the valve over any rea-sonable time period, then, is a function ofhow long the valve stays open comparedwith how long it is shut. This relationshipbetween on-time and off-time-called the

duty cycle-establishes the average rate offlow from the lower chambers to the tank,via the valve, and thus controls the mixture

strength in the way described above.Assuming a reference duty cycle of 50%

(the valve is open half the time and closedthe other half), then if the duty cycle is shift-ed to 45% (valve open 45% of the time,closed 55% of the time), the rate of flowthrough the valve will diminish, the pressurein the lower chambers will increase. . . andso on, and the mixture will become leaner. If

the engine is basically healthy, in practicethe electronic controller keeps the duty cyclevarying in the range of 45%-55%, in

102

response to information from the lambdasensor.

As mentioned in Chapter 2, however, alambda sensor needs to be heated to at least

3000 C (about 5700 P) before it functions.Thus, at cold engine startup, the sensor isinoperative, and the system operates inopen-loop mode. To cope with the uncer-tainties in mixture strength during cold startsand warm-up, some K-Iambda installationsare equipped with a thermo-time switch.Information from the thermo-time switch

that the engine is cold causes the controllerto supply a fixed duty cycle of about 60% aslong as the thermo-time switch is closed.The thermo-time switch self-heats suffi-

ciently to switch off before the lambda sen-sor warms up enough to become opera-tional, however, so at this point the con-troller reduces the duty cycle of the lambdavalve to somewhere between the cold start

value of 60% and the closed loop 45-55%value.

KE- JetronicWhile the addition of lambda control

enabled K-Jetronic systems to meet emis-sions limits up to the early 1980s, it was notconsistently able to meet the stricter limitsthen being introduced. To further tightencontrol over mixture strength, the samebasic system was refined by the addition ofnumerous other sensor inputs, and a muchexpanded role for the electronic control unit,which was now equipped with the same sortof electronic memory storing engineload/speed/fuel flow "maps" as used onintermittent systems. This "KE-Jetronic"first appeared on the 1984 Mercedes 190.

The array of new sensors resembles thaton intermittent systems. New sensorsinclude: an engine rpm signal, fed from theignition system; a full throttle and a closedthrottle switch; an engine temperature sen-sor; a potentiometer at the air metering vane,to monitor its position; and, on some vehi-cles, an air temperature sensor. These, plus

Components of the KE-Jetronic. (Robert Bosch Corporation)

the existinglambdasensor,all feedtheir sig-nals to the ECU, which modulates mixture

strengthin a way ratherdifferent from boththeK- andK-Iambdasystems.

The basic, all-mechanical, K-Jetronicadjusts mixture strength by varying the con-trol pressure acting on the top of the controlplunger, while keeping the pressure dropacross the barrel slits constant. The K-Iamb-

da system continues to use the control pres-sure as the main variable, particularly whenthe engine is cold, but then superimposes afurther control-driven by the lambdasen-sor-by deliberately varying the pressuredifference across the slits. In contrast, the

KE-Jetronic achieves control over air/fuel

mixturesentirelyby changesin the pressuredrop across the barrel slits; the control forceis constant- indeed, it is full primary sys-tem pressure, maintained that way by theprimary pressure regulator. The device thatexecutes the changes in air/fuel ratio, oninstructions from the ECU, is the pressureactuator, a small, light colored box on theside of the fuel distributor.

Basic System Operation & ComponentsAt root, the principle is the same as that in

K-Iambda systems: for a given position ofthe control plunger, the rate of fuel flow to

103

from control-pressureregulator

K (control pressure)from fuel pump

KE (system pressure)

The basic K-Jetronic adjusts mixture strength by varying the control pressure acting on the top of the control plunger, while keeping thepressure drop across the barrel slits constant. In contrast, the KE-Jetronic achieves control over air/fuel mixtures entirely by changes inthe pressure drop across the barrel slits; the control force is full primary system pressure, maintained by the primary pressure regulator.

the injectors depends on the pressure dropacross the barrel slits, which depends on thedifference between the primary system pres-sure and the pressure in the upper chambers,which last in turn depends on the pressure inthe lower chambers. The trick, then, lies inmodulating the pressure in the lower cham-bers. While K-Iambda systems achieve thisby starting with the fixed primary systempressure going into the lower chambers andvarying the resistance to fuel flowing out, theKE systems invert this; they impose a fixedresistance (through a narrow orifice) on theway out and vary the pressure going in.

This is achieved by a thin metal plate thatobscures the inlet to the pressure actuator to

104

an extent that depends on the position of theplate. When close to the supply port, theplate restricts flow considerably, so the rateof fuel supply to the lower chambers is pro-portionally reduced. Because fuel is alwaysflowing out from the lower chambersthrough the fixed restriction of the narroworifice, hindering the in-flow will drop thepressure in the lower chambers. Because theupper chambers are always at lower cham-ber pressure minus the spring force, thepressure in the upper chambers will drop,too, so the pressure drop across the slits willincrease, so more fuel will flow for a givenplunger position, so the air/fuel mixture willbe richened.

Injector fuel Diaphragm

Electricalconnector

Actuatorfuel

Actuator fuel Pressureactuator

KE systems impose a fixed resistance to flow out of the lower chambers of the differential pressure valves and vary the pressure goingin by having a powerful electromagnet bend a thin steel plate.toward or away from the supply port.

(In view of the fact that the chambersanddiaphragms in the fuel distributor are termed"differential pressure valves," it is a matterof some potential confusion that Bosch alsorefers to the flow controlling capability ofthe pressure actuator's control plate in termsof "differential pressure." Technically, ofcourse, this is perfectly correct-theobstruction to flow creates a pressure differ-ence between the inlet to the pressure actua-tor and the outlet to the lower chambers,

thus a "differential pressure" exists betweenthese points-but the possibility of confu-sion remains.)

The movement of the metal plate in thepressure actuator is accomplished by anelectromagnet that pulls the plate down toobstruct the inlet according to the currentsupplied to the magnet. That current, andthus the mixture strength, is governed by the

electronic control unit. With zero current

flow- which only occurs with the enginestopped or if there is a component failure-the plate is at its most distant from the fuelinlet to the actuator, so the mixture will be

very lean, though still rich enough that awarm engine will continue to run. This is theso-called "limp home" mode.

As long as the system is working correct-ly, the actuator magnet current on a runningengine with early versions of KE-Jetronicwill be some small positive value, say 1 mil-liamp (ma). On a warm idling engine, thecurrent may be somewhere in the order oflOma; for cold starting and full load, the cur-rent may be as much as 70ma. Later ver-sions have different actuator current ranges.On the KE3-Jetronic, the current typicallyranges from +lOma (full rich) to -lOma (fulllean), while the corresponding range for

105

Air-shrouded injectors (right) allow a little of the engine's intake air to pass through the injectors themselves, mixing with the fuel as itdoes so. Very much like the emulsion tube(s) in a carburetor, this improves atomization. (Robert Bosch Corporation)

KE-Motronic systems is more like +23ma to-16ma.

The ECD determines the current to be fed

to the pressure actuator on the basis of infor-mation received from the various sensors

mentioned above. After about 1985, thefunctioning of the cold start injector wasalso incorporated into the ECD on someinstallations, so the thermo-time switch waseliminated. Improved control over the warmup phase is also achieved, by virtue of theengine coolant sensor (and, on someMercedes Benz models, intake air tempera-ture). Acceleration enrichment is also dealt

with in a more subtle way. The "overswing"

106

of the air meter still provides the basicmomentary enrichment, but the ECD tailorsthis further, according to the rate of move-ment of the metering vane, as reported bythe potentiometer there. On the KE3-Jetronic, the ECD also has control over theauxiliary air bypass (also called the idlespeed stabilizer valve).

Pressure Regulator-The KE-Jetronicdiffers from earlier K-systems in two otherrespects. First, a different type of primarysystem pressure regulator is fitted. The reg-ulator is no longer integrated into the fueldistributor but is mounted separately.Because there is neither a control pressure

regulator nor a lambda valve, the return linefrom the fuel distributor is connected to the

primary system pressure regulator, fromwhere a single return line to the tank carriesboth the excess from the distributor and that

from the regulator.Air-Shrouded Injectors-The second dif-

ference, common to most but not all KE sys-tems, is the use of air-shrouded injectors. Inthese, a small amount of intake air from apoint upstream of the throttle plates isallowed to "leak" into the manifold throughpassages within the injector itself. This airbecomes entrained with the spray of gaso-line and helps to atomize the fuel. This pro-vision has greatest effect at idle, when man-ifold vacuum is high (and thus the pressuredifference between the supply point and thepoint of delivery is greatest); the small sizeof the "leak" means it has diminishing effectas engine speed and load rise and the totalvolume of air inhaled grows larger, and theleak delivers ever less as vacuum drops.

KE3-Jetronic and KE-MotronicIn addition to the enhanced mixture con-

trol achieved on the KE-Jetronic by numer-ous sensors and the much greater authorityafforded the ECU, KE3-Jetronic and KE-

Motronic systems also integrate control overignition timing. A fundamental differencebetween these two is that the ignition "brain"on KE3 systems is a separate package fromthe fuel "brain"; on the KE-Motronic bothfunctions are integrated into the same ECU.

Both these newer systems also include"diagnostics"-the system constantlychecks itself for faults. A coolant tempera-ture that rapidly fluctuates, for example, issurely a sensor fault rather than a real reflec-tion of what is going on. Likewise, a lamb-da sensor that insists that the mixture is

excessively lean even if the ECU has rich-ened the mixture much further than its pro-gramming says should be necessary proba-bly implies an air leak somewhere.

107

Preventative MaintenanceNote: The informationin the first few

paragraphs,up to the heading"No UserServiceable Parts Inside, " is identical to theadvice given in Chapter 5. However, it isrepeated herefor simplicity.

Before attributing some operating fault tothe fuel injection system, be sure the remain-der of the engine is in sound order. It is quitepointless to start to troubleshoot the injec-tion system if the spark plugs are years old,the rubber intake ducting is cracked, or anexhaust valve is burned.

Spark PlugsAs in the days of carburetors and point-

and-breaker ignition systems, the fIrst diag-nostic test should be to remove the plugs andexamine them-the removed plugs canreveal a great deal about engine condition.The insulator should be a light gray-to-tancolor. An insulator that is bone-white-or

worse, blistered - indicates excessive lean-

ness, or perhaps a plug of the wrong heatrange; a blackened insulator may be theresult of a too-cold plug, or excessively richrunning, or may be a product of oil fouling,because of worn rings or valve sterns, aplugged PCV valve, or even simply an over-filled oil pan. To distinguish between carbon(fuel) fouling and oil fouling, rub the plugagainst the heel of your hand-oil foulingwill leave a greasy smudge; carbon foulingwill not.

Any mechanical damage to the plug - acracked insulator, a broken side electrode-

implies .detonation, which may have dam-aged much more than the plug. A plug that istruly wet with gasolineimpliesa non-fIringcylinderthathas continuedto receivefuel.

Check Gap-Also, check the gap onremovedplugs. If the plugs have seen anysubstantial amount of service, the gap is sureto be larger than on a new, correctly gappedone. This widening of the gap results fromerosion by the hot gasses within the cylinder.An excessive plug gap can make unsustain-able demands on the ignition system - thevoltage required to jump the exaggeratedgap may cause the secondary (high-tension)voltage to rise so high that the spark seeksanother path to ground, perhaps punching amicroscopic hole right through the insula-tion on the plug or coil wires, leaving a leakpath to ground that remains even after theplugs are replaced.

Replacing Plugs-The replacement inter-val for spark plugs suggested by the factoryis likely to be highly optimistic; except forplatinum tipped plugs, they should bereplaced annually, as should the air and fuelfIlters. Check the gap on new plugs beforeinstalling them even if they corne pre-gapped, and take care when replacing thefuel filter that the act of removal and

replacement does not allow dirt to enter thesystem. High-tension wires should bereplaced every couple of years, likewise thedistributor cap and rotor, if applicable.

Routine Checks

Other routine service inspections that

109

should be taken care of before going furtherare checking ignition timing and, on engineswithout hydraulic lifters, the valve clear-ances. Inspect all rubber air trunking forcracks and other sources of air leaks.

Assuming all is in order so far, the next stepis a compression check. This will revealworn rings or leaking valves. Compressionthat is uniformly down by even as much as20- 30psi relative to the factory figure is notmuch to worry about, but variations betweencylinders of that much is cause for concern.

Gas Pains

Many fuel injection troubles can be avoid-ed, or at least long postponed, simply bypaying attention to the quality of fuel used,and where and when it is bought. While theowner's manual may make clear that regulargas of about 87 pump octane is suitable, andwhile the engine may not be able to takeadvantage of the higher octane (about 92) ofpremium fuel, the premium fuel from mostnational gasoline brands contains a moreaggressive detergent additive package thandoes their regular fuel. Clogged injectors areone of the more common causes of griefwith fuel injection systems-all systems,not just Bosch. Higher detergency of the fuelhelps prevent these faults. Even supplement-ing a standard diet of regular with a tankfulof premium every few weeks helps.Aftermarket detergent additives may also beeffective.

Running out of gas is a pain, argumentenough for following the advice implicit inthe old saw that "it costs no more to drive

around with a full tank than a near emptyone," but running most of the time with thegauge showing 1/4 tank or less is especiallypoor practice with fuel injection systems.The larger the air space above the fuel in thetank, the greater the amount of water thatcondenses out of the air, and the greater thesusceptibility of steel and iron componentsin the system to rusting. Apart from thedirect consequences of this corrosion, tiny

110

flakes of rust can play havoc if they get intothe system.

Note, too, that it is a good idea to keep aneye on when your habitual gas station getsits deliveries. The replenishment of the sta-tion's underground storage tanks stirs up rustand sediment in their tanks that may wind upin yours. Better to fill up the day before orthe day after.

"No User Serviceable Parts Inside"

No matter what the results of the diagnos-tic tests described below, bear in mind that

in most cases the only remedy for some-thing out of spec is component replacement.Parts that cannot be repaired or adjustedinclude, where applicable: primary systemfuel pump; lift pump; system pressure regu-lator; accumulator; control pressure regula-tor; lambda valve; pressure actuator; injec-tors; cold start injector; auxiliary air bypass(idle speed stabilizer); and all sensors,except for the air meter.

While the air meter can be dismantled and

certain adjustments or repairs carried out, asdescribed below, the only serviceable com-ponent in the fuel distributor is the controlplunger. At that, all that can be done here isto clean the plunger of varnish and toreplace the a-ring seal at its lowest part.Otherwise, the fuel distributor is essentiallynonserviceable. Bear in mind, too, that with-

out the highly specialized equipment towhich service technicians in dealershipshave access, there is absolutely nothing youcan do with any electronic control modules;you cannot even test them. All you can do ifan electronic controller is suspect is to sys-tematically eliminate all other componentsas potential culprits and, as a final resort,replace the controller.

Troubleshooting ChecksDespite the dismayingly long list of things

that cannot be done, there are neverthelesssome troubleshooting procedures that can behelpful. These should be carried out in a log-

WARNING!Gasolineis highly flammableand potentiallyexplosive.It can be ignitedby an

electricalsparkor by contactwith hot engineparts. Use extremecare when work-ing on any engine'sfuel system.Workonly in a wellventilatedarea,ban smokingor any open flame from the work area, and ensure a fully charged large capacity fireextinguisher is close to hand. Many fuel injection system circuits remain underpressure even when the engine is stopped and the fuel pump is de-activated; beforeloosening any fuel fitting, wrap a rag around it.

ical way, and with an understanding of howthe system is supposed to work and whatdifferent components do in various circum-stances. For example, if an engine startsreadily from cold and runs well while firststarting to warm up, but then runs progres-sively richer as it reaches operating temper-ature, guzzling gas and spewing blacksmoke, one plausible culprit is a cold startinjector stuck open. In the same way, if theidle speed is way too high when warm, ortoo low when cold, one of the first places tolook might be the idle speed stabilizer/auxil-iary air bypass. In general, if only one part ofthe engine's operating regime is affected,look first at components whose function isdirectly related to operation in that regime.

Air MeterAfter the air filter, the air meter is the com-

ponent furthest "upstream." Because it andthe control plunger are the primary means ofcontrolling mixture strength, here is a goodplace to start. It also helps that the meter, asa rule, is fairly accessible! A basic and sim-ple, "almost-no-tools" check is the freemovement and centering of the meteringvane, and its rest position.

Remove the rubber trunking that connectsthe air meter to the throttle body. Next,assuming an updraft sensor, use a smallmagnet to tug at the bolt head in the centerof the metering vane. Once you can get yourfinger under the edge of the vane, lift it andlet it drop. It should rise and fall again freely,and bounce once or twice. If it is sticky

going up, but falls freely, the problem isalmost certainly with the plunger. If you aresure that the plunger has not been damagedthrough careless handling (a dropped one islikely scrap), then the problem is probablygum and varnish from fuel deposits on theplunger.

Clean Plunger-To clean the plungerrequires removing it from the fuel distribu-tor, but unless the engine has been stoppedfor a very long time (hours), there is likely tobe residual pressure in the system, and thispressure has to be relieved. Bearing in mindthat a spray of fuel will likely erupt, wrap arag around the fitting and relieve the resid-ual pressure by cracking open the mostaccessible line on the pressure side of thesystem. Catch as much as possible of theemerging fuel in a container and return it tothe tank.

Remove all the other fuel fittings, againcatching spilled fuel, and the bolts thatsecure the fuel distributor.Although the rea-son you are doing this is because the plungerseems to be sticking, Murphy is everywhereand the plunger may just fall out of its ownaccord, so get a finger under the plungerbefore lifting the fuel distributor. A droppedplunger is quite likely a junk plunger andyou can't buy them separately, you have toget a whole new distributor. Keep an eyeout, too, for a spring above the plunger, usedon versions of K-Jetronic after about 1983,

and note that on KE systems, the plungercannot fall out-it is retained by a stopscrew. Before removing this screw to free

111

the plunger, be sure to measure the depth ofthe screw below the rim of its surroundingnut. After reinstalling the plunger, be surethe stop screw is turned in to give this samedepth.

Thoroughly clean the plunger and inspectit for signs of rusting. If all seems well,replace it, reattach the fuel distributor, usingnew gaskets, and retest for free movement.If the plunger still sticks after cleaning, youwill have to replace the fuel distributor.

Sensor Vane-If the sensor vane rises

freely but sticks near the bottom of its trav-el, it may be rubbing against the side of theair funnel at its narrowest part. This calls forrecentering the vane. The bolt that securesthe vane to the lever is smaller than the hole

it passes through, so the vane can be recen-tered by loosening this bolt. It is usually pos-sible to judge a correctly centered vane by"eyeball," or you can run a 0.004" feelergauge all around the edge of the vane. Re-check that you haven't disturbed the center-ing while retightening the bolt.

Establishing the rest position of the meter-ing vane requires that there be residual pres-sure on the system. To ensure there is, re-start the engine, let it run for a few minutes,and switch it off again.

On K-Jetronic and K-Iambda systems, theedge of the vane at rest should be at or justbelow (not more than about 0.020" below)the top edge of this narrowest part of thefunnel. On updraft sensors, this measure-ment is made at the edge of the plate nearestthe fuel distributor; on downdraft versions,measure at the edge furthest from the dis-tributor. Adjustment of this rest position ismade by bending the wire clip on the leafspring (NOT the leaf itself) that forms thelower travel stop for the vane.

On KE-systems, there are two criticalpositions for the vane when the engine isstopped. The first, called the "rest" or "zero"position, is where the vane sits naturally; thesecond, called the "basic" position, liesslightly above the rest position on updraft

112

meters, slightly below it on downdraft ones.When the vane on updraft meters is liftedmanually from the rest position, there willbe a small amount of free movement (about0.04-0.08") before a definite contact is felt,as the vane's lever takes up the small amountof clearance between itself and the bottom

of the control plunger; this is the basic posi-tion. (This clearance assures that the plungera-ring seals firmly against its seat.)

In the rest position, most of the verticalside of the narrowest part of the funnelshould be visible on updraft meters. Consultyour own vehicle's manual to determine theexact distance of the vane below the topedge of this cylindrical area-it varies fromone vehicle to another, although 0.07-0.08"is typical. The measurement on updraftmeters is made at the edge of the vane clos-est to the fuel distributor. If this distance is

not within spec, bend the wire clip on theleaf spring, as described above for K- and K-lambda systems.

As noted, on updraft sensors the basicposition lies 0.04-0.08" above the rest posi-tion, bringing the sensor vane to the top ofthe narrowest portion of the funnel. If therest position is correct but the basic positionis out of spec, remove and adjust the plungerstop screw.

On downdraft meters, these relationshipsare inverted. At the rest position, the vaneshould be level with the top of the cylindri-cal portion of the funnel; the basic positionis 0.04-0.08" below that. In this case, thereference point is at the edge of the vane fur-thest from the fuel distributor, and the mea-surement is made there.

The rest position of the vane on downdraftsystems is set by a stop pin that protrudesthrough the air meter housing. This pin is apress fit and can be (carefully!) tapped downor prised up, as required. Again, if the restposition is correct but the basic position isout of the specification for your application,the clearance can be adjusted at the plungerstop screw.

Valve closed

Systempressure

t

Control

pressureregulator

Fuel Flow and Pressure Checks

With the air vane centered and correctlypositioned, attention can be turned to thefuel system. The checks here involve testingboth flow rates and pressures. For volume(flow) tests, a graduated container of aboutone liter capacity will be needed. For pres-sure tests, a pressure gauge with a capacityof at least 100psi is needed, together withhoses and fittings, to connect it to the vari-ous circuits to be tested (threads on K-Jetronic fuel fittings are 12mm11.5), plus ashut-off valve and a Tee-fitting. Assemblethe gauge, fittings, and shut-off tap so thatthe single line leading from the gauge termi-nates in a Tee; fit the shut-off valve in one ofthe two lines leading from the Tee.

On K- and K-Iambda systems, relieve thesystem pressure, then disconnect, at bothends, the return line from the system pres-sure regulator on the fuel distributor to thecontrol pressure regulator. Attach the gaugeline that contains the shut-off valve to the

System pressureregulator

control pressure regulator; attach the otherline to the port at the fuel distributor. Thissetup allows testing the full primary systempressure, the control pressure, and the resid-ual pressure maintained in the system whenthe pump is stopped.

Remove the fuel tank cap. Unplug theelectrical connections at the control pressureregulator and the auxiliary air device, thenbypass the fuel pump relay, to allow thepump to be energized with the enginestopped. While a great many differentmeans to do this are spelled out in the man-uals for various makes and models of vehi-

cle, simply feeding the pump directly fromthe battery with a jumper wire works inevery case, but take care that the polarity iscorrect.

Primary System Pressure-With theshut-off valve closed, no fuel can return tothe tank, so the gauge will read primary sys-tem pressure-the fuel pump is workingagainst the "dead-end" of the closed valve.

A single test setup permitsmeasuring main systempressure (with valveclosed), control pressure(with valve open, see nextpage), and residual pres-sure, by using a "Tee" fittingand a shut-off valve (seetext). Be sure to use agauge with a high enoughpressure capacity-at least90 psi.

113

Specifications vary from one model toanother-check your vehicle's manual-buta typical primary system pressure is about75psi. If the primary system pressure isslightly out of spec, it can be adjusted at thesystem pressure regulator by adding orremoving shims under the spring that closesthe regulator plunger/piston. Each shimadded or removed increases or reduces pres-sure by 2 to 2 1/2 psi, respectively.

Primary System Flow-If the systempressure is very much below the specifiedvalue, the problem may lie with the fuelpump, or may be the result of a clogged fil-ter or a clogged or pinched line. After chang-ing the filter as a matter of course, thepump(s) can be checked by measuring thefuel delivery rate. Carefully relieve systempressure, disconnect the return line at thepressure regulator, attach a length ofreplacement line, and run the pump, catch-ing the discharged fuel in a graduated vessel.Again, specific values vary from one vehicleto another, but the general range is from750cc to about 11OOccper thirty seconds.(The specific figure for any given modelappears to be pretty much in proportion tothe power output per cylinder, which makessense when you think about it.)

If there is zero delivery, first look to theobvious: confirm that there is fuel in the

tank! If in doubt, add a gallon of fuel. Next,confirm that the pump is running. In a rea-sonably quiet environment, you should beable to hear it. If the pump is not running,check the electrical supply to the pump,starting with its fuse. If the fuse is OK,check for electrical power at the pump. Avoltmeter with one probe applied to one ofthe fuel pump ternlinals and the other to agood chassis ground should show batteryvoltage (near 12 volts) at one pump termi-nal, and zero at the other. (You will have tofold the rubber boot around the connector

out of the way to gain access to the electri-cal terminals.) Low voltage at the pump islikely the result of corroded terminals. If

114

there is adequate power to the pump but itdoes not run, the pump is defective and willhave to be replaced.

If the pump is receiving about twelve voltsand runs, but its delivery measured at thepressure regulator is much below spec,inspect connecting lines for damage and/orobstructions. Disconnect both ends of the

fuel supply line from the pump to the pres-sure regulator and blow compressed airthrough it. BEWARE THE FIRE HAZARDOF A SPRAY OF FUEL FROM THEOPEN END OF THE LINE. If the line is

clear but the delivery remains far belowspec, remove and clean the coarse strainerinside the fuel tank. If all this fails, the pumpis defective.

Control Pressure-Assuming the primarysystem pressure is within specification, con-trol pressure should be checked. With thepump running and the gauge hooked up asdescribed above, open the shut-off valve inthe gauge setup. Fuel is now free to flowfrom the fuel distributor to the control pres-sure regulator, and from there back to thetank. In this configuration, the gauge setupis simply measuring the pressure in this, thecontrol circuit.

As explained in the previous chapter, thecontrol pressure regulator on K- and K-lambda systems varies the control pressureaccording to temperature. If the engine is atroom temperature or less, the control pres-sure should be low, to provide the enrich-ment needed for cold starting and warm up.Confmn that the electrical supply to theheater in the control pressure regulator isdisconnected. On turning the shut-off valve,the reading on the gauge should sink fromthe system pressure to a lower value, typi-cally 20-25 psi, but within the limitsexpressed in the manual for your car for coldcontrol pressure.

If the cold control pressure is too low, theproblem almost certainly lies in the controlpressure regulator; if too high, the controlpressure regulator may again be the culprit

Controlpressure

t

-f

f

Controlpressure If

regula tor

To check control pressure, the valve must be open.

or, less likely, an obstructed return line fromthe control pressure regulator to the tank. Todetermine which is the case, WITH THEPUMP STOPPED disconnect the return line

from the control pressure regulator to thetank and, with the pump running, allow thefuel to flow into a container. If the cold con-

trol reading is now within spec, the returnline is clogged. If there is no change, thecontrol pressure regulator is defective.

Assuming the cold control pressure to beOK, with the pump running reconnect theelectrical supply to the control pressure reg-ulator. As the electrical heating element inthe regulator begins to warm up the leafspring within the regulator, the gauge read-ing should gradually rise to the warm con-trol pressure value specified for your car,typically 50-55 psi. Recall that the tapering

System pressureregula tor

off of cold start/warm up enrichment as theregulator gradually raises the control pres-sure takes some time-perhaps a couple ofminutes. If the control pressure remains lowand unchanged after several minutes, checkthat the control pressure regulator is receiv-ing something close to battery voltage. Ifthere is no power to the regulator, trace theelectrical connection to it; if there is about12 volts at the electrical connection, yet thecontrol pressure remains low and unchangedafter a few minutes, then the control pres-sure regulator is defective.

Residual Pressure-Recall from the sys-tem description in Chapter 6 that K-Jetronicsystems include an accumulator that main-tains a residual pressure in the system for atleast ten minutes after the pump hasstopped. The purpose of this is to ease hot

115

restarts by ensuring that the lines arecharged; this also helps to reduce vaporlock.

With the gauge still connected asdescribed above for the control pressurechecks-that is, with the shut-off valve

open-shut off the pump. The gauge read-ing should drop from the warm control pres-sure (assuming the control pressure regula-tor was allowed to warm up) to some valuesufficiently high as to achieve the intendedpurpose, but sufficiently low that it allowsthe check valves in the injectors to close,preventing them from dribbling fuel into theengine. Values for residual pressure immedi-ately after pump shut down vary rather morewidely than other system pressure; most,though, are in the range of 20-40 psi.

The "dead end" against which the accu-mulator is pushing is, at one end, the nowfully seated piston in"thepressure regulator;the other end of the line-at the pump-isshut by the check valve in the pump.However, there are several other leakagepaths that can cause the residual pressure todwindle away, either the pump stops imme-diately or more gradually, but still wellbefore the (usually) ten minute minimum. Ifthe residual pressure sinks below the valuespecified in your car's manual in less thanthe designated time, these numerous poten-tialleak points can be systematically trackeddown.

First, after shutting off the pump, close theshut-off valve at the gauge. If the residualpressure with the valve open rapidlydropped below the specified value, but thepressure holds steady for several minuteswith the valve closed, then the problem iswith the control pressure regulator.

If the pressure continues to drop with theshut-off valve closed, pinch closed the linefrom the pump to the accumulator. If thepressure now holds steady, the pump checkvalve is defective and needs to be replaced.This is surely the most common cause ofloss of residual pressure.

116

If the pressure still drops with the pump-to-accumulator line squeezed closed, thenlikewise pinch shut the line to the cold startinjector. If that ends the leakage, the coldstart valve is defective; replace it.

If the pressure continues to drop with boththese lines clamped shut, close off the returnline from the fuel distributor. If that ends the

leakage, the source of the leakage is eitherthe system pressure regulator or the fuel dis-tributor itself. In either case, the fuel distrib-utor has to be replaced.

If all this fails to stem the drop in residualpressure, about the only remaining candi-dates are one or more injectors with defec-tive check valves, or external leaks from one

or more fuel fittings. With pressure on thesystem, carefully inspect all fittings forseepage.

Injector Flow, Pattern, LeakageThe continuous flow injectors used on

K-systems are less prone to trouble thanthe intermittent injectors used in L-systems.Nevertheless, they do sometimes becomepartially or even completely clogged, espe-cially if the fuel regularly used lacks ade-quate detergency. Partial clogging mayresult in a spray pattern that is lopsided, or adischarge that is hardly atomized at all-more like a stream than a spray-or maysimply reduce the rate of delivery. Leakageis less common.

Taking care not to kink the lines, removethe injectors and temporarily plug the holesthey came out of. Set each injector into themouth of a graduated vessel. A reasonabledegree of accuracy is needed here, so gradu-ated cylinders (available at any laboratorysupply outfit, or ask your druggist) arepreferable to domestic measuring cups, etc.It should be pretty obvious that plastic ves-sels are preferable to glass ones!

For safety's sake, disable the primary igni-tion circuit. Now run the pump; no fuelshould flow from the injectors as long as theair metering vane is in the rest position.

Manually lift the air metering vane, allow-ing the injectors to discharge into the gradu-ated cylinders, and observe the spray patternfrom each. It should be cone shaped andsymmetrical, and the amount flowed byeach injector should be the same within tenpercent or less. A slight degree of asymme-try in the spray pattern is acceptable, as longas the delivery volumes match, but pro-nounced lopsidedness or a stream ratherthan a spray requires that the injectors becleaned or replaced.

If the vehicle has seen a lot of low speed,low power operation, partially cloggedinjectors may clean themselves if run atmaximum flow rate for a few seconds. Lift

the air sensor vane all the way up and hold itthere for ten seconds. If that doesn't solve

the problem, a prolonged soak in solventmay work. Failing that, replace the defectiveinjector(s).

Basic Mixture Strength/CO AdjustmentAbout the only truly "tuneable" aspect of

K-Jetronic systems is the provision foradjusting the relative positions of the meter-ing vane and the control plunger. The vane'slever is, in fact, made in two parts, with anadjusting screw that moves the lever con-nected to the vane relative to the lever that

drives the control plunger. This screw isaccessible through a small hole on the top ofthe air meter housing, usually closed offwith an "anti-tampering" plug.

Because this adjustment raises or lowersthe control plunger relative to the air meter-ing vane, it effectively modifies the mixturestrength. Note that while the adjustment isalways carried out at idle, its effect is feltthroughout the engine's operating enve-lope-in that sense, it resembles anadjustable main jet on a carburetor. . . (ah!memories of Stromberg 97s!)

An exhaust gas analyzer (CO meter) isneeded to perform this adjustment, togetherwith an unusually long 3mm Allen wrench.After the anti-tampering plug is pried out,

this wrench fits through the hole andengages the adjusting screw. Before youstart, however, set the idle speed using theidle air bypass screw. Note, too, that the COmeter has to "read" the exhaust gas upstreamof the catalytic converter-remember, theconverter is doing its best to oxidize the COto CO2' There is usually a port or pipe or fit-ting on or near the exhaust manifold for thispurpose.

With the engine fully warm and an appro-priate idle speed set, adjust the mixture untilthe meter reports a CO value correspondingto the figure on the EPA placard in theengine compartment. If the placard is miss-ing, aim for 1.5 percent. If you are going to"rev" the engine, to blowout the cobwebs,REMOVE THE ALLEN WRENCH before

touching the throttle-remember, the meter-ing vane will lift when the airflow increases,so its lever will rise, and the wrench will getbound up on the edge of the access hole, andbend things.

K-Lambda Mixture Strength/COAdjustment

On K-Jetronic systems with a lambda sen-sor, the lambda sensor/electroniccontrolllambda valve will attempt to bringthe mixture to the stoichiometric value no

matter what you do with the COlbasic mix-ture adjustment. The fix is simple- beforecarrying out the adjustment, disconnect thelambda sensor, obliging the electronic con-trol to operate open loop.

Recall from Chapter 6 that the lambdavalve controls fuel flow at the fuel distribu-

tor by cycling between open and closed, andthe electronic controller steers the systemtoward stoichiometry by adjusting the dutycycle (on time vs off time) of the lambdavalve. There is a purpose-made meter thatcan "read" this duty cycle, but it can also beread by an ordinary ignition dwell meter.With the dwell meter set on the four cylinderscale, a 50% duty cycle will read as 45degrees (50% of 90 degrees). This provides

117

DVvELL

(on 4-cylinder scale)

918 27 36 45 54 63 72

810 9

0 100

DUTY CYCLE

An ordinary dwell meter, set on the four-cylinder scale, can read the duty cycle of the lambda valve. If the basic mixture strength is cor-rect, the meter will oscillate around the 45 degree (= 50% duty cycle) point.

DWELL

(on 4-cylinder scale)

918 27 36 45

0

7281

9

0

DUTY CYCLE

A reading around 60-70 degrees (about 55-65% duty cycle) means the lambda valve is trying to drop the control pressure to rich en upan excessively lean mixture.

the opportunity for'a second check on bothmixture strength and lambda system func-tioning; it also permits mixture strength/COto be adjusted on K-Iambda systems withoutneed for a CO meter.

With the engine fully warmed up and theidle speed set, connect the dwell meter to thelambda valve. If the meter oscillates around

the 45-degree mark (that is, indicating a dutycycle averaging 50%), then the mixture

118

(on

918

00

0

DWELL

4-cylinder scale)

45 54 63 7281

90

100

DUTY CYCLE

A reading around 25-35 degrees (about 30-40% duty cycle) means the lambda valve is trying to raise the control pressure to lean outan excessively rich mixture.

strength is right on the button and no adjust-ment is needed. A high figure on the meter(say 55 or 60 degrees) indicates the lambdavalve is staying open more, attempting todrop the control pressure and thereby richenthe mixture-the basic mixture is too lean.

Conversely, a low figure on the dwell meter(say 30 or 35 degrees) indicates the systemis trying to correct for an over-rich basicmixture setting.

If a CO meter is available, then both cor-rect mixture strength and correct functioningof the lambda sensor and valve and the elec-

tronic control is absolutely confirmed if theCO reading is the same with the lambda sen-sor connected (closed loop) and disconnect-ed (open loop).

KE-Jetronic and KE-Motronic Mixture

Strength/CO AdjustmentThe remarks above regarding plain-vanil-

la K-Jetronic systems apply also to KE-sys-terns, with a few exceptions. First, it is nec-essary to check the current to the pressureactuator. With the engine fully warmed upand the lambda sensor connected, the needleof a milliameter should swing back and forthover a narrow range, typically 9-llma,though this varies from model to model.This confirms the actuator and ECD are

doing their jobs, constantly correcting themixture back to stoichiometry.

On some early KE-systems, mixturestrength can be adjusted exactly the sameway as on other K-systems, again using a COmeter. If it proves impossible to get both COand actuator current in spec at the same time,look for air or exhaust leaks. On later KE-

Jetronic systems and KE-Motronic systemsthere is no provision for either idle speed orbasic mixture strength adjustment-it is allhandled internally by the ECD.

119

On most KE-Jetronics, andlater systems, a rotary idleactuator (left) replaces thepassive, free-standing auxil-iary air device (right). Thelater device looks like asmall electric motor. (RobertBosch Corporation)

120

Auxiliary Air ValvePlain K-Jetronics and some with lambda

control use an auxiliary air valve to providethe extra air a cold engine needs to maintainan idle speed high enough that the enginewill not stall. As described in Chapter 3, thisis simply a small rotary disc valve thatallows air to bleed around the nearly closedthrottle (it is sometimes called the "auxiliaryair bypass"). The position of the disc is gov-erned by a bimetallic coil. When the coil iswarm, the valve is completely closed; whenvery cold, it is fully open, with a smooth andgradual transition between the two as theengine warms up. In very cold weather theauxiliary air bypass valve can take eight orten minutes to move from fully open to fullyclosed.

The valve is not serviceable, but its condi-tion can be initially diagnosed by symptom:An idle that is appropriate when cold butexcessively high when warm points to thisvalve being stuck partly or fully open.Conversely, if the warm idle seems right butthe engine requires some throttle opening toavoid stalling when cold, then the valve maybe stuck closed.

If the valve is suspect, check, fIrst, bypinching shut the connecting hose. On acold engine, this should drop the idle speed;on a hot one it should make no difference. If

the valve fails this test, sight through thevalve. Because the valve is powered as longas the ignition is on, the heating elementshould completely close the valve within tenminutes. Ensure that 12 volts is getting tothe valve, allow time for it to warm up, thenlook through to confmn the valve is closed.To confmn the valve is opening when cold,pop it into a freezer for ten minutes andagain sight through it-there should be aclear, round passageway through it.

Idle Speed RegulatorMost KE-Jetronics and all subsequent

Bosch continuous injection systems modu-late the auxiliary idle air bypass in a differ-ent way. It remains in principle a valve thatbypasses more or less air around the throttleplates, and its effect is most significant dur-ing warm up. However, because its moresophisticated control enables it to maintainan appropriate idle speed, irrespective oftemperature or engine age and condition, it

(on

DWELL

4-cylinder scale)

is renamed an "idle speed regulator," orsometimes "rotary idle actuator."

In this device, the moveable valve that

varies the size of the air opening is driven bya component that looks like an electricmotor. Indeed, in construction it essentiallyis, but in action it never turns more than 90

degrees, its electrical components receivingpulses from the ECD that cause it to "dither"back and forth slightly around an averageposition that gives the idle speed pro-grammed into the ECD.

Analogous to the function of the lambdavalve, the position of the valve, and thus theamount of air that passes, depends on theratio of on-time to off-time-its "dutycycle," in other words. A long duty cycle-more on-time than off-drives the valve fur-

ther open; a shorter duty cycle closes it fur-ther. This duty cycle can be measured with a

dwell meter, set on the four-cylinder scale.The exact values under various circum-

stances vary considerably among differentvehicles, so check your vehicle's specs in itsmanual. As a general guide, the duty cyclewith a warm, idling engine will be some-where around 30 degrees on the dwell meter.Adding a fair-sized load, such as by turningon an electric rear window defroster, shouldhave little effect on the idle speed, but theduty cycle should increase somewhat.

To simulate cold start operation, discon-nect the engine temperature sensor. Disablethe fuel pump (pull the fuse), and crank thestarter. Dwell should be dramaticallyincreased; expect about twice the idling fig-ure. A defective idle speed regulator cannotbe repaired; it must be replaced.

A dwell meter can also be

used to check out the rotaryidle actuator. The metershould read somewherearound 30 degrees (about35% duty cycle) on a warm,idling engine. Cranking theengine with the temperaturesensor disconnected and thefuel pump disabled simulatesa cold start; the meter shouldshow a large increase in dutycycle-say 60% (correspond-ing to about 55 degrees).

121

27 36 45 6318 72

9 810 \ \ I , I 9

40 50

,I '-.J

bU90

0 100

DUTY CYCLE

-

The issue of performance modifica-tions to Bosch EFI systems is verymuch a good-newslbad-news situa-

tion. On the one hand, Bosch produces asmall number of specialized EFI systems forrace-only, high-performance engines, andthe power outputs achieved can hardly bebettered. Moreover, many of the compo-nents in these systems are essentially thesame stock, bread-and-butter parts found onmillions of passenger cars. It is obvious,then, that in practice as in principle, BoschEFI can extract the last drop of power anddriveability out of a performance engine, forreasons we considered at length in Chapters1 and 2. On the other hand, you can't buyone...unless you happen to be a top-rank rac-ing team with a signed contract with Bosch!

Again, the potential exists to tailor aBosch-based EFI system to match any mod-ified engine. But it cannot be done with ordi-nary tweaking and tuning; trying to applyconventional "hop-up" techniques will gen-erally gain you nothing but grief. In addi-tion, the fuel system is regarded by law-makers as part of a vehicle's emissions con-trol apparatus. ANY modification to the fuelsystem of a street vehicle may render youillegal. Many jurisdictions now require anannual emissions test; if you flunk, youpark.

That said, if you have read the first twochapters, you should understand by now thatthe power an engine can potentially produceis established by how much air it can inhaleand make use of per minute. That, in turn, is

determined by fundamental factors-dis-placement, rpm, camshaft timing, compres-sion ratio. Whether carburetors or a fuel

injection system-of whatever type andbrand - the only requirement of the fueldelivery system is to supply and atomize aquantity of fuel that matches the quantity ofair being breathed at any moment.

System CapacityIf, in the process of measuring the airflow,

the fuel system (carb or FI) restricts that air-flow, then it will prevent realizing the fullpotential of the engine. Again, if the fuelsystem supplies more fuel, or less, thanappropriate for that air quantity, less thanpeak performance will result. The fuel sys-tem, in other words, might reduce power,but it cannot increase it beyond the potentialinherent in the engine's design.

More good-newslbad-news: if you makeminor modifications to improve the air han-dling capacity of a street engine, say by fit-ting a low-restriction air cleaner, or a free-flow exhaust system, or by taking a light cutoff the cylinder head to raise the compres-sion, or by matching ports, then there isalmost certainly enough "head-room"-reserve capacity - in your existing BoschEFI system to match the slightly increasedrate of airflow, with the same precision ithandled the stock motor. You'll have a sweet

running motor with a little more power. Gooff the deep end with a huge overbore, astroker crank, big valves and a cam withlobes the shape of a Hershey bar, and the

123

Bosch makes purpose-built racing fuel injection systems for many professional racing teams. No, youcan't buy one! (Mercedes Benz)

system will choke; the engine quite possiblywon't run at all.

To explain, a stock Bosch EFI system isdesigned to have enough range to cover allconceivable operating conditions for a pro-duction car engine. These can span from-40° F at sea level, in which case the air is

dense, so the engine inhales many oxygenmolecules per mouthful, to 130° F on Pike'sPeak, in which case the air is "thin" with

many fewer oxygen molecules per bite. Thepower produced in the first case would bemore than twice that in the latter! The sys-

124

tern can cope with this range, with a littlemore added as a safety margin. Bewarethere are limits to this, however. On a cold

day at low altitude with high atmosphericpressure, a modified engine capable of pro-ducing, say, 50% more power than stockmight run right out of the available excesscapacity.

System LimitsIgnoring secondary functions, like cold-

starting enrichment and idle stabilization, astock L- or LH system might run out of

600 rpm

80

70

1

50

3000 rpm

80

90

70

50

Injectorsopen

20

30

Injectorsclosed

Injectorsopen

20

30

Injectorsclosed

"head-room" in one of two basic ways(assuming here we are talldng about themaximumrather than the minimumcapaci-

ty). In closed loop mode, the ECU willincrease the quantity of fuel supplied so thatthe lambda sensor reports a stoichiometric

Intermittent injection sys-tems vary the amount of fueldelivered solely by varyingthe "on-time"to "off-time"ofthe electromagnetic injector.At idle, the injectors may onlybe "on" five percent of thetime; at medium revs andload, they may spend morethan half the time "on."

125

6000

90

80

70

50

rpm

Injectorsopen

20

30

Injectorsclosed

At a given supply pressure, the upper limit to rate of fuel delivery may be set by the injectors-eventhough they spend almost all the time "on," they can only just keep up with the demands of an enginemaking maximum power. Bosch recommends that their duty cycle should not exceed 85% or so.

mixture. The limit to this is probably theflow capacity of the injectors. A heavilymodified engine, such as one that has had itsdisplacement greatly increased, mightdemand so much fuel that, at peak torque,the injectors are "on" virtually all the time-they are essentially working as a continuousflow system-yet are still unable to flowenough fuel to meet the engine's needs. Asimilar problem will be encountered if afuel, such as methanol, is used that has astoichiometric air/fuel ratio greater than thatfor gasoline. In practice, the ECU will"know" that more fuel is being demandedthan called for under the worst case, accord-ing to its calibration, and so may regard thelambda sensor as defective and revert to

126

operating on its maps, which would obvi-ously be completely inappropriate.

As long as more than about four to sixvolts is applied to its solenoid coil, an injec-tor will remain "on," flowing fuel, so this100% duty cycle ("on" 100% of the time;"off" 0% of the time) is theoretically attain-able, at least for a short time, but Bosch con-tends that the injectors' duty cycle shouldnot exceed 85%, presumably to avoid over-heating the coils.

Running Open LoopIn open loop mode-that is, with a dis-

connected or inoperative lambda sensor, orwith the engine at full throttle (on mostBosch intermittent systems)-the ECU reg-

--

ulates the injectors' duty cycle based on"maps" stored in its memory. These mapswere developed by dyno tests on a represen-tative stock engine, so a modified enginehaving a greater air capacity than that refer-ence engine at any particular rpm will nec-essarily run lean. Conversely, at certainspeeds a modified engine may handle lessair than the stock one, depending on thenature of the modifications. Lumpy camsthat move the torque peak higher in the rpmrange, for example, will reduce torque atlower speeds below that of a stock engine.The maps know nothing about this, ofcourse, so continue to provide an amount offuel suitable for the stock engine that hap-pens to move more air at that speed, so themixture supplied is too rich.

Provided the basic shape of the torquecurve remains essentially unchanged - suchas would be the case with a moderate dis-

placement increase, but no other modifica-tions- then the maps will produce a mixturestrength that is fairly uniformly lean. In thatcase, an increase in fuel pressure proportion-al to the hike in displacement should pro-duce a serviceable mixture strength.Governed by the maps, the injectors "on"times will be the same as they were, but theywill flow more fuel per cycle because of theincreased supply pressure. Adjustable after-market fuel pressure regulators are availablefrom a wide variety of sources. Note that aregulator need not necessarily be intendedfor a Bosch system, as long as it is ade-quately sized (so it does not itself present arestriction to flow) and can be adjusted tothe desired pressure.

Care should be exercised here; anyincrease in supply pressure brings with it anincreased risk of leaks. In the worst case, ahose or other component might burst. This isclearly hazardous. In addition, the addedload may damage the fuel pump. Boschplumbing components can obviously toler-ate pressures somewhat above their nominalratings, but it could be dangerous and

1.Fullloadcontact

2.Contouredswitchingplate3.Throttle-valveshaft4. Idlecontact5. Electricalconnection

1 52 3 4

At idle and at full-throttle, lambda controlled systems ignore the sensor and revertto "open-loop" operation, using their preprogrammed maps to decide the length ofinjection pulses. The change between modes is signalled by the throttle switch.(Robert Bosch Corporation)

expensive to find out just where their truelimits are. Note, too, that a hike in supplypressure will richen the mixture under allconditions when the engine is operatingopen loop and the ECU reverts to its maps.While this may be harmless as far as itaffects cold starting, the idle mixture maybe so rich as to foul plugs and cause otherdriveability problems.

For off-road use, and assuming no catalyt-ic converter is fitted, an engine running inopen loop mode will benefit from a slightdegree of enrichment anyway. Recall that aconverter needs a mixture held very near

127

a. Richmixture(air deficiency)

b. leon mixture(excessair)

mV a.....

b.....

1000

800

1 1,1 1,2

600

400

200

00,8 0,9

Raising the fuel pressure will increase the flow rate through an injector when it is "on," but this hot-rodding technique will not work onan engine with a working lambda sensor. The sensor will detect the rich mixture and reduce the "on" time accordingly! (Robert BoschCorporation)

128

stoichiometry in order to work properly, butthe mixture strength for peak power is some-what richer than that. This "tweak" cannot

be applied to an engine running in closed-loop mode, however-the ECU, based onthe "rich mixture" signal it is getting fromthe lambda sensor, will simply dial back theinjectors' "on" time to correct for theincreased flow due to the raised system pres-sure.

Fooling the ECUIncreased system pressure can, however,

deal with the problemmentionedearlierof

the fuel demands of a heavily modifiedengine in closed-loop mode exceeding theflow capacity of the injectors. As long as alambda sensor is fitted and the injectors cankeep up with the demand, the ECU will keeptailoring the injectors "on" time to match theengine's air consumption, pretty muchregardless of what the shape of the torquecurve is. While this at first appears to be apromising line of approach, note again thatthe ECU/lambda sensor's goal is a mixturethat is somewhat leaner than that giving bestpower. (And, again, you might bump intothe ECU's preset limits.)

---

Other makeshift means of richening themixture include persuading the ECU that theengine is cold, inducing it to provide a rich-er mixture even when the engine is hot. Onestandard "trick" (a foolish one) is to discon-nect the coolant sensor. On older L-Jetronic

systems, this will work but, as noted above,the mixture will be enriched under all cir-

cumstances, not just at full throttle. (Andnote that the stock system already enrichesthe mixture beyond stoichiometry when thethrottle position sensor indicates full throt-tle.)

On more recent LH-systems, you are like-ly to discover that the ECU detects this tom-foolery. For example, it will not long be sat-isfied with information from the air temper-ature sensor that the air is at 100°F while the

coolant temperature sensor reports that thecoolant remains "cold" after many minutesof running. It will conclude the sensor isdefective. . . and revert to its maps!! It isessentially impossible to "fool" the comput-er-at least in a productive way-and it isstupid to try.

Activating the Cold-Start Injector-Again, some folks have attempted to dealwith a modified engine that needs more fuelat full throttle than provided by the maps(remember, this is what the ECU uses inopen-loop mode, which most systems revertto at full throttle) by activating the cold-startinjector. Yes, more fuel will be supplied,and, yes, the average mixture will thus beenriched. But the position of the cold-startinjector, and the fact that there is just one,pretty much guarantees that some cylinderswill get a lot more than their share and oth-ers a lot less.

Installing NewMapsThe only real solutionto this conundrum

is for the engine to run open-loop,using anew set of maps. With informationfrom athrottlepositionsensorand an enginespeed(or "tach")signal,you have the informationneededto determinehowmuchfuelto inject

Tuned-length intake pipes can provide a degree of "acoustic ramming," pumpingup the torque curve at an rpm that depends on the length of the pipes. Long ones,as on this Audi twin cam V8, beef up the bottom end of the range; short ones addtop end power at the expense of low end torque. (VW Canada)

at any instant, provided you know what theengine's appetite is under any possible com-bination of those factors. That can be

achieved electronically with a set of "maps"particular to the engine, established duringdyno testing.

Eliminating Airflow Meter-Thisapproach has the further advantage that itallows elimination of the airflow meter. (Forall-out race use, it is not really necessary todirectly measure airflow at all, if the elec-tronic "maps" are sufficiently accurate.)While the central airflow sensor- whether

vane-type or the true mass-metering hot-wire and hot-film types - is an asset in aproduction engine, it can be a liability forrace-only use because it prevents the bestpossible exploitation of ram-tuning of theintake pipes; for that, individual stacks areneeded. Getting rid of the air meter alsoeliminates the slight restriction to airflowimposed even by the LH-Jetronic's massflow meter. Purpose-built race engines using

129

~~~~

~ ./70~/' ;:

~/r~ :-

~ .~ :/

~\vr-

~

l]D0.10 D

j0.50 D

-----

PERCENT OFRAM EFFECT

13%

85%

97%

Most of the benefit of tuned "ram" pipes is lost if the entry to the pipe is sharp edged. A smoothlyradiused edge is necessary to get maximum benefit; a large radius is better than a tight one.

Bosch racing EFI systemsoperateon thisbasis.

Dispensingwith the air meter does meanthat racers have to take account of air densi-

ty, just as tuners of carbureted engines do.Higher air density than that prevailing at thetime the maps were compiled requires a cor-responding overall enrichment, and vice-versa. As long as the initial calibration isperformed at an altitude where the enginewill usually be operated and on a day with"typical" barometric pressure, this is a mat-ter that need only concern cost-no-objectrace teams. They swap chips in the ECU toachieve this last fraction of a percent tuning.

In the past, this concept of a custom-

130

---

tailored set of ECU maps was understoodto be theoretically ideal, but was essentiallyunattainable in practice because of the"deep-tech" involved in reprogramming themaps in a stock Bosch ECU. For starters, thelocations of the individual data points in theelectronic memory sometimes differ widelybetween very similar car models, so elec-tronic wizards turned junkyard scroungerswould have to essentially start from scratchevery time around. And Bosch will mostdefinitely NOT release the "codes"! They(or perhaps more correctly, the OEM manu-facturers that sell cars using their EFI sys-tems) are legally obliged to certify theirvehicles' conformance with emissions limits

for 50,000 miles, and to do that they have toensure the systems are tamper-proof.

Given the impracticality of reprogram-ming the Bosch ECD, the only logical alter-native is an aftermarket ECD that is capableof being programmed by the user. A handfulof companies now offer these, together withthe software necessary to allow the pro-gramming to be done from a desktop com-puter. In early versions, a fair bit of comput-er savvy was necessary to do the program-ming, but the more recent versions turn thewhole business into a "point-and-click"operation.

Aftermarket ECU's

Among the first to ride to the rescue in thisway was the Australian firm Haltech. Theirmost recent control module- the E6K, as ofthe time of writing-incorporates an igni-tion timing module, in the same way asBosch's own Motronic systems. In Britain,the Autocar Electrical Equipment Co. pro-duces something very similar under thebrand name "Lumenition." Both systems aresufficiently flexible that they are compatiblewith not just Bosch hardware (injectors,pumps, sensors, etc.) but with componentsfrom virtually any EFI system.

The programming is done in real-time,with the engine under load on a dyno.Working with a straightforward screen dis-play, you simply point and click to increaseor decrease the amount of fuel delivered at

some particular load point. The process isthen repeated for all load points in each rpmrange. Ignition timing data points areentered in a similar way. As long as youhave the tuning savvy to know which direc-tion to adjust mixture strength and sparktiming, the programming procedure is with-in the capability of anyone half-bright withaccess to an engine dyno-no rocket scienceis really required.

A British firm, Jenvey Dynamics, offers throttle bodies in both side-draft and down-draft configurations. These are suitable for use with do-it-yourself EFI systemsusing Bosch fuel-handling and electrical components, and an aftermarket ECU,such as Haltech's. (Jenvey Dynamics Ltd)

131

AutocarUnder the Lumenition brand, Autocar also

offers an aftermarket throttle body assemblythat is a direct bolt-on replacement forWeber carburetors, in both side-draft anddown-draft versions, intended for use withtheir ECU (but equally compatible in princi-ple with Haltech or other ECUs) and Boschor other fuel system hardware. Autocar arealso the source for components under theMicro Dynamics brand name, having takenover that company. (It is worth mentioninghere that the Micro Dynamics rising-ratepressure regulator, previously offered forboth intermittent and for continuous EFI

systems, is no longer available. Autocarcites a combination of insufficient demand

and some technical shortcomings for dis-continuing this product).

Apart from aftermarket ignition parts,including a spark retard module for tur-bocharged EPI engines, Autocar offers theMicro Dynamics PIC5-a supplementaryelectronic module and extra injector that isprimarily intended for temporary full-throt-tle enrichment for turbocharged engines.Autocar correctly points out that while thePIC5 can be fitted to a normally aspiratedengine, it is definitely not advisable becauseof the same problems noted above withregard to use of the cold-start injector as ameans of supplementary enrichment-uneven distribution of mixture strength fromcylinder-to-cylinder.

Racing Gasoline"There is nothing special about gasoline,"

one fuel engineer has explained, "it is sim-ply something that you can make from crudeoil that engines can run on." And unlikemany industries that produce a saleableproduct from a natural source, fuel refinersthrow nothing away- there is no sawdustand bark left over after the tree is sawn into

lumber, no chaff left over after the wheat ismilled. These folks sell absolutely every-thing!

132

Refining is partly a process of sorting thebad stuff from the good stuff, and partly amatter of converting some of the rubbishinto saleable product. But this is an expen-sive business, so the process is carried onlyfar enough to make a serviceable fuel.

The hundreds of different hydrocarboncompounds that make up a typical gasolineblend have widely differing properties.Their specific gravities vary from 0.500 to0.879; their octane number can be anywherefrom 0 to 120; they have different heat ener-gy contents- from 0.24 to 0.39 BTU per lb.

One characteristic that distinguishes adesirable hydrocarbon for high performanceuse is a high octane number. Another is ahigh heating value. Although there is only aslight difference between one hydrocarbonand another in the amount of heat producedby burning, say, one cubic foot of air/vapormix, there is nevertheless a difference. Thereis something like 3-4% more power to behad by burning a pure sample of the mostenergetic hydrocarbon, compared to typicalpump fuel.

By a cruel fluke of nature, the hydrocar-bons that rank highest in heating value aregenerally the worst in terms of octane, with afew significant exceptions. The exceptionsare found among the aromatics- benzene,toluene, and various xylenes-and the alky-lates (also called isoparaffms). Aviation gas,for instance, has lots of alkylates, and manypump super premiums contain plenty of aro-matics. (No benzene, though, beyond a meretrace-benzene is a known carcinogen.)These high-quality components are compara-tively difficult and expensive for a refinery toproduce, however, so for most of a humanlifetime refiners added tetraethyllead (TEL),which works like a negative catalyst, slowingdown the chemical reactions that are the root

cause of detonation. But TEL-long knownto be lethally toxic-was banned from gaso-line, perhaps not so much because of its directeffect on human health but rather because

you cannot meet emission standards without

a catalytic converter, and leaded fuel "poi-sons" the converter, rendering it ineffective.

Lead AdditivesAnd that's where the race fuel wizards

come in. Working solely with hydrocarbons,it might be just barely possible to produce alOO-plus octane gasoline, but it certainlywouldn't make economic sense. Here, racefuel producers do the same as the majorrefineries do- they use non-hydrocarbonadditives. Government regulations for pumpgas permits limited amounts of certain ofthese additives; gasoline rules for manyclasses of motorcycle, oval track, and roadracmg are more generous.

While many, if not most, racing gasolinescontain a significant amount of lead, thereare some unleaded racing gasolines, basedmostly on alkylates and aromatics. Tosqueeze the last bit of octane out of thesecarefully selected base stocks, most unlead-ed race gas has a lead substitute added. Theadditive, called MMT -methylcyclopenta-dienyl manganese tricarbonyl is the activeingredient in many octane-boosting addi-tives. Many premium pump fuels also con-tain MMT, although this varies from state-to-state.

Even though MMT is about the best leadreplacement available, it doesn't work near-ly as well as lead, so even if you start offwith the select mixture of hydrocarbonsfound in race gas, you're still scraping fordecent octane numbers no matter how much

MMT you add. To bump up the detonationresistance further, unleaded race gasolinesalso contain a fat hit of MTBE-methyl ter-tiary butyl ether which, by itself, has ahealthy octane number of 114 (ResearchMethod). Not only does MTBE jack upoctane, it is also an oxygenate-it containsoxygen-and so directly increases power. Afuel carrying its own oxygen adds to theamount the engine inhales, increasing powerroughly in proportion. As much as 11%MTBE can be found in premium unleaded;

some racing UL's contain twice that much.Because of the oxygenate content, the

mixture has to be richened a little. A motor

that is borderline lean on straight hydrocar-bon gasoline can slip right over the edgewhen it gets more oxygen than there is fuelto oxidize. As explained above, the pro-gramming in recent Bosch electronic FI sys-tems has enough "head-room" to cope withthe added fuel flow requirements of theseoxygenated fuels. Systems with a lambdasensor will simply crank up the mixturestrength until the sensor reports stoichiome-try. The engine will then be ingesting justenough additional hydrocarbon molecules tobum in the few percent more oxygen mole-cules provided by the fuel, and power willbe increased in proportion.

In drag racing, there's another problemwith the oxygen content. NHRA definesgasoline a little differently than the rulemakers in many other forms of racing. Byspecifying a maximum dielectric (electricalinsulating) value for gasoline, the NHRArules effectively place a strict limit on theamount of any added oxygen bearing com-pounds, including MTBE.

AvgasBefore the development of race-only

gasolines, some performance enthusiastssoughtincreasedpowerfrom aviationgaso-line- Avgas.In fact,Avgashas little if anyedge over ordinary pump fuel in terms ofheatingvalue;any increasedperformanceitmightprovideis attributablealmostentirelyto the fact thatAvgas'shigheroctaneallowsa highercompressionratio to be used with-out runningintodetonation.

Types of AvgasAvgas comes in 2 (or maybe 3) grades;

each is identified by a performance num-ber - not to be confused with an octanenumber. In contrast to the method used toestablish the Research and Motor octane

numbers (see the Sidebar "Doing Octane

133

Numbers" in Chapter 1), the test proceduresfor establishing the anti-knock properties ofaviation gasolines vary not compressionratio, but inlet air pressure, in reflection ofthe fact that the tests were specifically devel-oped for highly supercharged aircraftengines. The tests are also conducted at ahigher rprn.

Two numbers fallout from this test: alower number at a lean air/fuel ratio and a

higher one established with a very rich mix-ture (a supercharged aircraft engine at take-off power setting is running as much as 60percent richer than stoichiometric!). Thetwo numbers are separated by a slash, e.g.100/130.

The most humble grade of Avgas islabelled 80/87, and is colored red. In termsof octane rating - whether Research, Motor,or "pump" numbers-this would certainlyrate lower than even pump regular unleaded.It is intended for the few remaining ancientaircraft engines of 6: 1 or 7: 1 compressionratio, and thus is extremely rare; it is entire-

ly unsuitable for any automotive use, any-way.

By far the most cornmon type of Avgas isdesignated 1O0LL-1O0 low-lead. Thisreplaces 100/130, which had a high leadcontent and was colored green. Tested by theMotor and Research methods, 100LL (blue)Avgas rates at about 100 octane on both

scales. Finally, and scarcely worth mention-ing, is (or perhaps more correctly was) pur-ple colored 115/145-it is essentiallyunavailable unless you happen to be the USAir Force.

Low-lead does not mean no lead, so 100LL

Avgas cannot be used in a vehicle that depends ona catalyticconverter.For race-only use, Avgas is afeasible alternative to purpose-specific racinggasoline. It is arguably easier on plastics and elas-tomers (synthetic rubbers and rubber-like materi-als) than racing gasolines containing a high pro-portion of aromatics, and being free of oxy-genates, should requITe almost no change of over-

all air/fuel ratios compared to ordinary pump gas.

134

---~

Methanol andMethanol/Gasoline Blends

The first racing use of methanol was in the1920s, and it is still preferred by many rac-ers for its cool-running, engine-friendlyqualities, plus its ability to substantiallyincrease power and torque. Methanol showsan octane rating of way over 100, which per-mits much higher compression ratios, and somore power.

Alcohol's cooling properties are also asource of increased power. Methanol takesmuch more heat to evaporate than gas (473BTU/lb vs. 130-140 BTU/lb) and you canmultiply that difference by two or more,since it takes twice as much alcohol as gas tomake a chemically correct mixture with air,and it ignites even when it is 40% richer thanthat. All told, that makes for a sizeablerefrigeration effect, and so a denser chargein the cylinder.

Unfortunately, methanol is somewhatcorrosive to many metals, and severelydegrades various plastics and elastomers.According to Bill Roth, who wasinvolved in the development of the origi-nal D-Jetronic system, no Bosch system,indeed "no fuel injection system is resistantto methanol, ".over the long term, althoughhe acknowledges that short term exposure isprobably OK.

To take full advantage of the internal cool-ing properties of alky, methanol burnersmay consume nearly three times as muchfuel as comparable gasoline-fuelled racers.This thirst can be a problem: the addedweight of the alky may hinder performanceand/or demand more time spent in the pitsfor refuelling. In some cases it may even beworth sacrificing some power to save onweight or pit-stops.

In the '30s, for instance, methanol/ben-zene/gasoline blends were popular for use inOffy engines racing at Indy. Likewise, untila gasoline-only rule was implemented in1957, Formula One teams favored alcohol-

--- -

based fuels, yet few if any of them ever ranmethanol straight-all the blends used bythe top teams also included a significantamount of gasoline.

There is reason to believe that the gas wasnot in there purely to reduce the fuel load. Insome engines, a gasoline/methanol blendmay provide more power than methanolalone. For example, on straight meth the1954 Maserati Formula One motor deliv-

ered about 230 bhp, not bad for 154 unsu-percharged cubic inches, nearly 50 yearsago. But when the alcohol was cut withgasoline, benzol and acetone, the power roseto 242 bhp.

Problems With Blends

Given the problems of arranging to doubleor triple the fuel handling capacity of an EFIsystem, this notion of using amethanol/gasoline blend seems worth con-sideration. Mixtures of gasoline andmethanol, however, are tougher on certainrubber and plastic fuel system componentsthan either the gas or the alcohol alone. Incertain proportions, a gasoline/methanolmix has a mighty appetite for magnesium,and the presence of other fuel compo-nents-even small amounts of lubricatingoil or minor impurities in the methanol, andespecially added aromatics-can have fur-ther serious solvent and corrosion effects.

There are other problems with blendinggasoline and methanol. For one thing, whilea little bit of water in straight methanol mayactually help performance, even a smallamount of water will cause a mixture of

methanol and gas to separate into two parts,with the gas floating on top of the alky.Evenif the alcohol is completely anhydrous (dry)when you buy it, it has the nasty habit ofsoaking up water from the atmosphere, somost people who run methanol ormethanol/gas blends have learned to usetightly sealed containers, and to tape overtank vents whenever the race vehicle is not

running. And because methanol does not

have the slight lubricating properties ofgasoline, don't forget to add a dash of alco-hol-soluble oil to the race fuel, to keep thevalve stems and cylinder walls happy.

Co-Solvents- There are certain blendingagents that help prevent this tendency ofgas/alky mixtures to separate. Among themare other alcohols like isopropanol (IPA)and tertiary butyl alcohol (TBA). IPA andTBA are readily available, and since they areboth alcohols, they may be legal as blendingagents under some racing rules. Unfor-tunately, they are only marginally effectiveas co-solvents for gas/alky mixtures.

Aromatics-Other, more effective co-sol-vents include the aromatic hydrocarbonslike benzene, toluene, and xylene. Benzeneis a pretty decent fuel itself, but it was a largepart of the mix in some older fuel blendslargely because of its effectiveness at co-dis-solving methanol and gasoline when there issome water present. Benzene has since beenfound to cause cancer, so it is difficult tobuy, especially in drum quantities.

The dangers of benzene and the relativeineffectiveness of IPA and TBA brings usback to toluene, xylene, and a few other aro-matic hydrocarbons as gas/methanol/waterblending agents. Toluene and xylene are alsoreadily available, but these are powerful sol-vents - they will eat certain fuel systemcomponents.

What can be said for certain here is that

Bosch-based EFI systems have been usedsuccessfully to deliver methanol and gaso-line/methanol blends, but that componentlife is shortened. Further, to achieve evenmodest survivability, the system has to bepurged completely after each race weekend.After draining the tank, the engine is thenrun until it stalls from lack of fuel, then thesystem flushed through with straight gaso-line, to remove all traces of the methanol.

135

-- - --

Born in Britain, Forbes Aird

moved to Canada at the age ofnine. He is gradually getting

used to the climate. For several years,Forbes's career alternated betweenadministrative/research work in various

academic institutions, and self-employ-ment-which included several yearsoperating a small specialty reinforcedplastics shop, and a financially doomedattempt to build a limited productionsports car using RP stressed skin con-struction.

First published in 1965, he began writ-ing full time in 1986. Despite strong evi-dence to the contrary, he retains the con-viction that he can make a living in thisway. Over 100 of his automotive, aero-nautical, and science articles have

appeared in more than a dozen differentmagazines in Canada and the U.S. In1991 he received a Science JournalismAward from The Canadian Science

Writers' Association; in 1992 theInternational Motor Press Association

bestowed on him the prestigious KenPurdy Award. Forbes is the author ofeleven books, including HPBooks'Fiberglass and Other Composite Mater-ials, High Performance Hardware, andAerodynamics for Racing and Perfor-mance Cars.

Forbes works and lives in Toronto with

his wife, Kate. When not slaving over ahot keyboard, he spends much of histime cooking Thai and Indian curries.

137