linear technologylinear technology - analog.com · linear technologylinear technology may 2000...
TRANSCRIPT
LINEAR TECHNOLOGYLINEAR TECHNOLOGYLINEAR TECHNOLOGYMAY 2000 VOLUME X NUMBER 2
Monolithic SynchronousStep-Down RegulatorsPack >500mA OutputCurrent in anMS8 Package by Jaime Tseng and
Gary Shockey
IntroductionThe quest to pack more power intoportable electronic devices whileshrinking their size has placedincreased demands on power man-agement products. Not only must theybe physically smaller, but they mustalso retain the power handlingcapability of their older, larger coun-terparts. The LTC1877, LTC1878 andLT®1612 are the first of a new genera-tion of monolithic synchronousstep-down switching regulators ca-pable of supplying more than 500mAof output current in an MS8 package.Their internal synchronous switchesincrease efficiency and eliminate theneed for external Schottky diodes,saving external components andboard space. Optimized for battery-powered applications, the LTC1877works with a supply range of 2.65V to10V, the LTC1878 works with sup-plies of 2.65V to 6V and the LT1612works with supplies of 2V to 5.5V.This wide operating supply range cov-ered by the two parts allows the use ofa single or dual Li-Ion battery or 2- to6-cell NiCd and NiMH battery packs.
LTC1877/LTC1878 vs LT1612The LTC1877and LTC1878 aredesigned in Linear Technology’s high
performance BiCMOS process andare optimized for ultrahigh efficiencyat low load currents. The DC supplycurrents of both parts are only 10µAwhile maintaining the output voltage(using Burst Mode™ operation) at noload. This enables both parts to main-tain better than 90% efficiency overthree decades of output load current.The LT1612, designed in LinearTechnology’s high speed bipolar pro-cess, is optimized for low voltageoperation and is a lower cost alterna-tive to the LTC1877/LTC1878. TheLT1612 can still regulate the outputwhile supplying 500mA to the loadwith input supply voltages as low as2V. The internal frequency is set to800kHz, compared to 550kHz for theLTC1877. The DC supply current is150µA. Placing the part in shutdownreduces the current to 1µA.
A Detailed Look at theLTC1877/LTC1878The LTC1877 and LTC1878 are twonearly identical parts in that theyhave exactly the same functionality,architecture and pinout. What dis-tinguishes the two parts is theirmaximum supply voltages: the
, LTC and LT are registered trademarks of Linear Technology Corporation. Adaptive Power, Burst Mode, C-Load,FilterCAD, Hot Swap, LinearView, Micropower SwitcherCAD, Multimode Dimming, No Latency ∆Σ, No RSENSE,Operational Filter, OPTI-LOOP, Over-The-Top, PolyPhase, PowerSOT, SwitcherCAD and UltraFast are trademarks ofLinear Technology Corporation. Other product names may be trademarks of the companies that manufacture theproducts.
continued on page 3
IN THIS ISSUE…COVER ARTICLEMonolithic Synchronous Step-DownRegulators Pack 500mA OutputCurrent in an MS8 Package..................................................1Jaime Tseng and Gary Shockey
Issue Highlights ....................... 2
LTC® in the News .......................2
DESIGN FEATURESNew Rail-to-Rail Output Op AmpsBring Precision Performance toLow Voltage Systems .................6Alexander Strong and Gary Maulding
DC Accurate, Rail-to-Rail Active RCLowpass Filter Replaces DiscreteDesigns—One Resistor Value Setsthe Cutoff Frequency............................................... 11Doug La Porte
SOT-23 Switching Regulator withIntegrated 1 Amp Switch DeliversHigh Current Outputs ............. 17Albert Wu
Dual Operational AmplifierCombines 16-Bit Precision withHigh Speed............................................... 18Kris Lokere
New Instrumentation Amplifier:Single-Resistor Gain Setand Precision Front End MakeAccuracy Easy ........................ 21Glen Brisebois
DESIGN IDEAS.......................................... 23–36complete list on page 23
New Device Cameos ..................37
Design Tools ............................39
Sales Offices ............................40
Linear Technology Magazine • May 20002
EDITOR’S PAGE
Issue HighlightsOur feature article this issue
introduces three new synchronousstep-down regulators. The LTC1877/LTC1878 and LT1612 are newmonolithic synchronous step-downswitching regulators capable of sup-plying 500mA of output current in anMS8 package. Their internal synchro-nous switches increase efficiency andeliminate the need for externalSchottky diodes, saving componentsand board space. Optimized for bat-tery-powered applications, theLTC1877/LTC1878 work with sup-ply ranges of 2.65V to 10V and 2.65Vto 6V, respectively, and the LT1612works with supplies of 2V to 5.5V.
Another power product introducedin this issue is the LT1930. TheLT1930 is the only SOT-23 switchingregulator in the industry with anintegrated 1A switch. The LT1930uses a constant frequency, internallycompensated, current mode PWMarchitecture. Its 1.2MHz switchingfrequency allows the use of tiny, lowcost capacitors and low profile induc-tors. With an input voltage range of2.6V to 16V, the LT1930 is a good fitfor a variety of applications. Theonboard switch features a low VCESATvoltage of 400mV at 1A, resulting invery good efficiency even at high loadcurrents.
This issue is strong on signal con-ditioning products, including anumber of new op amps, a newinstrumentation amp and an RC activelowpass filter.
The LT1677 and the LT1881,LT1882, LT1884 and LT1885 are pre-cision op amps designed for use inlow voltage systems. The LT1677 is arail-to-rail input, rail-to-rail output,single-supply version of the industry-standard LT1007. It features thelowest noise available for a rail-to-railop amp. Low noise is combined withoutstanding precision: the CMRR andPSRR are 130dB, the offset voltage isonly 20µV and the open-loop gain istwenty million. The LT1677 is unity-gain stable and has a gain bandwidthproduct of 7.2MHz.
LTC in the News…On April 19th, Linear Technology Cor-poration announced its financialresults for the third quarter of fiscalyear 2000. Robert H. Swanson,Chairman and CEO, stated, “TheMarch quarter was an outstandingquarter for us, as we achieved recordlevels of bookings, sales and profits,with sales increasing 14% and profits17% sequentially from the Decemberquarter. Our return on sales was arecord 41%. Demand from ourcustomers was very robust, escalat-ing during the quarter and increasingin all major geographical areas andall major end markets. Given thisrobust business activity, we expectthe upcoming June quarter to alsohave continuous sequential sales andprofit growth.” The Company reportedsales of $185,075,000 and net incomeof $75,867,000 compared with$49,828,000 a year ago. Net saleswere up 42% over last year.
The Company was named to theS&P 500 on March 31st. According toRobert H. Swanson, Chairman of theBoard and CEO, “We are honored tobe added to the prestigious S&P 500.This is a very significant milestonein the development of Linear Tech-nology Corporation. We were foundedin 1981 with the belief that a pureplay analog semiconductor companycould be both very profitable and alarge company. The S&P 500 listingacknowledges that the Company isone of the most significant 500 pub-licly traded companies, since it listscompanies that are considered lead-ing companies in leading industrieswithin the US economy. This willbenefit our shareholders in that itshould broaden the holding of ourstock. It is also a gratifying accom-plishment to the many dedicatedemployees whose efforts made ithappen.”
The LT1881 dual and LT1882 quadop amps feature 150pA input biascurrents, whereas the similar LT1884and LT1885 dual and quad op ampstrade slightly higher input bias cur-rents of 500pA for three times higherspeed. Their bias current specifica-tions, coupled with 50µV offsetvoltage, open-loop gains of over onemillion and high common moderejection, allows precision accuracyto be maintained in systems withhigh source impedances.
Another new op amp debuted inthis issue is the LT1469, a dualoperational amplifier that is optimizedfor accuracy and speed in 16-bit sys-tems. The amplifier settles in just900ns to 150µV for a 10V step. TheLT1469 also features the excellentDC specifications required for 16-bitdesigns. Input offset voltage is 125µVmaximum, input bias current is 10nAmaximum for the inverting input andminimum DC gain is 300V/mV.
LTC’s newest instrumentationamplifier, the LT1168, is a lowpower, single-resistor gain-program-mable instrumentation amplifier thatis easy to apply. With negligible 60µV(Max) offset voltage and ultrahigh 1TΩinput impedance, it can sense bridgeswith source impedances from 10Ω to100k without degrading the signal.Its 120dB CMRR at 60Hz is achievedeven with a 1k source impedanceimbalance. Matching the LT1168’sCMRR using discrete op amps wouldrequire the use of 0.001% resistors.
The LTC1563-2 and LTC1563-3comprise a new family of extremelyeasy to use, 4th order active RC low-pass filters. Their cutoff frequenciesrange from 256Hz to 256kHz whileoperating at supplies from as low as asingle 3V up to ±5V. The LTC1563also features rail-to-rail input andoutput operation with, typically, 1mVof DC offset. The design of the mostpopular filter responses is trivial, re-quiring only six resistors of identicalvalue and no external capacitors.
Our Design Ideas section featuresan SMBus fan controller for portable
devices, a VID-controlled 42A supplyfor the AMD Athlon™ processor, theconclusion of the ADSL Line DriverDesign Guide, begun in the Februaryissue, and a resistance-measuringcircuit using the new LT1168 instru-mentation amp. The issue concludeswith six New Device Cameos.Athlon is a trademark of Advanced Micro Devices, Inc.
Linear Technology Magazine • May 2000 3
DESIGN FEATURES
LTC1877 is designed for higher inputvoltage applications, whereas theLTC1878 is optimized for lower inputvoltage applications. For example, theLTC1877 provides up to 600mA ofoutput current at an input voltage of5V, whereas the LTC1878 providesthe same amount of current at aninput voltage of only 3.3V.
Both the LTC1877 and LTC1878incorporate a constant frequency,current mode step-down architecturewith on-chip power MOSFETs.
The LTC1877/LTC1878 includeprotection against output overvolt-age, output short-circuit and poweroverdissipation conditions. When anovervoltage condition at the output(>6.25% above nominal) is sensed,the top MOSFET is turned off untilthe fault is removed. When the outputis shorted to ground, the frequency ofthe oscillator slows to prevent induc-tor-current runaway. The frequencyslows to about 80kHz or one-seventhof the nominal frequency. The fre-quency returns to 550kHz (or the
external synchronized frequency)when VFB is allowed to rise to 0.8V.When there is a power overdissipationcondition and the junction tempera-ture reaches approximately 145°C,the thermal protection circuit turnsoff the power MOSFETs allowing theLTC1877/LTC1878 to cool. Normaloperation resumes when the tempera-ture drops by 10°C.
Burst Mode OperationThe LTC1877/LTC1878’s Burst Modeoperation is enabled by simply strap-ping the SYNC/MODE pin to VIN orconnecting it to a logic high (VSYNC/
MODE >1.2V). In this mode, the peakcurrent of the inductor is set toapproximately 250mA, even thoughthe voltage at the ITH pin (the outputof the error amplifier) would reflect alower value. The voltage at the ITH pindrops when the inductor’s averagecurrent is greater than the loadrequirement. When the ITH voltagedrops below approximately 0.6V, asleep signal is generated, turning offboth power MOSFETs. The ITH pin is
then disconnected from the output ofthe error amplifier and “parked” adiode voltage above ground. Duringthis time, the internal circuitry ispartially turned off, reducing the qui-escent current to 10µA; the loadcurrent is now supplied by the outputcapacitor. When the output voltagedrops by an amount dependent onthe output voltage (on the order of10mV for a 2.5V output), the ITH pinreconnects to the output of the erroramplifier, the top MOSFET is againturned on and the process repeats.
For frequency-sensitive applica-tions, Burst Mode operation isdisabled by connecting the SYNC/MODE pin to GND. In this case,constant-frequency operation is main-tained at lower load currents togetherwith lower output ripple. If the loadcurrent is low enough, cycle skippingwill eventually occur to maintainregulation. In this mode, the efficiencywill be lower at light loads, butbecomes comparable to Burst Modeoperation when the output loadexceeds 50mA.
1200
1000
800
600
400
200
02 10
VIN (V)
MAX
IMUM
OUT
PUT
CURR
ENT
(mA)
4 6 8
VOUT = 1.5V
VOUT = 2.5V
VOUT = 3.3V
VOUT = 5V
L = 10µHLTC1877
95
90
85
80
75
70
65
60
55
500.1 1.0 10 100 1000
OUTPUT CURRENT (mA)
EFFI
CIEN
CY (%
)
VIN = 10V
VIN = 7.2V
VIN = 5V
VIN = 3.6V
L1 = 10µHVOUT = 2.5V
LTC1877
100
80
60
40
20
00.1 1.0 10 100 1000
OUTPUT CURRENT (mA)
EFFI
CIEN
CY (%
)
VIN = 10V
VIN = 5V
VIN = 3.6V
L1 = 10µHVOUT = 2.5V
VIN = 7.2V
LTC1877
Figure 2. Maximum output current vsinput voltage
Figure 4. Efficiency vs load current for Figure3’s circuit (Burst Mode operation enabled)
Figure 5. Efficiency vs load current for Figure3’s circuit (Burst Mode operation disabled)
1.2
1.0
0.8
0.6
0.4
0.2
0.00 1 2 3 4 5 6 7 8 9 10
INPUT VOLTAGE (V)
R DS(
ON) (
Ω)
SYNCHRONOUS SWITCH
LTC1877
LTC1878
MAIN SWITCH
MAIN SWITCHSYNCHRONOUS SWITCH
7
6
1
2
5
8
3
4
SYNC/MODE
VIN
RUN
ITH
SW
PLL LPF
VFB
GND
LTC1877
CIN10µF10V
CITH220pF
L1 10µH
COUT47µF6.3V
R1877k
1%
R2412k
1%
CFW
20pF
VOUT2.5V/500mA
VIN3.6V–10V
CIN:COUT:
L1:
TAIYO YUDEN LMK325BJ106MN (408) 573-4150SANYO 6TPA47M (619) 661-6835SUMIDA CD54-100 (847) 956-0667
Figure 1. RDS(ON) for both switches vsinput voltage
Figure 3. 2.5V/500mA step-down regulator
LTC1877/LT1612, continued from page 1
Linear Technology Magazine • May 20004
DESIGN FEATURES
Frequency SynchronizationA phase-locked loop (PLL) on theLTC1877/LTC1878 allows the oscil-lator to be synchronized to an externalsource connected to the SYNC/MODEpin. The output of the phase detectorat the PLL LPF pin operates over a 0Vto 2.4V, range corresponding to400kHz to 700kHz. When locked, thePLL aligns the turn-on of the topMOSFET to the rising edge of thesynchronizing signal. Burst Modeoperation is disabled when theLTC1877/LTC1878 is synchronizedto an external source. Frequency syn-chronization is inhibited when thefeedback voltage, VFB, is below 0.6V.This prevents the external clock frominterfering with the frequency fold-back for short-circuit protection.
Low Input Supply andOperation in DropoutThe LTC1877/LTC1878 can operateon an input supply voltage as low as2.65V. However, the maximum allow-able output current is reduced at thislow voltage due to an increase in theRDS(ON) of the P-channel MOSFET. SeeFigure 1 for a graph of switch resis-tance vs input voltage. Figure 2 showsthe reduction in the maximum out-put current as a function of inputvoltage for various output voltages.
The LTC1877/LTC1878 is capableof turning the main P-channel MOS-FET on continuously (100% dutycycle) when the input voltage falls tonear the output voltage. In this drop-out mode, the output voltage isdetermined by the input voltage mi-nus the voltage drop across the
internal MOSFET and the inductorresistance.
2.5V/500mAStep-Down RegulatorA typical circuit using the LTC1877 isshown in Figure 3. This design sup-plies a 500mA load at 2.5V with aninput supply between 3.6V and 10V.The circuit operates at the internallyset frequency of 550kHz. A 10µHinductor is chosen so that theinductor’s current remains continu-ous during burst periods at low loadcurrent. For low output voltage ripple,a low ESR capacitor is used. All thecomponents shown in this schematicare surface mount and have beenselected to minimize the board spaceand height.
Efficiency ConsiderationsThe efficiency curves for the 2.5V/500mA regulator at various supplyvoltages are shown in Figure 4. Note
+
BURSTMODE
SHDN
MODE
SW
FB
VIN BOOST
VC GND
VIN2V
VOUT0.9V/500mA
C110µF
0.1µF
L1 10µH
R1105k
R2232k
100pF
C247µF3.15V
33k
330pF
LT1612
C1: TAIYO YUDEN JMK325BJ106MN (408) 573-4150C2: PANASONIC EEFCDOF680R (714) 373-7334L1: SUMIDA CD43-100 (847) 956-0667
80
70
60
50
40
30
20
101 10 100 1000
LOAD CURRENT (mA)
EFFI
CIEN
CY (%
)
VIN = 2V (LINEAR)
VIN = 2V (LT1612)
VIN = 3V (LT1612)
VIN = 3V (LINEAR)
BURSTMODE
SHDN
MODE
SW
FB
VIN BOOST
VC GND
VIN5V
VOUT3.3V/500mA
C110µF
0.1µF
L1 10µH
R11.0M1%
R2232k1%
20pF
C222µF
33k
680pF
LT1612
C1: TAIYO YUDEN JMK325BJ106MN (408) 573-4150C2: TAIYO YUDEN JMK325BJ226MNL1: SUMIDA CD43-100 (847) 956-0667
85
80
75
70
65
60
55
501 10 100 1000
LOAD CURRENT (mA)
EFFI
CIEN
CY (%
)
Figure 6. 2V to 0.9V step-down converter
Figure 7. Efficiency curves for the LT1612 2Vto 0.9V converter and a theoretical linearregulator
Figure 8. 5V to 3.3V/500mA converterFigure 9. 5V to 3.3V converter efficiencypeaks at 83%
Linear Technology Magazine • May 2000 5
DESIGN FEATURES
the flatness of the curves over theupper three decades of load currentand that the efficiency remains highdown to extremely light loads. Effi-ciency at light loads requires lowquiescent current. The curves are flatbecause all significant sources of lossexcept for the 10µA standby cur-rent—I2R losses in the switch, internalgate charge losses (to turn on theswitch) and burst cycle DC supplycurrent losses—are identical duringeach burst cycle. The only variable isthe rate at which the burst cyclesoccur. Since burst frequency is pro-portional to load, the loss as apercentage of load remains relativelyconstant. The efficiency drops off asthe load decreases below about 1mAbecause the non-load-dependent10µA standby current loss then con-stitutes a more significant percentageof the output power. This loss is pro-portional to VIN and thus its effect ismore pronounced at higher input volt-ages. Figure 5 shows the effect onefficiency of disabling Burst Modeoperation.
LT1612 DetailsLike the LTC1877/LTC1878, theLT1612 also uses constant-frequency,current mode control. It is capable ofBurst Mode operation or constant-frequency switching. Unlike theLTC1877/LTC1878, it uses bipolarpower transistors instead of MOSswitches. One of the main designchallenges was providing the abilityto operate with inputs as low as 2V.
To achieve this low voltage operationin a buck switching regulator, a bipo-lar NPN topside power switch isneeded, rather than a MOS device.Because the NPN power transistorrequires significant base drive cur-rent, efficiency is not as high as withits MOS counterpart. Thus, a smallsacrifice in efficiency is made for thislow voltage operation. The alternativeto using the LT1612 to step downfrom low voltages is to use a linearregulator, which, of course, has itsown disadvantages. Linear regulatorsare inherently inefficient because thepower device is operated in the linearregion. They must be physically largerto allow for dissipation of the extraheat generated and heat sinks areoften needed, even at modest outputpower.
ApplicationsFigure 6 shows the LT1612 convert-ing a 2V input down to 0.9V. Theinternal reference is set at 0.62V,which allows outputs below 1V. Thegraph in Figure 7 compares efficiencyfor the LT1612 to that of a theoreticallinear regulator. At an input voltage of2V and a load current of 200mA, theefficiency is 70% for the circuit inFigure 6 vs 45% for the linear regula-tor. At an input voltage of 3V, thelinear regulator’s efficiency is just30%, while that of the circuit usingthe LT1612 drops to 65%. This clearlyillustrates the power savings of theLT1612 as compared to a linearregulator.
The LT1612 is also well suited formore general purpose applications,such as the circuit shown in Figure 8.Here, the LT1612 is shown steppingdown 5V to 3.3V. Efficiency, graphedin Figure 9, reaches 83% at a loadcurrent of 300mA. Maximum outputpower for this configuration is nearly2W. Figure 10 shows transientresponse to a 300mA load step withBurst Mode enabled. If low noiseoperation is desired, the MODE pincan be pulled high, giving the responseseen in Figure 11. The low frequencyoutput voltage r ipple is noweliminated.
ConclusionThe LTC1877/LTC1878 and LT1612are well suited for medium to lowpower step-down applications withtight board space requirements. Thesesynchronous buck regulators candeliver 500mA of output current andcover the input voltage range of 2V to10V. The LTC1877/LTC1878 switchat 550kHz and offer the highestperformance possible with efficiencyexceeding 90%. An internal phase-locked loop al lows frequencysynchronization from 400kHz to700kHz. The LT1612 operates at800kHz and exhibits less than onefourth the power loss of a linear regu-lator at an input of 3V. All three partscome in the MS8 package and requireminimal external components, whichallows them to meet the tightest spacerequirements.
VOUTAC COUPLED
100mV/DIV
IL200mA/DIV
ILOAD10mA TO 300mA
0.1ms/DIV
VOUTAC COUPLED
200mV/DIV
IL200mA/DIV
ILOAD10mA TO 300mA
0.1ms/DIV
Figure 10. Transient response for Figure 9’s circuit Figure 11. With the MODE pin high, low frequency rippleat VOUT is eliminated
Linear Technology Magazine • May 20006
DESIGN FEATURES
New Rail-to-Rail Output Op AmpsBring Precision Performance toLow Voltage Systems by Alexander Strong
and Gary MauldingIntroductionLinear Technology has recentlyreleased several new high precisionop amps for use in low voltage sys-tems. The LT1677, LT1881, LT1882,LT1884 and LT1885 all operate onpower supplies from 3V or lower up to36V and have rail-to-rail output volt-age swing. These amplifiers allow highprecision circuits to be implementedon low voltage power supplies,including single positive supplies.Rail-to-rail output stages maintainthe output signal dynamic range byeliminating the base-emitter voltagedrops of conventional emitter-followeroutput stages. Offset voltages are trim-med to less than 80µV, with the lowtemperature drift and low noise to beexpected from bipolar transistordesigns. High open-loop voltage gainsmaintain this accuracy over the out-put swing range.
The LT1677 is a rail-to-rail input,rail-to-rail output, single-supply ver-sion of the industry-standard LT1007.It features the lowest noise availablefor a rail-to-rail op amp: 3.2nV/√Hzand 70nV peak-to-peak 0.1Hz to 10Hznoise. An important feature in lowvoltage, single-supply applications (aslow as 3V) is the ability to maximizethe dynamic range. The LT1677’sinput common mode range can swing
100mV beyond either rail and theoutput is guaranteed to swing towithin 170mV of either rail whenloaded with 100µA. Low noise is com-bined with outstanding precision: theCMRR and PSRR are 130dB, the off-set voltage is only 20µV and theopen-loop gain is twenty-five million(typical). The LT1677 is unity-gainstable and has a gain bandwidth prod-uct of 7.2MHz. Figure 1 shows theinput and output of an LT1677 infollower mode (gain = 1) using a single3V supply. The output clips cleanly atthe rails with no phase reversal, evenwhen the input exceeds the rail by0.5V. This has the advantage of elimi-nating lockup in servo systems.
The LT1881 dual and LT1882 quadop amps feature 150pA input biascurrents, whereas the similar LT1884and LT1885 dual and quad op ampstrade slightly higher input bias cur-rents of 500pA for three times higherspeed. This series of amplifiers bringsthe performance of the LT1112 to lowvoltage applications that need thewide rail-to-rail output dynamicrange. The graph of Figure 2 showsthe input bias currents of the LT1884over the common mode range of –14Vto 14V. This low stable bias currentbehavior, when coupled with 50µV
offset voltage, open-loop gains of overone million and high common moderejection, allows precision accuracyto be maintained in systems withdifficult source impedances.
Table 1 highlights key performancespecifications for these amplifiers.Each of these amplifiers provideshigher precision operation than waspreviously available in a rail-to-railoutput swing amplifier.
Selecting the Right AmplifierWhen choosing one of these amplifi-ers for an application, it is necessaryto consider the signal levels and sourceimpedance of the signal source. Lowimpedance, low level sources will usu-ally operate best with the LT1677amplifier. The ultralow 3.2nV/√Hznoise of the LT1677 will not obscurelow amplitude signals. High gain canbe used without introducing DCerrors, an important feature in lowsupply voltage applications. Othernatural applications for the LT1677occur when the input signal rangeextends to either power supply rail.The LT1677 maintains good DCaccuracy and noise performance withthe inputs at either power supply rail.As source impedance increases theLT1881 dual or LT1882 quad ampli-
1000
750
500
250
0
–250
–500
–750
–1000–15 –10 –5 0 5 10 15
COMMON MODE VOLTAGE (V)
INPU
T BI
AS C
URRE
NT (p
A)
TA = 25°C
IB–
IB+
Figure 1. Input (left) and output (right) of an LT1677 configured as a voltage follower withinput exceeding the supply voltage (VS = 3V, input = –0.5V to 3.5V)
3V
2V
1V
0V
–0.5V50µs/DIV
3V
2V
1V
0V
–0.5V50µs/DIV
Figure 2. LT1884 input bias currentvs common mode voltage
Linear Technology Magazine • May 2000 7
DESIGN FEATURES
fiers become the better choice. Theseamplifiers have an input noise cur-rent that is less than one tenth of theLT1677’s. The input bias currentsare as low as those of most FET inputdevices and they maintain their low IBat high temperatures where FET leak-age currents increase exponentially.The input offset voltage and tempera-ture drift are far superior to those ofJFET input amplifiers. The LT1881and LT1882 also operate at only 1mAsupply current per amplifier.
The LT1884 dual and LT1885 quadamplifiers have input bias and offsetcurrents almost as low as the LT1881and LT1882, but have approximatelythree times faster AC response. Theseamplifiers can be employed in thesame types of applications as theLT1881/LT1882, where AC response
has greater value and the cost in DCaccuracy is minimal. Supply currentis the same 1mA per amplifier.
Low NoiseRemote Geophone AmplifierSmall signal applications require highgain and low noise, a natural for theLT1677. Its 1kHz noise is 100% testedand is guaranteed to be less than4.5nV/√Hz. Figure 3 is a 2-wire remotegeophone preamp that operates on acurrent-loop principle and, as such,has good noise immunity. A low noiseamplifier is desired in this applica-tion because the seismic signals thatmust be resolved are extremely smalland require high gain. The LT1677amplifies the geophone signal by onehundred and transmits it back to the
operator by modulating the currentthrough R12. U2 is an LT1635 micro-power rail-to-rail op amp andreference configured as a stable cur-rent source of 5mA, which powers theLT1677 and another LT1635, thistime configured as a 3V shunt regula-tor. The idling current through R10 isset up from the voltage at the emitterof Q2 (3V) and the voltage at theemitter of Q3 (1.85V from the ratio ofR6 and R7). This places about 1.15V(zero TC, since Q1 temperature com-pensates Q2) across R10, therebypulling an additional 7mA from themain supply through Q2. Of the total12mA across the receiver resistor R12,7mA is being modulated by allowing apeak signal of ±1.5V about the 3Vbias point across R12.
Table 1. Key performance specifications
retemaraP 7761TL 1881TL 2881TL 4881TL 5881TL
noitarugifnoC elgniS lauD dauQ lauD dauQ
)xaM(egatloVtesffO Vµ06 Vµ08)edargA(Vµ05
Vµ08 Vµ08)edargA(Vµ05
Vµ08
tnerruCsaiBtupnI)xaM(
An02 Ap005)edargA(Ap002
Ap005 Ap009)edargA(Ap004
Ap009
tnerruCtesffOtupnI)xaM(
An51 Ap005)edargA(Ap002
Ap005 Ap009)edargA(Ap003
Ap009
nommoCtupnIegnaRedoM
V EE otV7.1+V CC V1–
V EE VotV1+ CC V1– V EE VotV1+ CC V1– V EE VotV1+ CC V1– V EE VotV1+ CC V1–
)noisicerPdecudeR( V EE otV1.0–V CC V1.0+
gniwStuptuOIL Aµ001=
V EE V071.0+V CC V071.0–
V EE V60.0+V CC V032.0–
V EE V60.0+V CC V032.0–
V EE V60.0+V CC V032.0–
V EE V60.0+V CC V032.0–
esioNegatloVtupnI)pyT(
zH√/Vn2.3 zH√/Vn41 zH√/Vn41 zH√/Vn5.9 zH√/Vn5.9
esioNtnerruCtupnI)pyT(
zH√/Ap3.0 zH√/Ap30.0 zH√/Ap30.0 zH√/Ap50.0 zH√/Ap50.0
egatloVylppuSegnaR
V04otV7.2 V63otV7.2 V63otV7.2 V63otV7.2 V63otV7.2
reptnerruCylppuS)xaM(reifilpmA
Am5.3 Am9.0 Am9.0 Am9.0 Am9.0
htdiwdnaBniaG)pyT(tcudorP
zHM2.7 zHM1 zHM1 zHM2 zHM2
)pyT(etaRwelS sµ/V7.1 ,sµ/V51.0sµ/V11.0–
,sµ/V51.0sµ/V11.0–
,sµ/V5.0sµ/V4.0–
,sµ/V5.0sµ/V4.0–
,niaGpooLnepORL )pyT(k01=
Vµ/V25 Vµ/V1 Vµ/V1 Vµ/V1 Vµ/V1
C DAOL , AV 1+=)xaM(
Fp0001 Fp0001 Fp0001 Fp003 Fp003
Linear Technology Magazine • May 20008
DESIGN FEATURES
Buffered PrecisionVoltage ReferenceFigures 4a and 4b, respectively, showthe gain (VOUT/VIN) and gain linearityfor the LT1677 sinking or sourcingcurrent into a 600Ω load with a single5V supply. A horizontal trace indicateshigh gain; a straight trace indicatesconstant gain vs output voltage—thisis excellent gain linearity. The tracefor the ground-referenced load is morehorizontal; this indicates higher loop
gain, due to the additional gain of thePNP output stage. Gain and gain lin-earity are improved by increasing theload resistor. Figure 4c shows thegain for a higher load resistance at a±15V supply (note the change ofvertical scale).
When teamed up with a precisionshunt voltage reference such as theLT1634 (Figure 5), the LT1677, used
as a precision buffer, can enhancethe reference voltage without signifi-cantly increasing the error budget.The LT1677 is used to make a 2.5Vvoltage source from a single 5V supply.The tolerance of the LT1634BCS8-2.5 is ±1.25mV (0.05%); the LT1677adds only a ±60µV offset voltage. Theoutput impedance of the voltagesource for a wide range of sourcing or
–
+
–
+
–
+
U2LT1635
U1LT1635
U3LT1677
2
3
81
4
67
2
3
81
4
6
2
3
67
7
R1 1k
R21k
R343.2Ω
C20.01µF
R4 14k
R51k
Q12N3904
C547µF
R6200k
R7324k
C10.13µF
R8866Ω
GEOSPACEGEOPHONEMODEL 20DX630Ω, 10Hz(713) 939-7093
R9150k
C30.01µF
Q32N3906
Q22N3904
C42200pFR10
165Ω
R1120Ω
D1–D41N4148
×4
12V
R12249Ω
VOUT
5mA
7mA
CURRENTSOURCE
3V SHUNT REGULATOR
REMOTE 3V SUPLY
IMPOSSIBLETO MISWIRELOCAL END
4
+
0 1 2 3 4 5
OUTPUT VOLTAGE (V)
INPUTVOLTAGECHANGE
10µV/DIV
RL = 600Ω TO 0VVS = 5V, 0VTA = 25°C
INPUTVOLTAGECHANGE
10µV/DIV
0 1 2 3 4 5
OUTPUT VOLTAGE (V)
RL = 600Ω TO 5VVS = 5V, 0VTA = 25°C
Figure 3. Geophone amplifier
Figure 4a. Gain linearity of the LT1677 sinking current Figure 4b. Gain linearity of the LT1677 sourcing current
Linear Technology Magazine • May 2000 9
DESIGN FEATURES
sinking currents can be calculated bydividing the LT1677’s 80Ω open-loopresistance by its loop gain. This resultsin an output impedance of less than1mΩ. The dynamic impedance at10kHz drops from 20Ω (LT1634) toless than 0.01Ω (LT1677). The changein output voltage due to die tempera-ture is reduced thirty times by shiftingthe load current to the LT1677. Thetemperature coefficient of the LT1677,2µV/°C max, is negligible comparedto the 62.5µV/°C TC (25ppm/°C •2.5V) of the LT1634.
High-Side Current SensingFigure 6 is a precision high-side cur-rent sense amplifier that can operateon a single supply from 3V to 40V.The current flowing into the load pro-duces a voltage drop across the lineresistor RLINE. The LT1677 forces acurrent through RIN, which dupli-
cates the voltage drop across RLINE.This current is then converted backto a voltage across ROUT. By selectingappropriate values for these threeresistors, the transfer function canbe tailored to fit any application. Sincethe LT1677 can operate from eitherrail, a low-side current sense circuitcan also be realized.
Low Input Bias CurrentsFit Other ApplicationsThe applications described above ben-efit from the LT1677’s low input noiseand rail-to-rail inputs, but otherapplications require high DC accuracywith low input bias currents. TheLT1881/LT1882/LT1884/LT1885provide the appropriate answer inthese applications. Circuits that havehigh source impedances make theinput bias current and input offsetcurrent characteristics of the ampli-
fier important considerations. Theamplifiers’ input currents, acting onthe source impedance, generate a DCoffset error, limiting precision sens-ing of the source signal. The inputvoltage noise of the amplifier becomesa less important parameter becausethe noise generated by the high sourceimpedance will typically be larger thanthe op amp’s input noise. Input cur-rent noise is now the more importantamplifier noise characteristic. TheLT1881/LT1882/LT1884/LT1885have very low input noise current, asshown in Table 1. Figure 7 graphs thetotal system noise due to amplifierinput noise voltage, input noise cur-rent and the source resistanceJohnson noise. The graph shows thatthe LT1677 is the correct amplifierwhen source resistance is below 20k.Above 200k an LT1881, LT1882,LT1884 or LT1885 are the best choicesfor minimizing system noise. In the
0 1 2 3 4 5
OUTPUT VOLTAGE (V)
INPUTVOLTAGECHANGE
1µV/DIV
RL 10k TO 0VVS = ±15VTA = 25°C
–
+2.5VLT1677
5V
R1249k
LT1634-BCMS8-2.5
2
3
4
7
6
2 7
BC856B*
VOUT4
6
3
RIN1k
SOURCE
RLINE0.1Ω
ROUT20kLOAD VOUT
ILOAD= RLINE
= 2V/AMP
ROUTRIN
–
+LT1677
*ZETEX (516) 543-7100
100 1k 10k 100k 1M 10M 100M
10k
1k
100
10
1
SOURCE RESISTANCE (Ω)
TOTA
L 1k
Hz R
MS
VOLT
AGE
NOIS
E DE
NSIT
Y (n
V/√H
z)
–
+
RIN–
RIN+
RS = RIN– + RIN
+
RESISTORNOISE
LT1884
LT1881
LT1677
VN = √(VOP AMP)2 + 4KT • RS + 2q • IS • RS2
LT1677 3.2nV/√Hz 0.30pA/√HzLT1884 9.5nV/√Hz 0.05pA/√HzLT1881 14.0nV/√Hz 0.03pA/√Hz
Figure 4c. LT1677 gain linearity with a higher load resistance and supply voltage
Figure 5. 2.5V reference from a 5V supply
Figure 6. Precision high-side current sense amplifierFigure 7. 1kHz noise voltage vssource resistance
Linear Technology Magazine • May 200010
DESIGN FEATURES
intermediate range of 20k to 200k,the source resistor’s noise dominatesand all of the amplifiers will give nearlyequal system noise performance.
Input-Fault-ProtectedInstrumentation AmplifierFigure 8 is a fault-protected instru-mentation amplifier with shield drive.The 1M input resistors allow highvoltage faults to be tolerated withoutdamaging the amplifiers. An AC linefault will result in only 180µA of peakcurrent flowing into the LT1882’sinput pins. Normally, such a highinput resistance would result in hugeDC offsets due to the input bias cur-rents from the amplifiers. TheLT1882’s IB is a low 500pA max. Inaddition, by using the guaranteedmatching of IB between amplifiers Aand B or C and D, a worst-case IBmismatch of 700pA is guaranteed.This results in a worst-case offseterror of 700µV; typically this errorwould be less than 200µV. A pair ofLT1881 “A” grade devices may be usedto ensure a worst-case error less than300µV.
The LT1882’s input noise currentworking against the 1M sourceresistance generates a noise voltageof 42nV/√Hz. This compares to theprotection resistors’ Johnson noise of182nV/√Hz and the amplifier’s inputnoise voltage of 14nV/√Hz. Hence,the 1M protection resistors dominatethe system’s input noise. It is inter-esting to note that the LT1677 wouldcontribute 420nV/√Hz of input noise
due to its input noise current, domi-nating the system noise level. In highsource impedance applications, theLT1677 low noise amplifier will gen-erate more system noise than anLT1882, which has a higher inputvoltage noise level. This underscoresthe need for proper amplifier selec-tion based upon the application. Inhigh source impedance applications,input current noise is a more impor-tant parameter than input voltagenoise.
Low Voltage –50°C to 600°CDigital ThermometerThe circuit of Figure 9 is a digitalthermometer which uses a 1k RTDsense element in a linear-output,single-leg bridge configuration. AnLT1881 dual amplifier is used to pro-vide the negative bridge excitation aswell as output amplification and buff-ering. An LTC1287 A/D converter
digitizes the output. No reference isneeded; the bridge excitation and A/Dreference are the power supply. Thefixed bridge elements and output gainresistor are made with series andparallel combinations of an 8 × 2kresistor pack to get the best precisionat a reasonable price.
The LT1884’s low offset voltageand rail-to-rail output swing enablethe circuit to function properly. Thefirst amplifier, A1, is used to drive thenegative side of the bridge to force aconstant current excitation of thevariable resistance element. The con-stant current drive results in theoutput voltage being perfectly linearwith respect to the variable resis-tance. Since the LT1884 is able toswing to within to 50mV of the nega-tive supply, the full dynamic range ofthe transducer is available even whenoperating on low voltage supplies.The second amplifier provides voltagegain to the bridge output to use thefull-scale range of the A/D converter.The LT1884’s low offset voltage is animportant attribute of the gainamplifier.
±4.096V Swing 16-Bit VoltageOutput DAC on a ±5V SupplyThe final application, Figure 10, showsan LT1881 dual amplifier used as anI/V converter with the 16-bit LTC1597DAC. The first amplifier is used toinvert and buffer the LT1634 refer-ence. This amplifier has no troubleswinging to –4.096V even with lowsupply voltages. The LT1881’s excep-tionally low IB and offset voltage
RG/2
RG/2
20pFTRIM FORAC CMRR
10k
OUT
–IN
SHIELD
+IN
10k
1M
1M 3
5
10k
10k
10k
10k
GAIN = 2•10k
RG
–
+
–
+
–
+
–
+
1/4 LT1882
1/4 LT1882
1/4 LT1882
1/4 LT1882
10pF
–
+–
+
VREF+IN–IN
GND
VCC
CLK
DOUT
CS
5
2
3
4
8
7
6
1
LTC1287
10k0.1%
10k0.1%
VCC
A11/2 LT1881
A21/2 LT1881
VCC
RF 1kR2 4kR1 4k
R3 1kRT
1µF
VCC = 3.3V
V = ± 1.588mV/°CVCC
2
RT:R1–R3, RF:
OMEGA F4132 1kΩ RTD (203) 359-1660BI 698-3 2k × 8 RESISTOR NETWORK (714) 447-2345
Figure 8. Input-fault-protected instrumentation amplifier
Figure 9. –50°C to 600°C digital thermometer runs on 3.3V
continued on page 16
Linear Technology Magazine • May 2000 11
DESIGN FEATURES
DC Accurate, Rail-to-Rail Active RCLowpass Filter Replaces DiscreteDesigns—One Resistor ValueSets the Cutoff Frequency by Doug La Porte
No More Complex Equations,No More Precision Capacitors,No More Frustration The LTC1563-2 and LTC1563-3 forma family of extremely easy to use, 4thorder active RC lowpass filters (nosignal sampling or clock require-ments). They cover cutoff frequenciesranging from 256Hz to 256kHz whileoperating at supplies from as low as asingle 3V (2.7V minimum) up to ±5V.The LTC1563 also features rail-to-rail input and output operation with,typically, 1mV of DC offset. The designof the most popular filter responses istrivial, requiring only six resistors ofidentical value and no externalcapacitors. The cutoff frequency ofunity-gain Butterworth or Besselfilters is set with a single resistorvalue calculated using the followingsimple formula:
R = 10kΩ • (256kHz/fC)where fC is the cutoff frequency inHertz.
The LTC1563-2 is used for a But-terworth response, whereas theLTC1563-3 is used for a Besselresponse.
Beyond design simplicity, theLTC1563 facilitates manufacturability.For a discrete design to achieve the±2% cutoff frequency accuracy of theLTC1563, precision capacitors (2% orbetter) are required. These capacitorsare not readily available and canpresent a difficult and costly pur-chasing problem. A typical discretedesign also requires four resistor val-ues and four capacitor values—eightreels of components. This leads toeight times the purchasing, stockingand assembly costs and eight reels ofcomponents on the automatedassembly machine. Many large cir-cuit boards require more componentreels than the assembly machine canaccommodate leading to costly sec-ondary operations. The LTC1563decreases purchasing, stocking andassembly costs while removing sevencomponent reels from the assembler.
The LTC1563 is also very versatile.The proprietary architecture yieldseffortless design of the unity gainButterworth and Bessel filter respon-ses while still allowing complex,arbitrary filter responses with anygain desired. A Chebyshev, Gaussianor any other all-pole response, with or
without gain, can be obtained usingunequal valued resistors calculatedwith a more complex set of equations(for the best results use FilterCADversion 3.0). Designs ranging from adual 2nd order filter up to an 8thorder filter (two cascaded devices) arealso easily obtained. With the addi-tion of two capacitors, a singleLTC1563 can be used to implement a6th order lowpass filter. In anotherapplication, the two additionalcapacitors render a simple wideband,low Q bandpass filter.
Salient performance features of theLTC1563 family include the following:
256Hz ≤ fC ≤ 256kHz fC accuracy < ±2% (typ) Continuous time filter—no clock,
no sampling SINAD ≥ 85dB at 3VP-P, 50kHz—
compatible with 16-bit systems Rail-to-rail input and output
operation Output DC offset voltage ≤ ±1mV
(typ) DC offset drift ≤ ±5µV/°C (typ) Operates from a single 3V (2.7V
min) to ±5V supplies
1
2
3
4
5
6
7
8
16
15
14
13
12
11
10
9
LP
SA
NC
INVA
NC
LPA
AGND
V–
V+
LPB
NC
INVB
NC
SB
NC
EN
VOUT
V+
ENABLE
VIN
R R
R
R
R R
0.1µF
LTC1563-X 0.1µF
LP HS
FREQUENCY (Hz)10k1k
GAIN
(dB)
10
0
–10
–20
–30
–40
–50
–60
–70
–80100k 1M
R = 10MfC = 256Hz
R = 10kfC = 256kHz
100
10
0
–10
–20
–30
–40
–50
–60
–70
–801k 10k 100k 1M
FREQUENCY (Hz)
GAIN
(dB)
R = 10kfC = 256kHz
R = 10MfC = 256Hz
100
Figure 2. LTC1563-2 Butterworth response ofFigure 1’s circuit for R = 10M and R = 10k
Figure 3. LTC1563-3 Bessel response ofFigure 1’s circuit for R = 10M and R = 10k
Figure 1. Typical LTC1563-X single-supplyapplication
Linear Technology Magazine • May 200012
DESIGN FEATURES
Low power mode, ISUPPLY = 1mA(typ), fC ≤ 25.6kHz
High speed mode, ISUPPLY = 10mA(typ), fC ≤ 256kHz
Shutdown mode, ISUPPLY = 1µA(typ)
Narrow SSOP-16 package, SO-8footprint
Typical 4th OrderButterworth and BesselApplicationsFigure 1 illustrates a typical LTC1563single-supply application, with Fig-ure 2 showing the Butterworthfrequency response of the LTC1563-2and Figure 3 showing the Bessel fre-quency response of the LTC1563-3.As the R value is decreased from 10Mto 10k, the cutoff frequency increasesfrom 256Hz to 256kHz. For cutofffrequencies below 25.6kHz, signifi-cant power can be saved by placingthe LTC1563 into the low power mode.Connect the LP pin to the V– potentialto enable the low power mode. Allother applications should place thepart in the high speed mode by leav-ing the LP pin open or connecting it tothe V+ potential. The high speed mode,in addition to supporting higher cut-off frequencies, has a lower DC offsetvoltage and better output drive capa-bility than the low power mode. Theminimum supply voltage in the highspeed mode is 3V, whereas the lowpower mode supports 2.7V opera-tion. The shutdown mode is availableat all times. The EN pin is internallypulled up to the V+ potential, causingthe LTC1563 to default to the shut-down mode. To enable normal
operation, the EN pin must be pulledto ground. In the shutdown mode, thesupply current is typically 1µA and20µA maximum over temperature.
Performance SpecificationsThe LTC1563 family has outstandingDC specifications. The DC offset ofthe filter in the high speed mode istypically ±1mV, with a maximum off-set over temperature of ±3mV. TheDC offset is slightly greater in the lowpower mode at ±5mV maximum overtemperature on the lower supply volt-ages and ±6mV maximum overtemperature on a ±5V supply. The DCoffset drift is only 10µV/°C.
The LTC1563’s SINAD performancewith lower signals is dominated bythe noise of the part. Figure 4 is a plotof the noise vs the cutoff frequency.
The total integrated noise (over a band-width of twice the cutoff frequency) is32µVRMS at the lowest cutoff frequencyand increases to 56µVRMS at the high-est cutoff frequency. Figure 5 showsthe SINAD performance as functionof the input signal amplitude for a50kHz signal. The plot demonstratesthat distortion is not a significantfactor at smaller signal amplitudes.Distortion is only noticeable whenthe signal amplitude is very large,within about 1V of the supply rails.The distortion performance also holdsup well at higher frequencies, as Fig-ure 6 illustrates. The SINAD is nearlyflat over all frequencies, indicatingagain that noise is the determiningfunction. The SINAD is about 85dBfor the 3VP-P input, indicating thatthis part is suitable for 16-bit systems.
60
50
40
30
20
10
01k 10k 100k 1M
fC (Hz)
TOTA
L IN
TEGR
ATED
NOI
SE (µ
V RM
S)TA = 25°C
100
–40
–50
–60
–70
–80
–90
–1000.1 1.0 10
INPUT VOLTAGE (VP-P)
(THD
+ N
OISE
)/SIG
NAL
(dB)
UNITY GAINHS MODEfC = 256kHzfIN = 50kHz
3.3V SUPPLY
±5V SUPPLY
5V SUPPLY
–40
–50
–60
–70
–80
–90
–1001 10 100 200
FREQUENCY (kHz)
(THD
+ N
OISE
)/SIG
NAL
(dB)
VIN = 1VP-PVIN = 2VP-P
UNITY GAINHS MODEfC = 256kHz5V SUPPLY
VIN = 3VP-P
ngiseD troppuS0.3DACFtroppuS
FPLredrodn2 seY
FPLredrodn2lauD seY
)roticapaceno(FPLredrodr3 seY seY
seY
seY
)sroticapacowt(FPLredrodr3lauD seY seY
FPLredroht4 seY seY
)roticapaceno(FPLredroht5 seY seY
)sroticapacowt,trapeno(FPLredroht6–oduesP seY seY
)sroticapacon,strapowt(FPLredroht6dradnatS seY seY
)roticapaceno(FPLredroht7 seY seY
FPLredroht8 seY seY
)roticapaceno(FPLredroht9 seY seY
)sroticapacowt(ssapdnabdnabediW seY oN
Table 1. LTC1563 family configurations
LTC1563
Figure 4. LTC1563-X total integrated noisevs cutoff frequency
Figure 5. LTC1563-X THD plus noiseperformance vs input signal amplitude
Figure 6. LTC1563-X THD plus noise vsinput signal frequency
Linear Technology Magazine • May 2000 13
DESIGN FEATURES
Transfer FunctionCapabilitiesThe LTC1563 was designed to makeimplementing standard unity-gain,4th order Butterworth and Bessel fil-ters as simple as possible. Althoughthis goal was accomplished, the partalso maintains tremendous flexibility.Virtually any all-pole transfer functioncan be realized with the LTC1563-2.Table 1 lists the many filter configu-rations attainable with the LTC1563family. The actual transfer function(pole locations) of the filter is nearlyarbitrary.
4th OrderLowpass Filters with GainFilters responses other than Butter-worth and Bessel, filters with gainand higher order filters are easilyobtained with the LTC1563. Thedesign equations are more complexthan the simple unity-gain Butter-worth and Bessel equation. Refer tothe LTC1563 data sheet for the actualequations. For the best design result,use FilterCAD version 3.0 (or later).FilterCAD uses a very complex andaccurate algorithm to account for mostparasitics and op amp limitations.Using FilterCAD will yield the bestpossible design.
Figure 7 shows an LTC1563-2employed to make a 4th order,150kHz, 0.5dB ripple Chebyshev low-pass filter with a DC gain of 10dB. Thefrequency response is shown in Fig-ure 8. All of the gain is realized in thefirst section to achieve the lowestoutput voltage noise. Note that theschematic, and thus the PC boardlayout, is the same as a unity gainButterworth lowpass filter except thatthe resistor values are different. Thesame PC board layout could be usedto build an infinite number of lowpassfilters, each with a different gain,t ransfer funct ion and cuto f ffrequency.
8th Order Lowpass FilterDesigning an 8th order filter is just aseasy as a 4th order filter. Of course,the final solution uses twice as manycomponents but the design is essen-tially the same procedure. UsingFilterCAD makes the design proce-dure simple and straightforward.Figure 9 shows the schematic for a180kHz 8th order, Butterworth low-pass filter with a gain of 6dB alongwith a plot of its frequency response.The solution is quite compact, simpleto layout on the PC board and usesonly standard 1% resistors. A dis-crete solution would involve severalprecision capacitors and a much morecomplicated layout accompanied bynumerous layout-related parasitics.With the discrete solution, you maynot get the response that you expect,due to these parasitic elements.
1
2
3
4
5
6
7
8
16
15
14
13
12
11
10
9
LP
SA
NC
INVA
NC
LPA
AGND
V–
V+
LPB
NC
INVB
NC
SB
NC
EN
VOUT
5V
VIN
R319.31k
R3211.8k
R22 20k
R2173.2k
R1123.2k
R1220k
1µF
LTC1563-2 0.1µF
–5V
20
10
0
–10
–20
–30
–40
–50
–60
–7010k 100k 1M
FREQUENCY (Hz)
GAIN
(dB)
1
2
3
4
5
6
7
8
16
15
14
13
12
11
10
9
LP
SA
NC
INVA
NC
LPA
AGND
V–
V+
LPB
NC
INVB
NC
SB
NC
EN
5V
VIN
R319.09k
R3220.5k
R22 8.66k
R2117.4k
R118.87k
R128.66k
0.1µF
LTC1563-2 0.1µF
1
2
3
4
5
6
7
8
16
15
14
13
12
11
10
9
LP
SA
NC
INVA
NC
LPA
AGND
V–
V+
LPB
NC
INVB
NC
SB
NC
EN
VOUTR337.87k
R3410.2k
R24 16.5k
R2321.5k
R1416.5k
0.1µF
LTC1563-2 0.1µF
R1321.5k
20
0
–20
–40
–60
–80
–100
–12010k 100k 1M
FREQUENCY (Hz)
GAIN
(dB)
Figure 7. 0.5dB, 150kHz, Chebyshevlowpass filter with a DC gain of 10dB
Figure 8. Frequency responseof Figure 7’s circuit
Figure 9a. 180kHz 8th order, Butterworth lowpass filter with a DC gain of 6dBFigure 9b. Frequency responseof Figure 9a’s circuit
Linear Technology Magazine • May 200014
DESIGN FEATURES
Pseudo–6th OrderLowpass FilterA textbook, theoretical 6th order low-pass filter is composed of three 2ndorder sections, each realizing a com-plex pole pair. The LTC1563 has two2nd order sections to yield two com-plex pole pairs. A 6th order lowpassfilter can be made using two LTC1563parts (with half of the second partunused) or, by using an old filtertrick, it can be implemented with asingle part.
First a little background material(a bit of filter theory, but not too bad).Each 2nd order, complex pole paircan be defined by the parameters fOand Q. Figure 10 illustrates that eachcomplex pole pair has the same realcomponent and the complementaryimaginary components. The fOparameter is the magnitude of thepoles and the Q is a measure of howclose the poles are to the real orimaginary axis. The closer the polesare to the imaginary axis, the higherthe Q becomes, until Q reaches infin-ity when the poles are directly on theimaginary axis. As the pole pair movesclose to the real axis, the Q decreasesin value until the Q is at 0.5 whenthey are both on the real axis. A polepair with a Q of 0.5 is mathematicallyidentical to two 1st order, real poles.
Armed with a little filter knowl-edge, we are now ready for the trick.All 6th order filters have three fO andQ pole pairs. The pair with the lowestQ usually has a Q somewhere between0.5 and 0.6. The trick is to substitutetwo 1st order, real poles (the equiva-lent of a 2nd order section with a Q of
0.5) for the pole pair with the lowest Qvalue. To compensate for this bend-ing of the ideal mathematics, it isnecessary to go back and tweak someor all of the fO and Q values in thedesign. The resulting filter is no longerthe exact, ideal, theoretical math-ematical implementation, but withcare you can get a frequency responseand a step response that differ imper-ceptibly from the textbook plots. Thename “pseudo–6th order filter” issomewhat misleading, because thefilter is clearly of the 6th order (sixpoles with a final attenuation slope of–36dB/octave). It is only “pseudo” inthe sense that the filter does notconform to the classical, standardmathematical models. If your com-pany is shipping mathematics, stickwith the textbook 6th order response.However, realize that once you con-sider the component tolerances, you
still end up with an approximation.Most products really require filtersthat meet a specific set of perfor-mance criteria (for example, cutofffrequency, passband flatness, stop-band attenuation, step responseovershoot and step response settling)in a simple, reproducible and costeffective manner. The textbookresponses are just a convenient wayto synthesize a filter (or a good start-ing point in the filter design).
The classical response is tweakedmanually using the FilterCAD Cus-tom Design feature. Start with thetextbook 6th order lowpass filter. Openthe Step Response and FrequencyResponse windows by clicking on theappropriate buttons. In each responsewindow, save the trace to comparethe responses with the tweakeddesign. Next, click on the CustomResponse button, remove the low Q
IMAGINARY
REAL
CIRCLE OFCONSTANT fO
Q = 0.5
Q = ∞Q = 0.7
X
X2
X
X
X
Im
Re
fC = 100kHz
XX
X
XX
X
CLASICAL BUTTERWORTH
fO100kHz100kHz100kHz
Q0.51760.70711.9319
Im
Re
fC = 100kHz
XX
XX
PSEUDO-BUTTERWORTH
fO100kHz100kHz100kHz100kHz
Q————
0.73581.9319
X2
1
2
3
4
5
6
7
8
16
15
14
13
12
11
10
9
LP
SA
NC
INVA
NC
LPA
AGND
V–
V+
LPB
NC
INVB
NC
SB
NC
EN
VOUT
3.3V
VIN
R3117.8k
R3220.5k
R22 28.7k
R2132.4k
0.1µF
LTC1563-2 0.1µF
C12560pF
C11560pF
RA2 3.16k RB2 25.5k
RA1 3.16k RB1 29.4k
Figure 10. Locations of complex pole pairs offixed fO and different Qs
Figure 11. Pole locations for the classic Butterworth and pseudo-Butterworth lowpass filters
Figure 12. 100kHz, 6th order pseudo-Butterworth lowpass filter
Linear Technology Magazine • May 2000 15
DESIGN FEATURES
section and add two 1st order low-pass sections at the same fO. Finally,alter the new design’s fO and Qparameters as required, while moni-toring the frequency and step responsewindows, until the desired responsesare achieved. With some practice, anintuitive sense of which parametersneed adjustment is developed.
The procedure is clearly illustratedwith the following 6th order 100kHzButterworth lowpass filter example.Figure 11 shows the pole locationsand fO and Q values for a textbook 6thorder Butterworth lowpass filter andthe pseudo–6th order equivalent val-ues. The circuit implementation,Figure 12, of this transfer functionuses a standard 4th order type ofcircuit where the input resistor issplit into two resistors with an addi-tional capacitor to ground in betweenthe resistors. This “TEE” networkforms a single real pole. By making
this modification to both sections, apair of real poles is added to the 4thorder network forming the desiredpseudo–6th order filter. The TEE net-work technique is also used to givethe single real pole required in allodd-order responses.
Wideband Bandpass FiltersAlthough the LTC1563 family doesnot directly support classical band-pass filters, you can successfullyimplement a wideband bandpass fil-ter using the standard 4th orderlowpass circuit and a couple of addi-tional capacitors. The resulting filteris more of a highpass-lowpass type offilter than a true bandpass. You candesign outstanding, highly selectivebandpass filters using the LTC1562family of universal filter products (seeLinear Technology VIII:1 [February1998] and IX:1 [February 1999])—thebest continuous time filters in the
industry for bandpass and elliptichighpass or lowpass filters. Althoughthe LTC1562 is the best part for nar-rowband bandpass filters, if yourrequirements are less stringent, theLTC1563 family can provide a simple,cost-effective wideband bandpassfilter.
Figure 13 shows the schematic ofan LTC1563-3 used to make a simple,wideband bandpass filter centered at50kHz. The frequency response isshown in Figure 14. The design pro-cedure for this type of filter does notconform to any standard procedure.It is best to start with a standardlowpass filter and add two 1st orderhighpass sections to the transfer func-tion. Adjust the highpass corner andthe lowpass fC up and down until thedesired transfer function is achieved.The design in Figure 13 starts with astandard, unity gain, 4th order Bes-
1
2
3
4
5
6
7
8
16
15
14
13
12
11
10
9
LP
SA
NC
INVA
NC
LPA
AGND
V–
V+
LPB
NC
INVB
NC
SB
NC
EN
VOUT
5V
VIN
R3120k
R3220k
R22 20k
R2120k
R1120k
R1220k
0.1µF
LTC1563-3 0.1µF
680pF
680pF
–5V
10
0
–10
–20
–30
–40
–50
–601k 10k 100k 1M
FREQUENCY (Hz)
GAIN
(dB)
1
2
3
4
5
6
7
8
16
15
14
13
12
11
10
9
LP
SA
NC
INVA
NC
LPA
AGND
V–
V+
LPB
NC
INVB
NC
SB
NC
EN
5V
VIN
5V
R3182.5k
R3278.7k
R22 137k
R21243k
0.1µF
LTC1563-2 0.1µF
C11560pF
R12 137k
RA1 26.7k RB1 215k
++
+
+
+
1
2
3
4
5
6
7
8
9
10
34
AIN+
AIN–
VREF
REFCOMP
AGND
AGND
AGND
AGND
DVDD
DGND
VSS
35
36
33
32
31
30
27
11 TO 26
29
28
AVDD
AVDD
SHDN
CS
CONVST
RD
BUSY
OVDD
OGND
LTC1604
560pF
2.2µF
47µF
49.9Ω
10µF
10µF
10µF
10µF
5V OR 3V
16-BITPARALLEL BUS
µPCONTROLLINES
5V10Ω
–5V
–5V
+
10µF
Figure 13. 50kHz wideband bandpass filter Figure 14. Frequency responseof Figure 13’s circuit
Figure 15. 0.1dB, 5th order, 22kHz Chebyshev lowpass filter driving an LTC1604 16-bit ADC
Linear Technology Magazine • May 200016
DESIGN FEATURES0
–20
–40
–60
–80
–100
–120
–1400 36.58 73.15 109.73 146.30
FREQUENCY (kHz)
AMPL
ITUD
E (d
B)
fSAMPLE = 292.6kHzfIN = 20kHzSINAD = 85dBTHD = –91.5dB
sel lowpass filter with a cutoff fre-quency of 128kHz. Each section isthen AC coupled through the 680pFcapacitors. Each capacitor, workingagainst its input resistor, realizes a1st order highpass function with thecutoff frequency (11.7kHz in thisdesign) defined by the followingequation:
fC = 1/(2 • π • R • C) (R = R11 or R12)The resulting circuit yields a
wideband bandpass filter that usesonly one resistor value and onecapacitor value. Note that althoughFilterCAD does not provide completesupport for this type of filter with theLTC1563, you can still use theprogram’s Custom Design mode toset the required transfer function.After the transfer function has beenchosen, use FilterCAD to design thelowpass part of the filter and then usethe simple highpass formula above tocomplete the circuit.
Driving 16-Bit ADCsThe LTC1563 is suitable for 16-bitsystems. Figures 3 through 5 showthat the LTC1563’s noise and SINADis commensurate with 16-bit dataacquisition systems. These measure-ments were taken using laboratoryequipment and well-behaved loadingcircuitry. Many circuits will perform
well in this environment only to fallapart when driving the actual A/Dconverter. Many modern ADCs have aswitched capacitor input stage thatoften proves to be difficult to drivewhile still maintaining the converter’s16-bit performance. The LTC1563succeeds with only a little help from aresistor and a capacitor.
Figure 15 shows the LTC1563-2configured as a 5th order, 22kHz,0.1dB ripple Chebyshev lowpass fil-ter driving an LTC1604 16-bit ADC.The converter operates with a292.6kHz sample clock. The 22kHzChebyshev filter has attenuation ofabout 96dB at the Nyquist frequency.This is a very conservative antialias-ing filter that guarantees aliasing willnot occur even with strong input sig-nals beyond the Nyquist frequency.
Figure 16 shows a 4096 point FFTof the converter’s output while beingdriven by the LTC1563 at 20kHz. TheFFT plot shows the fundamental20kHz signal and the presence ofsome harmonics. The THD is –91.5dBand is dominated by the second har-monic at –92dB. The remainingharmonics are all well below the–100dB level. There are also somenonharmonically related spurs thatare artifacts of the ADC. In the filter’spassband, the noise level is slightlyhigher than the converter’s. The peak-ing of the noise level at the cutofffrequency, mostly due to the high Qsection, is a typical active filter char-acteristic. In the stopband, the noiselevel is essentially that of the con-verter. The end result is a SINADFigure of 85dB, about 4dB less thanthe converter alone.
ConclusionThe LTC1563 continuous time, activeRC lowpass filter is an economical,simple to use, yet versatile part thatmeets the high performance stan-dards required in today’s highresolution systems. Beyond ease ofuse, the part’s design simplicity leadsto ease of manufacture. This combi-nation of features makes discretelowpass filters and expensive, bulkyfilter modules obsolete.
Figure 16. FFT of the ADC output datafrom Figure 15’s circuit
–5V
5V
VOUT–4.096VTO 4.096V
LTC1597
DAC
–5V
R2
RCOM ROFS
ROFS
REF
R1
R1
5V1.6k
LT16344.096V
5V
–
+
–
+1/2 LT1881
1/2 LT1881
3 2 1 23 4
5
6
7
VCCRFB
RFB
IOUT1
AGND
33pF
minimize the introduction of errorsdue to the amplifiers. This is espe-cially important when operating theDAC at low supply voltages. An LSBof DAC output current is only 3nA.The low IB of the LT1881 keeps thiserror to less than 0.2LSB of zero-scale offset.
ConclusionLinear Technology’s new rail-to-railoutput precision operational amplifi-ers provide the proper amplifier forany low voltage, high precision appli-cation. The user must understandthe requirements of the application,particularly the source impedance,to make the proper choice as towhether the LT1677, LT1881/LT1882or LT1884/LT1885 is the best ampli-fier for a given application.Figure 10. 16-bit voltage-output DAC on a ±5V supply
LT1677/LT188X, continued from page 10
Linear Technology Magazine • May 2000 17
DESIGN FEATURES
SOT-23 Switching Regulator withIntegrated 1 Amp Switch DeliversHigh Current Outputs by Albert Wu
The LT1930 is the only SOT-23switching regulator in the industrythat includes an integrated 1A switch.The LT1930 utilizes a constant fre-quency, internally compensated,current mode PWM architecture. Its1.2MHz switching frequency allowsthe use of tiny, low cost capacitorsand low profile inductors. With aninput voltage range of 2.6V to 16V,the LT1930 is a good fit for a variety ofapplications. The onboard switch fea-tures a low VCESAT voltage of 400mV at1A, resulting in very good efficiencyeven at high load currents.
Figure 1 shows a typical 3.3V to 5Vboost converter using the LT1930.
The circuit can provide an impressiveoutput current of 480mA. The effi-ciency remains above 83% over awide load current range of 60mA to450mA, reaching 86% at 200mA. Themaximum output voltage ripple ofthis circuit is 40mVP-P, which corre-sponds to less than 1% of the nominal5V output. Figure 2 is an oscilloscopephotograph of the transient response.The lower waveform represents a loadstep from 200mA to 300mA, themiddle waveform shows the inductorcurrent and the upper waveformshows the output voltage. The outputvoltage remains within 1% of the nomi-nal value during the transient steps
GND
SHDN FB
VIN SWLT1930
5 1
3
2
4
L1 5.6µH D1
R140.2k
R213.3k
C210µFSHDN
C14.7µF
VIN3.3V
VOUT5V/480mA
C1: TAIYO-YUDEN X5R JMK212BJ475MG (408) 573-4150C2: TAIYO-YUDEN X5R JMK316BJ106MLD1: ON SEMICONDUCTOR MBR0520 (800) 282-9855L1: SUMIDA CR43-5R6 (847) 956-0666
++
and displays a well damped responsewith little ringing.
Another typical application is a 5Vto 12V boost converter, as shown inFigure 3. This circuit can provide300mA of output current with effi-ciencies as high as 87%. Themaximum output voltage ripple ofthis circuit is 60mVP-P, which corre-sponds to 0.05% of the nominal 12Voutput. Figure 4 is an oscilloscopephotograph of the transient response.The lower waveform shows a loadcurrent step from 200mA to 250mA.The middle waveform displays theinductor current and the upper wave-form shows the output voltage. The
Figure 1a. 3.3V to 5V/450mA step-up DC/DC converter
Figure 2. Transient response of Figure 1a’s circuit
Figure 1b. Efficiency of Figure 1a’s circuit
continued on page 20
90
85
80
75
70
65
60
55
500 100 200 300 400
LOAD CURRENT(mA)
EFFI
CIEN
CY (%
)
50050 150 250 350 450
20µs/DIV
VOUT0.1V/DIV
AC COUPLED
IL10.5A/DIV
AC COUPLED
300mA
200mALOAD CURRENT
Linear Technology Magazine • May 200018
DESIGN FEATURES
Dual Operational Amplifier Combines16-Bit Precision with High Speed
by Kris LokereIntroductionThe LT1469 is a dual operationalamplifier that has been optimized foraccuracy and speed in 16-bit sys-tems. The amplifier settles in just900ns to 150µV for a 10V step. TheLT1469 also features the excellentDC specifications required for 16-bitdesigns. Input offset voltage is 125µVmaximum, input bias current is 10nAmaximum for the inverting input andminimum DC gain is 300V/mV. TheLT1469 specifications are summa-rized in Table 1.
This article presents two applica-tions of the LT1469 in 16-bitdata-conversion systems. The firstapplication is with a fast current-output digital-to-analog converter(DAC), such as the LTC1597. Thedual LT1469 amplifier allows thisDAC to operate in bipolar, 4-quad-rant multiplying mode. The secondapplication illustrates the use of thisdual amplifier as a buffer for a differ-ential analog-to-digital converter(ADC), such as the 333ksps LTC1604.
16-Bit 4-Quadrant DACwith 2.4µs Settling TimeThe fastest, most precise way toachieve 16-bit digital-to-analog con-version is using a current-output DACfollowed by a precision amplifier forcurrent-to-voltage conversion. Figure1 shows the LT1469 used in con-junction with the LTC1597 16-bitcurrent-output DAC. The first ampli-fier is used as the current-to-voltage(I/V) converter at the output of theDAC. The second amplifier is used toinvert the reference input voltage. Allthe resistors are internal to the DACand precisely trimmed. With a fixed10V reference input (such as could beprovided by the LT1021-10), the ref-erence inversion allows a bipolaroutput swing, that is, from –10V to10V. In addition, because of the highbandwidth and low distortion of the
LT1469, this configuration allows thereference input to be a variable sig-nal, such as a sine wave, for full4-quadrant multiplication operation.Figure 2 shows signal-to-(noise plusdistortion) measurement results ofthis circuit.
The key AC specification of thecircuit in Figure 1 is settling time,since this limits the DAC update rate.In an optimum configuration, the set-tling time of the LT1469 alone is ablistering 900ns. In Figure 1, settlingtime is limited by the need to compen-sate for the DAC output capacitance,which, for the LTC1597, varies from70pF to 115pF, depending on theinput code. This capacitance at theamplifier’s inverting input combineswith the internal feedback resistor toform a zero in the closed-loop fre-quency response in the vicinity of100kHz–200kHz. Without a feedbackcapacitor, the circuit will oscillate. A15pF feedback capacitor stabilizes the
circuit by adding a pole at 880kHz.This 12kΩ||15pF feedback networkincreases the settling time. The theo-retical minimum for the settling timeto 16-bit accuracy for a 1st orderlinear system is –ln(2–16) = 11.1 timeconstants set by the 12kΩ and 15pF,which equals 2.0µs. Figure 1’s circuitsettles in 2.4µs to 150µV for a 20Vstep.
The important DC specifications ofthis bipolar DAC circuit are integraland differential nonlinearity (INL andDNL), zero error and gain error. Theamplifiers contribute to these errorsthrough their input offset voltage(VOS), finite DC gain (AVOL) and invert-ing input bias current (IB–). Sinceboth amplifiers have their positiveinputs tied to ground, the noninvert-ing input bias current does not add toany errors. With this key applicationin mind, the design of the LT1469 isoptimized for a low IB–.
retemaraP eulaV
egatloVtesffOtupnI )xaM(Vµ521
tnerruCsaiBtupnIgnitrevnI )xaM(An01
tnerruCsaiBtupnIgnitrevninoN )xaM(An04
niaGCD Vm/V003
RRMC )niM(
)niM(
Bd69
RRSP )niM(Bd001
noitarapeSlennahC )niM(Bd001
egatloVesioNtupnI zH√/Vn5
tnerruCesioNtupnI Ap6.0 zH√/htdiwdnaBniaG zHM09
etaRwelS sµ/V22
sn009
sµ4.2
Bd5.69–
)xaM(Am2.5
Table 1. LT1469 specifications summary
(emiTgniltteS AV )petSV01,Vµ051,1–=
,7951CTLhtiw(emiTgniltteS CF )petSV02,Fp51=
V01rofDHT P-P zHk001,
V,tnerruCylppuS S )reifilpmAreP(V51±=
Linear Technology Magazine • May 2000 19
DESIGN FEATURES
The INL and DNL of the LTC1597are hardly affected by the surround-ing amplifiers. Figure 3 showsmeasured results of INL better than0.25LSB and DNL better than 0.1LSB,which is outstanding for 16-bit per-formance.
The effect of the amplifier’s VOS, IB–and AVOL on the system’s zero errorand gain error is a function of thenoise gain and DAC resistance. Theexact design equations have beenpresented in Linear TechnologyDesign Note 214. For a –10V to 10Voutput swing, the LSB of this 16-bit
system is 20V/216 = 305µV. Relativeto this LSB, the LT1469 worst-casespecifications lead to a zero error of3.6LSB and a full-scale gain error of4.9LSB. These numbers are insignifi-cant compared to the inherent DACspecifications.
With its low 5nV/√Hz input voltagenoise and 0.6pA/√Hz input currentnoise, the LT1469 contributes onlyan additional 23% to the DAC outputnoise voltage. The optional lowpassfilter at the output allows the designerto trade off resolution for settlingtime. A lower cutoff frequency elimi-nates wideband noise, as shown inFigure 2, whereas a higher cutofffrequency, such as the 1.6MHz shownin Figure 1, contributes only 0.1µs tothe settling time.
Single-Ended-to-Differential16-Bit ADC BufferFigure 4 illustrates the use of theLT1469 as a buffer for the LTC1604differential 16-bit ADC. The impor-
15pF
2kLTC159716-BIT
DAC INPUTS
REF
15pF
51pF
VOUT
–
+
–
+
1/2 LT1469
1/2 LT1469
12k12k12k12k
–15V
15V
5
6
8
7
3 2 1 4 5
6 2
3
4
1FREQUENCY (Hz)
90
SIGN
AL/(N
OISE
+ D
ISTO
RTIO
N) (d
B)
70
50
40
10 1k 10k 100k110
100
60
80
100
VREF = 6VRMS
500kHz FILTER
80kHz FILTER30kHz FILTER
tant amplifier specifications for thisapplication are low noise and lowdistortion. The LTC1604 16-bit ADCsignal-to-noise ratio (SNR) of 90dBimplies 56µVRMS noise at the input.The noise of the two amplifiers and100Ω/3000pF lowpass filter is only6.4µVRMS. The total noise includes acontribution from the source resis-tance. For a high value RS of 10kΩ,this amounts to 11.8µVRMS. Clearly,both noise sources taken together arestill well within the requirement for16-bit precision.
An advantage of driving theLTC1604 differentially is that the sig-nal swing at each input can bereduced, which reduces the distor-tion of both the ADC and the amplifier.For the ADC, a full-scale input meansthat AIN
+ – AIN– = ±2.5V. In single-
ended mode, with AIN– grounded, this
means that AIN+ must swing ±2.5V.
When driving both inputs differen-tially, each input must swing onlyhalf that amount, that is, ±1.25V. TheLTC1604 total harmonic distortion(THD) is a low –94dB at 100kHz. Thebuffer/filter combination alone has2nd and 3rd harmonic distortion bet-
DIGITAL INPUT CODE0
–1.0
INTE
GRAL
NON
LINE
ARIT
Y (L
SB)
–0.8
–0.4
–0.2
0
1.0
0.4
16384 32768
–0.6
0.6
0.8
0.2
49152 65535
DIGITAL INPUT CODE0
–1.0
DIFF
EREN
TIAL
NON
LINE
ARIT
Y (L
SB)
–0.8
–0.4
–0.2
0
1.0
0.4
16384 32768
–0.6
0.6
0.8
0.2
49152 65535
2k
VIN
2k
10pF
100Ω
100Ω
3000pF
3000pF
AIN+
LTC160416-BIT333kspsADC OUTPUTS
AIN–
–
+
–
+
1/2 LT1469
1/2 LT1469
5V
–5V
5V
–5V
RS 5
6
2
3
8
4
7
1
1
2
Figure 1. 16-bit DAC I/V converter and reference inverter
Figure 2. Signal to (noise plus distortion) forFigure 1’s circuit (code = all zeros)
Figure 3a. INL for Figure 1’s circuit
Figure 3b. DNL for Figure 1’s circuit Figure 4. Differential ADC buffer
Linear Technology Magazine • May 200020
DESIGN FEATURES
ter than –100dB for a ±1.25V, 100kHzinput, so it does not degrade the ACperformance of the ADC. Typical per-formance is shown in Figure 5.
Another advantage of operating indifferential mode is that commonmode errors of the ADC can bereduced. In single-ended mode, theADC sees a common mode signal atits inputs that is one-half of the inputsignal. With the LTC1604’s minimumCMRR of 68dB, this can result insignificant gain and offset errors atthe ADC output. In differential mode,only the LT1469 amplifiers see a com-mon mode at their inputs, whichresults in negligible errors thanks tothe 96dB CMRR of these amplifiers.The common mode signal at the ADCinput is now always 0V.
The buffer also drives the ADCfrom a low source impedance. With-out a buffer, the LTC1604 acquisitiontime increases with increasing sourceresistance above 100Ω and thereforethe maximum sampling rate must be
0
–20
–40
–60
–80
–100
–120
–1400 20 40 60 80 100 120 140 160
FREQUENCY (kHz)
AMPL
ITUD
E (d
B)
fSAMPLE = 333kspsVIN = ±1.25VfIN = 100kHzVS = ±5V
reduced. With the low noise, low dis-tortion LT1469 buffer, the ADC canbe driven at the maximum speed fromhigher source impedances withoutsacrificing AC performance.
The DC requirements for the ADCbuffer are relatively modest. The inputoffset voltage, CMRR and noninvert-ing input bias current through thesource resistance, RS, affect the DCaccuracy, but these errors are an
insignificant fraction of the ADC off-set and full-scale errors.
ConclusionThe LT1469 provides two fast andaccurate amplifiers in a single 8-leadSO or PDIP package. The unrivaledcombination of speed and accuracymake it the component of choice formany 16-bit systems.
Figure 5. 4096 point FFT of ADC output for Figure 4’s circuit
GND
SHDN FB
VIN SWLT1930
5 1
3
2
4
L1 10µH D1
R1115k
R213.3k
C24.7µFSHDN
C12.2µF
VIN5V
VOUT12V/300mA
C1: TAIYO-YUDEN X5R LMK212BJ225MG (408) 573-4150C2: TAIYO-YUDEN X5R EMK316BJ475MLD1: ON SEMICONDUCTOR MBR0520 (800) 282-9855L1: SUMIDA CR43-100 (847) 956-0667
++
output voltage remains within 1% ofthe nominal value during both tran-sient steps.
These applications demonstratethat the LT1930 is the industry’s high-est power SOT-23 switching regulator.In addition to step-up or boost con-verters, the LT1930 can be used insingle-ended primary inductance con-verters (SEPIC) and flyback designs.The LT1930 is pin compatible withboth the low power LT1613 and themicropower LT1615, providing asimple upgrade path for users of theolder parts who need more power.
Figure 3a. 5V to 12V/300mA step-up DC/DC converterFigure 3b. Efficiency of Figure 3a’s circuit
Figure 4. Transient response of Figure 3a’s circuit
LT1930, continued from page 1790
85
80
75
70
65
60
55
500 100 200 300 400
LOAD CURRENT(mA)
EFFI
CIEN
CY (%
)
50 150 250 350
VOUT0.2V/DIV
AC COUPLED
IL10.5A/DIV
AC COUPLED
250mA
200mALOAD CURRENT
20µs/DIV
Linear Technology Magazine • May 2000 21
DESIGN FEATURES
New Instrumentation Amplifier: Single-Resistor Gain Set and Precision FrontEnd Make Accuracy Easy by Glen Brisebois
The LT1168 is a low power,single-resistor gain-programmableinstrumentation amplifier that is easyto apply. It is a pin-compatible upgradefor the AD620 (and the INA118 with aresistor-value change) and is avail-able in C and I grades. With negligible60µV (Max) offset voltage and ultra-high 1TΩ input impedance, it cansense bridges with source impedancesfrom 10Ω to 100k without degradingthe signal. Its 120dB CMRR at 60Hz(AV = 100V) is achieved even with a 1ksource impedance imbalance. Match-ing the LT1168’s CMRR using discreteop amps would require the use of0.001% resistors. The LT1168’srobust inputs meet IEC1000-4-2Level 4 ESD tests with the addition oftwo external 5k series resistors. These5k resistors will contribute only anegligible 6µV of DC error.
Typical Application: AbsolutePressure Meter (Barometer)Figure 1 shows just how easy it is toapply the LT1167. The LT1097, inconjunction with the LT1634-1.25reference and R1, provides a preci-sion current drive for the bridge. D1ensures that the common mode input
range of the LT1097 is not exceededover temperature. With the currentdrive and the sensor bridge in place,the rest of the circuit is composedentirely of the LT1167 and its gain-set resistor, R2. The sensor gives afull-scale output of 50mV (±1%) at30psi (just over two atmospheres) witha 1.235V reference. With a 1.25Vreference, the full-scale output valuescales to 50.61mV. In order to get aconvenient and easily read output of100mV per 1psi, R2 should be chosento set the gain to 3V/.05061V = 59.28.Or, with 2.036 inches of mercuryequal to 1psi (referred to 39°F), changeR2 to set a gain of 120.7 for a conve-nient output of 100mV per 1inHg. Or,again, with 6.8947kPascals equal to1psi, set the gain to 408.7 for anoutput of 100mV per kiloPascal. Thesingle-resistor gain adjustment makes
it easy to accommodate the desiredunits effortlessly.
Setting the GainThe relation between the LT1168 gainand the gain-set resistor is given bythe following equations: RG = 49.4kΩ/(Gain – 1), and equivalently, Gain =(49.4kΩ/RG) + 1. In order to takeadvantage of the high CMRR withouttaking gain, simply do not connect anRG. This configures the LT1168 as aprecision, buffered, unity-gain differ-ence amplifier. If precision gain isdesired, use a 0.1% precision resistorin order not to degrade the 0.1% gainerror specification of the LT1168. Totarget the desired gain exactly, pick astandard 0.1% value that is slightlysmaller than required. Then, in orderto “top up” the total RG to the requiredvalue, use a small value standard 1%resistor in series, as shown in Figure2. As long as the 1% resistor is arelatively small value, say 1/20 of the0.1% value, the overall precision willnot be degraded. Table 1 shows prac-
niaGderiseDRG
)laciteroehT(R 1G
%1.0R 2G
"pU-poT"%1 niaGtnatluseR
82.95 Ω36.748 Ω548 Ω61.2 82.95
96.201 Ω27.214 Ω214 Ω1 16.021
27.804 Ω26.121 Ω121 Ω0 26.904
Table 1. Examples of series resistors for precision gain
–
+LT1097
5V
–5V
5V
–5V
C10.01µF
D1 1N4148
5V
LT1634-1.25
R1CURRENTSET~2k
2
3
7
4
6R2
3
2
81
7
45
6VOUT
6
5
4
2
1
3
ABSOLUTE PRESSURE SENSORIC SENSORS (MEASUREMENT SPECIALTIES)#1220-030A(408) 432-1800
GAIN R2 100mV =59.3 847.6Ω 1psi
102.7* 412.7Ω 1in Hg408.7* 121.6Ω 1kPa
*SEE TEXT FOR ACHIEVINGLARGE OUTPUT SWING
–
+LT1168
R333k
BRIDGE4× ~4k
RG10.1%
RG21%
2
18
3
RG1 = LARGER VALUE PRECISION RESISTORRG2 = SMALLER VALUE (A FEW OHMS) "TOP-UP" RESISTORRG = RG1 + RG2
GAIN = 49,400ΩRG
+1
REF
–
+LT1168 VOUT
6
5
Figure 1. Simple barometer Figure 2. Targeting gain precisely
Linear Technology Magazine • May 200022
DESIGN FEATURES
tical resistor values for the gains speci-fied above. Note that most transducersdo not have 0.1% accurate sensitivityanyway, so most applications woulduse a 1% resistor and then perform atwo-point calibration on the overallsystem.
Achieving LargeOutput SwingsFigure 3 shows the block diagram ofthe LT1168. Note that the last stage isa unity-gain difference amplifier. Thismeans that the output voltage (acrossthe Output and the Ref pins) willappear internal to the LT1168 acrossnodes Vx and Vy, centered aroundthe common mode input voltage(minus one VBE). In order to supportfull output swings, whether due tohigh applied gains or large differen-tial inputs, the common mode inputvoltage should not be too close toeither rail. For example, if the inputsare centered around ground and theRef pin is grounded, then under exci-tation Vx and Vy will swing aroundground (minus one VBE) and the out-put swing will be limited by the normaloutput stage limitation of about 1.2Vfrom either rail. However, if the com-mon mode input voltage is at 10V on±15V supplies, Vx and Vy will swingaround 9.3V. In this case, the outputswing will be limited by the outputs of
the intermediate amplifiers, A1 andA2, as they reach their upper limitsabout 1V from the rail. Thus, the finaloutput swing in this case will be lim-ited to ±4.6V even though the suppliesare ±15V. With regard to the lowerrail, if the common mode input volt-age is at –10V on ±15V supplies, thenVX and VY will swing around –10.7V.The final output swing in this casewill be limited to ±3.3V even thoughthe supplies are ±15V.
As a rule of thumb, in order toachieve the full output swing, keepthe common mode input voltage withinthe middle one-third of the supplies.For example, in the case of Figure 1,with gain = 408.7, the full-scale out-put voltage at 30psi (204.4kPa) wouldbe 20.44V. Providing the LT1168 with+24V/–5V supplies and, given thatthe bridge section of the sensor typi-cally has about 3V of excitation acrossit, replacing D1 with a 12V Zenerdiode will approximately center thecommon mode input voltage. Thisallows the LT1168 to achieve full 0V(vacuum) to 20.44V (204.4kPa, orabout 2.2 atmospheres) output swing.For information on using the LT1168on a single 5V supply, see “Designingthe LT1167 Instrumentation Amplifierinto a Single 5V Supply Application,”in Linear Technology IX:2 (June 1999).
Q1
RG
2
OUTPUT6
REF5
7
–
+A1
–
+A3
VB
R124.7k
R3400Ω
R4400Ω
C1
1
RG 8
R730k
R830k
R530k
R630k
DIFFERENCE AMPLIFIER STAGEPREAMP STAGE
+IN
–IN
3
–
+A2
VB
R224.7k
C2
V+
V–
V–
V+
V–
Q2 V–
V+
4 V–
VX
VY
FREQUENCY (kHz)
GAIN
(dB)
0.01
60
50
40
30
20
10
0
–10
–201 100 10000.1 10
G = 1000
G = 100
G = 10
G = 1
VS = ±15VTA = 25°C
FREQUENCY (Hz)0.1 1 10 100 1k 10k 100k
COM
MON
MOD
E RE
JECT
ION
RATI
O (d
B)
160
140
120
100
80
60
40
20
0
G = 1000
G = 100
G = 10
G = 1
VS = 15VTA = 25°C1k SOURCE IMBALANCE
AC Performanceat Low PowerThe LT1168 is not limited to DC andlow frequency applications. Figures 4and 5 show the frequency responseand common mode rejection ratio ofthe LT1168 for various gains. Thehigh frequency performance rendersthe LT1168 applicable in accelerom-eter and balanced audio line receiverapplications, for example, in additionto the usual slower applications. ThisAC performance is achieved with alow supply-current budget of only0.5mA.
ConclusionThe LT1168’s precision front end andsimple gain programmability makeaccuracy easy to achieve. The ACperformance of the internal amplifierresults in both extended differentialmode bandwidth and excellent com-mon mode rejection. Good commonmode rejection is essential forextracting signals from a hostile worldof interference.
Figure 3. LT1168 block diagram
Figure 4. LT1168 gain vs frequency
Figure 5. Common mode rejection ratio vsfrequency (1k source imbalance)
Linear Technology Magazine • May 2000 23
DESIGN IDEAS
SOT-23 SMBus Fan Speed ControllerExtends Battery Life and Reduces Noise
by David Canny
IntroductionBattery run times for notebook com-puters and other portable devices canbe improved and acoustic noisereduced by using Linear Technology’sLTC1695 to optimize the operation ofthese products’ internal cooling fans.The LTC1695 comes in a SOT-23package and provides all the func-tions necessary for a system controlleror microcontroller to regulate thespeed of a typical 5V/≤1watt fan via a
2-wire SMBus interface. By varyingthe fan speed according to the system’sinstantaneous cooling requirements,the power consumption of the coolingfan is reduced and battery run timesare improved. Acoustic noise ispractically eliminated by operatingthe fan below maximum speed whenthe thermal environment permits.Designers also have the option ofcontrolling the temperature in por-table devices by using feedback from
a temperature sensor to control thefan speed.
Figure 1 shows a typical applica-tion. Fan speed is easily programmedby sending a 6-bit digital code to theLTC1695 via the SMBus. This code isconverted into an analog referencevoltage that is used to regulate theoutput voltage of the LTC1695’sinternal linear regulator. The systemcontroller can enable an optional boostfeature that eliminates fan start-upproblems by outputting 5V to the fanfor 250ms before lowering the outputvoltage to its programmed value.Another important feature is that thesystem controller can read overcur-rent and overtemperature faultconditions from information storedin the LTC1695. The part’s SMBusAddress is hard-wired internally as1110 100 (MSB to LSB, A6 to A0) andthe data code bits D0 to D6 are latchedat the falling edge of the SMBus DataAcknowledge signal (D6 is a Boost-Start Enable bit and D5 to D0 translateto a linearly proportional output volt-age, 00–3F hex = 0V–5V). The
VCC
GND
SCL
VOUT
SDA
1
2
3
5
4
VCC
GND
SMB1
SMB2
1
2
5
4
LTC1695
LTC1694
SYSTEMCONTROLLER
5V
5V
3.3µF 4.7µFSUNONKED0502PFB1-85VDC FAN*0.6W
*(949) 583-9802
++
LTC1695OUTPUT VOLTAGE
2V/DIV
100ms/DIV
Figure 1. SMBus fan-speed controller
Figure 2. Fan start-up voltage profile
continued on page 25
DESIGN IDEASSOT-23 SMBus Fan SpeedController Extends Battery Lifeand Reduces Noise .................. 23David Canny
LTC1709 Low Cost, High Efficiency42A Converters with VID ControlReduce Input and OutputCapacitors ...............................24Wei Chen
ADSL Line Driver Design Guide,Part 2 ..................................... 26Tim Regan
Measure Resistances Easily,without Reference Resistor orCurrent Source ....................... 36Glen Brisebois
Linear Technology Magazine • May 200024
DESIGN IDEAS
+
RUN/SS
SENSE1+
SENSE1–
EAIN
PLLFLTR
PLLIN
NC
ITH
SGND
VDIFFOUT
VOS–
VOS+
SENSE2–
SENSE2+
ATTENOUT
ATTENIN
VID0
VID1
NC
TG1
SW1
BOOST1
VIN
BG1
EXTVCC
INTVCC
PGND
BG2
BOOST2
SW2
TG2
PGOOD
VBIAS
VID4
VID3
VID2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
C12470pF
1000pF
C8 1.2nF
C7 120pF
R7 15k
R6 15k
R3 10k
INTVCC
R2 2.7k
R451k
C5 0.01µF
C4 0.1µF
C102.2µF
C60.47µF
C190.1µF
C180.1µF
C11000pF
C2 0.1µF
LTC1709EG-8
C140.47µF
C131µFQ5
Q3
Q6
Q4
Q1 Q2
Q8
D1 BAT54A
D2 BAT54SA B
D2, C18 AND C19 ARE NEEDED ONLY IF VIN IS <5V;OTHERWISE, THEY CAN BE OMITTED AND POINTS A AND B SHORTED
C170.1µF
C1110µF6.3V
R9 10Ω
100k5V
Q7
C31µF
C161µF
C201µF
R1051Ω
R1151Ω
COUT
CIN
5VIN+
5VIN–
VOUT+
VOUT–
VOSENSE–
VOSENSE+
L11µH
L21µH R8
0.002Ω
R50.002Ω
+
+R1 10Ω
VID0 VID1 VID2 VID3 VID4 PGOOD
4 RUBYCON ALUM ELECT CAPACITORS 1500µF AT 6.3V (714) 668-89986 RUBYCON ALUM ELECT CAPACITORS 1500µF AT 6.3V OR 4 SANYO OS-CON 2R5SP1200M (619) 661-6835SUMIDA CEPH149-1R0MC (847) 956-0667FAIRCHILD FDS7760A (408) 822-2126OR SILICONIX Si4874 (800) 554-5565FREQUENCY = 200kHz
CIN:COUT:
L1, L2:Q1 TO Q8:
LTC1709 Low Cost, High Efficiency 42AConverters with VID Control ReduceInput and Output Capacitors by Wei Chen
IntroductionThe LTC1709-8/LTC1709-9 are dual,current mode, PolyPhase™ control-lers that drive two synchronous buckstages out of phase. This architecturereduces the number of input andoutput capacitors without increasingthe switching frequency. The rela-tively low switching frequency andintegrated high current MOSFETdrivers help provide high power-conversion efficiency for low voltage,high current applications. Because ofthe output ripple current cancella-tion, lower value inductors can be
used, resulting in a faster load tran-sient response. This, plus the 5-bitVID table, makes these devices par-ticularly attractive for CPU powersupply applications. Two VID tablesare available to comply with theVRM 8.4 (LTC1709-8) and VRM9.0(LTC1709-9) specifications.
Design ExampleFigure 1 shows the schematic dia-gram of a 42A power supply for theAMD Athlon microprocessor. Withonly one IC, eight tiny SO-8 MOS-
FETs and two 1µH low profile, sur-face mount inductors, an efficiency of86% is achieved for a 5V input and a1.6V/42A output. Greater than 85%efficiency can be maintained through-out the load range of 3A–42A, asshown in Figure 2. Because of the lowinput voltage, the reverse recoverylosses in the body diodes of the bot-tom MOSFETs are not significant. NoSchottky diodes are required in par-allel with the bottom MOSFETs inthis application.
Figure 1. Schematic diagram of a 42A power supply using the LTC1709
Linear Technology Magazine • May 2000 25
DESIGN IDEAS
Table 1 compares the input andoutput ripple currents for single-phaseand 2-phase configurations. A 2-phaseconverter reduces the input ripplecurrent by 50% and the output ripplecurrent by 75% compared to a single-phase design. The reduction in thecost and size of the input and outputcapacitors is significant.
Figure 3 shows the measured loadtransient waveform. The load currentchanges between 2A and 42A with aslew rate of about 30A/µs. Outputcapacitor type and size requirementsare dominated by the total ESR of theoutput capacitor network. Six low costaluminum electrolytic caps (Rubycon,1500µF/6.3V) are needed on the
LOAD CURRENT (A)0
50
EFFI
CIEN
CY (%
)
60
80
90
100
10 20 25 45
70
5 15 30 35 40
VIN = 5VVOUT = 1.6VfS = 200kHz
VOUT50mV/DIV
ILOAD20A/DIV
10µs/DIV
sesahPelppiRtupnI
A(tnerruC SMR )elppiRtuptuO
A(tnerruC P-P )
1 7.91 9.01 1
2 1.01 9.21 owtsesutiucricesahp-elgnisehttahtsemussA
A24edivorpotlellarapnisrotcudniA12/Hµ0.1.tuptuo
Table 1. Comparison of input and output ripplecurrent for single-phase and dual-phase configurations (L = 1µH, fS = 200kHz)
output to meet this requirement. Themaximum output voltage variationsduring the load transients are lessthan 200mVP-P. Active voltage posi-tioning was employed in this designto keep the number of output capaci-tors at six (refer to Linear TechnologyDesign Solutions 10 for more detailson active voltage positioning). R4 andR6 provide the output voltage posi-tioning with no loss of efficiency. IfOSCON caps are used, four 1200µF/2.5V (2R51200M) capacitors will besufficient.
ConclusionThe LTC1709 based, low voltage, highcurrent power supply described aboveachieves high efficiency and smallsize simultaneously. The savings inthe input and output capacitors,inductors and heat sinks help mini-mize the cost of the overall powersupply. This LTC1709 circuit, with afew modifications, is also suitable forVRM9.0 applications. Refer to LinearTechnology Application Note 77 formore information on the PolyPhasetechnique.
Figure 2. Efficiency vs load currentfor Figure 1’s circuit
Figure 3. Load transient waveforms at 40Astep and 30A/µs slew rate
SMBus Fan, continued from page 23
LTC1694, which also appears in Fig-ure 1, is a dual SMBus accelerator/pull-up device that may be used inconjunction with the LTC1695.
Boost-Start Timer,Thermal Shutdown andOvercurrent Clamp FeaturesA DC fan typically requires a startingvoltage higher than its minimum stallvoltage. For example, a Micronel 5Vfan requires a 3.5V starting voltage,but once started, it will run until itsterminal voltage drops below 2.1V (itsstall voltage). Thus, the user needs toensure that the fan starts up properlybefore programming the fan voltage toa value lower than the starting volt-age. Monitoring the fan’s DC currentfor stall conditions does not helpbecause some fans consume almostthe same amount of current at thesame terminal voltage in both stalledand operating conditions. Anotherapproach is to detect the absence of
fan commutation ripple current. This,however, is complex and requires cus-tomization for the characteristics ofspecific brands of fans. The LTC1695offers a simple and effective solutionthrough the use of a boost-start timer.By setting the Boost-Start Enable bithigh via the system controller, theLTC1695 outputs 5V for 250ms to thefan before lowering the voltage to itsprogrammed value (see Figure 2 forthe start-up voltage profile).
During a system controller Readcommand, bits 6 and 7 in the databyte code are defined as the ThermalShutdown Status (THE) and the Over-current Fault (OCF), respectively. Therest of the data byte’s register (bits 0to 5) are set low during host readback. The LTC1695 shuts down itsPMOS pass transistor and sets theTHE bit high if die junction tempera-ture exceeds 155°C. During anovercurrent fault, the LTC1695’s over-current detector sets the OCF bit highand actively clamps the output cur-
rent to 330mA. This protects theLTC1695’s PMOS pass transistor.Under dead short conditions (VOUT =0), although the LTC1695 clamps theoutput current, the large amount ofpower dissipated on the chip will forcethe LTC1695 into thermal shutdown.These LTC1695 dual protection fea-tures protect the IC and the fan and,more importantly, alerts the host tosystem thermal management faults.During a fault condition, the SMBuslogic continues to operate so that thehost can poll the fault status data.
ConclusionThe LTC1695 improves battery runtimes and reduces acoustic noise inportable equipment. In addition, itprovides important performance andprotection features by controlling theoperation of the equipment’s coolingfan. It comes in a SOT-23 packageand is easily programmed via theSMBus interface.
Linear Technology Magazine • May 200026
DESIGN IDEAS
ADSL Line Driver Design Guide, Part 2by Tim Regan
Part one of this article appeared inLinear Technology X:1 (February 2000)and is also avai lable on theLinear Technology web site atwww.lineartech.com/ezone/dsl.html.It discusses the different DSL stan-dards, characteristics of the DSLsignals, the design of differential driv-ers for DSL and the requirements foramplifiers used in this application.
Design Calculations, Volts,Amps and Power DissipationIt is very important to consider thepower requirements of the line driverin DSL applications. Although a nomi-nal power level of 100mWRMS or lessinto a 100Ω load does not seem to bea lot of power, the driver must handlelarge peak signals and thereforerequires a larger than nominal powersupply voltage. This increases boththe power dissipation in the driverpackage and the peak current capa-bility needed from the power supply.This issue becomes most critical incentral office designs, where manyDSL ports are included on a singlecard powered from one supply. Addi-tionally, the heat generated by thedrivers must be handled properly toensure reliable operation.
This section will provide the calcu-lations necessary to determine thevoltages, currents and power dissipa-tion for an ADSL driver of eitherstandard. It can be quite useful toplace these equations in a spread-sheet to allow quick observation ofthe effect of different design variableson the overall system. Assuming thata wide band, low distortion driver hasbeen selected (the LT1795 and LT1886are excellent choices), the three mostimportant system issues to considerare the total supply voltage, the peakoutput current and the driver powerdissipation required.
For these calculations, the RMSvoltages required are treated as DClevels for the purpose of estimating
the power dissipation. In an actualDSL design this approach overesti-mates the typical power dissipationwith a DMT signal by 10% to 20%because a data transmission is notalways at the maximum output powerlevel. The DSP intelligence built intothe system automatically adjusts thetransmitted power level and frequencyspectrum for each connection made.With shorter phone-line loops, thetransmitted power is reduced; withlonger loops, not all of the channelsare used and the number of data bitsper channel is reduced. The maxi-mum transmitted power is providedwhen the connection loop length is inthe range of 4000 feet to 10,000 feet
and there happens to be a significantlevel of noise interference and/or lowline impedance conditions. Design-ing to handle the conservativeestimate provides a margin of safetyfor reliable operation.
The Input VariablesBefore a design can begin, the follow-ing information must be known:which DSL standard is to be used,Full Rate or G.Lite, whether upstream(CPE) or downstream (CO). Thesesame equations apply for any DSLstandard (HDSL and HDSL2 forexample) with some changes to theinput parameters (see Table 1).
lobmyS retemaraP noitpircseDrofseulaVlacipyT
LSDA
P ENIL )mBd( rewoPeniL enilehtnotupebotrewopSMR
mBd02)OC,etaRlluF(
mBd3.61)OC,etiL.G(
mBd31dnaetaRlluF()EPC,etiL.G
RAP rotcaFtserC TMDehtrofoitaregareva-ot-kaePlangis 3.5
Z ENILeniL
ecnadepmIehtfoecnadepmicitsiretcarahC
enil Ω001
n oitaRsnruT gnilpuocenilehtfooitarsnrutehTremrofsnart rehgihro1:1
P SSOL )mBd( ssolnoitresnI remrofsnartehtfossolrewopehTdesugnieb mBd2otmBd2.0
V RHmoordaeHegatloV
noitarutastuptuoehtfonoitcnufAevitagendnaevitisop(segatlov
-daeH.desurevirdehtfo)gniwsowtehtforegralehteciwtsimoor
.segatlovnoitarutas
V5otV2
IQtnecseiuQ
tnerruC
)langistupnion(tnecseiuqlatoTsitahtrevirdehtfotnerrucylppus
.daolehtotdetrevidtonAm03otAm01
e NI egatloVtupnIlaitnereffidkaepo-tk-aepmumixaMgolana(EFAehtmorfegatlovtupni
)dnetnorfV5.4otV5.1 P-P
Table 1. Input variables
Linear Technology Magazine • May 2000 27
DESIGN IDEAS
Basic System RequirementsThe following equations determinethe essential operating requirementsindependent of the driver amplifierused in the design:
Line Power in Watts:
PLINE(W) = 10 • 1mWPLINE(dBm)
10 (1)example: 20dBm = 100mW.
RMS Line voltage:eLINE(RMS) = PLINE(W) • ZLINE (2)
Power to the primary of thetransformer:
PPRI(dBm) = PLINE(dBm) + PLOSS(dBm)
PPRI(W) = 10 • 1mW
(3)PPRI(dBm)
10
Impedance of primary of thetransformer:
ZPRI = ZLINE
n2(4)
Transformer termination resistors:
RBT1, RBT2 = ZPRI
2(5)
Primary RMS voltage:
ePRI(RMS) = PPRI(W) • ZPRI (6)Transformer primary RMS current:
IPRI(RMS) =ePRI(RMS)
ZPRI(7)
Driver amplifier RMS output voltage:
eAMPLIFIER(RMS) =
• ePRI(RMS)ZPRI + (2 • RBT)
ZPRI))
(8)
This is the RMS voltage betweenthe two amplifier outputs. If the RBTresistors are properly sized this volt-age is twice the RMS voltage of thetransformer primary.
Peak driver amplifier output current:
IPEAK =
= IPRI(RMS) • PAR
EPRI(RMS) • PARZPRI
(9)
The peak current handling capa-bility is key to selecting the driveramplifiers.Power supplied by the driveramplifiers:
POUT = eAMPLIFIERS(RMS) • IPRI(RMS) (10)
Overall line driver voltage gain:
AV(TOTAL) = eLINE(RMS) • 2 • PAR
eIN(11)
Differential amplifier voltage gain:
AVDIFF(AMPLIFIERS) =
eAMPLIFIERS(RMS) • 2 • PAReIN
(12)
The turns ratio of the transformerused is critical to the overall design.Figure 1 illustrates the minimum totalsupply voltage across the driver andthe peak driver output currentrequired as a function of the turnsratio. These are the absolute mini-mum requirements based on an idealamplifier that has 0V headroom andis therefore able to swing fully toeither supply voltage rail, and an idealtransformer, with zero insertion powerloss. A practical implementation willrequire a larger supply voltage, as
determined from the next section.Trying to design a system with lesssupply voltage or current capabilityusing conventional transformertermination resistors will result inclipping and transmission data errors.
Figure 1 also compares the differ-ent ADSL standards with the centraloffice, downstream, Full Rate ADSL,which requires the most current andvoltage. The reduced line powerrequirements for the downstreamG.Lite and the upstream Full Rateand G.Lite modems produce designswith lower voltage and currentrequirements.
Important DriverCharacteristics: HeadroomVoltage and QuiescentCurrentTo determine the required supply volt-age, power consumption and powerdissipation of the driver, the head-room voltage and required quiescentcurrent of the driver amplifier mustbe considered.
Minimum total supply voltage forthe amplifiers:
VSUPPLY(MIN) = EAMPLIFIER(RMS) (13)• PAR + VHR
The actual supply voltage for thedriver amplifier must be set above theminimum peak-to-peak amplifier out-put swing to provide for the headroomvoltage to prevent peak signal clip-ping. Using a supply voltage greaterthan this minimum value will increasethe power dissipation in the driveramplifiers.
1 1.5 2 2.5 3 3.5 4
40
35
30
25
20
15
10
5
0
800
700
600
500
400
300
200
100
0
TURNS RATIO (n)
MIN
IMUM
TOT
AL S
UPPL
Y VO
LTAG
E (V
)
MINIM
UM DRIVER PEAK OUTPUT CURRENT (m
A)
FULL RATE VS
UPSTREAM VS
FULLRATE IPEAK
G.LITE IPEAK
UPSTREAM IPEAK
G.LITE VS
VCC
VCC
VEE
VEE
VSAT–
VSAT+
+
–+
–LOAD
AMPLIFIEROUTPUTSTAGE
VHR–
VHR+
RSAT+
RSAT–
TEMPERATURE (°C)–50
OUTP
UT S
ATUR
ATIO
N VO
LTAG
E (V
)
V+
–1
–2
–3
–4
4
3
2
1
V–
0 50 75–25 25 100 125
VS = ±15V
RL = 2k
RL = 25Ω
RL = 25Ω
RL = 2k
VSAT+
VSAT–
RSAT– =
∆VOUT∆IOUT
RSAT+ =
∆VOUT∆IOUT
∆VOUT
∆VOUT
Figure 1. Minimum peak-to-peak driveroutput voltage and peak output currentrequired, ideal amplifier and transformer
Figure 2. Typical output stage model and common data sheet curvesare used to determine amplifier headroom voltage
Linear Technology Magazine • May 200028
DESIGN IDEAS
The headroom voltage of an ampli-fier is determined from either theguaranteed specification for outputvoltage swing or from characteristiccurves showing output saturationvoltage vs output current or vs tem-perature with different load currents.The headroom voltage is the differ-ence between the supply voltage railand the maximum output voltageswing, both positive and negative, fora given load current. Figure 2 showsa simple model for determining anamplifier’s output saturation voltagesand an example of a useful data sheetcurve.
During large signal transients, thetransistors in the output stage of theamplifier will fully turn on to pull theoutput as close as possible to thesupply voltage rails. The limitation onhow close the signal can swing can bemodeled as a fixed voltage drop acrossthe transistor being driven with aresistance in series. This resistanceincreases the voltage swing limitationin proportion to the amount of loadcurrent the transistor must source orsink. The combined total of the fixedvoltage drop and the voltage acrossthe resistor is called the output satu-ration voltage. The values to use tomodel this characteristic can bedetermined from a data sheet curve.Figure 2 shows the curve that appearson the LT1795 data sheet.
This curve shows the positive andnegative amplifier saturation voltagesvs junction temperature with two dif-ferent values of load resistance. DSLline drivers typically run warm, so thearea of interest on the curve will be inthe range of junction temperature
around 50°C. To determine the fixedvoltage part of the model for the posi-tive output swing, VSAT
+, evaluate thetop curve with RL = 2k. From thecurve it can be seen that the outputwill swing to within 1.2V of the posi-tive supply. Because the curve wasgenerated using supplies of ±15V, theload current at 50°C is only 13.8V/2kΩ or 7mA. To determine the valuefor the series resistance in the model,determine the change in output satu-ration voltage with a change in loadcurrent. At the same 50°C junctiontemperature point, evaluate the uppercurve with RL = 25Ω. With this loadthe output swings to within 1.8V ofthe positive rail and the load currentis 13.2V/25Ω or 528mA. The seriesresistance is then ∆VSAT/∆IOUT (0.6V/521mA), which is 1.15Ω. From thesevalues, the positive amplifier satura-tion voltage will be 1.2V + 1.15Ω •IPEAK where the value of IPEAK dependson the particular modem design.Applying the same approach for theamplifier swing towards the negativerail results in saturation voltage modelparameters of 1.2V in series with aresistance of 2.2Ω.
With these values modeling theoutput saturation characteristics ofthe LT1795, at any level of peak out-put current the output stage willsaturate or clip when swingingtowards the negative supply before itwill clip on the positive swing, due tothe higher effective series resistancevoltage drop. Transmission errors canoccur if either output swing excur-sion clips, so when sizing the totalsupply voltage requirement for thedriver the total headroom voltage of
the amplifier, VHR, should be twicethe larger of the two output satura-tion voltages. This will ensure thatthe output will not clip at all duringmaximum peak signal conditions.
With VSUPPLY set large enough toprevent signal clipping the total powerconsumption from the supplies canbe determined with Equation 14:
Power consumption of the com-plete line driver:
PIN = VSUPPLY • (IQ + IPRI(RMS)) (14)= (VEXTRA + VHR + VAMPLIFIER(RMS) • PAR)• (IQ + IPRI(RMS))
This equation introduces two newterms, VEXTRA and IQ. VEXTRA is thetotal additional power supply voltageabove VSUPPLY(MIN) that is actually usedto power the driver amplifiers. Forexample, if the minimum total supplyvoltage for a design is determined tobe 20V (or ±10V) but the actual sup-plies available are ±12V, then theVEXTRA term will be 24V – 20V or 4V.The total power consumption of eachline driver is very important whensizing the power supply for both volt-age and current capability to be usedin the system. This becomes mostsignificant when multiple DSL portsare to be powered from a predesignedpower supply. The power supply couldbecome the limiting factor to the num-ber of ports allowable.
The quiescent current, IQ, is basi-cally the operating supply current ofthe driver amplifiers. This is the cur-
IOUTPUT STAGESOURCING
IOUTPUT STAGESINKING
IBIAS
IBIAS
ISUPPLY+
ISUPPLY–
INPUTS
LOAD
ILOAD
IQ = ISUPPLY+ – ILOAD
30
25
20
15
10
5
0ACTU
AL A
MPL
IFIE
R QU
IESC
ENT
CURR
ENT
(mA)
LOAD CURRENT (mA)
0 100 200 300 400 500100200300400500SINKING SOURCING
IPRI(RMS)
IQ (NO SIGNAL)
1 1.5 2 2.5 3 3.5 4
4500
4000
3500
3000
2500
2000
1500
1000
500
0
TURNS RATIO (n)
P DIS
S (m
W)
FULL RATE CO (±15V)
FULL RATE CO (±12V)G.LITE CO (±15V)
CPE (±15V)CPE (±12V) G.LITE CO (12V)CPE (12V)
G.LITE CO(±12V)
FULL RATE G.LITE CPEIQ 28mA 18mA 14mAVHR 3V 2.5V 2VPLOSS 0.5dBm 0.5dBm 0.5dBm
ASSUMPTIONS
Figure 3. Much of an amplifier’s quiescent current is transferred to the load current
Figure 4. Driver power dissipation vsturns ratio: a practical implementation
Linear Technology Magazine • May 2000 29
DESIGN IDEAS
rent required to bias the internal cir-cuitry of the amplifiers. In general,high speed, high output currentamplifiers that process signals withvery low distortion require signifi-cantly more operating current thangeneral-purpose amplifiers. This cur-rent adds to the power consumptionand power dissipation of the driverpackage, because it must always besupplied whether there is signalapplied or not. However, the powerdissipation in the driver for the quies-cent current is not just a fixed DCpower of IQ • VSUPPLY. As seen in Figure3, much of the quiescent current isdiverted to the amplifier output stageand becomes part of the load currentwhile processing a signal. The curveshown is again for the LT1795 driver.With no load, all of the 30mA quies-cent current flows from the positivesupply through the amplifier to thenegative supply. However, when theload is sourcing or sinking 500mA,only 12mA flows through the ampli-fier, the remaining 18mA is taken bythe output stage and diverted tobecome part of the load current. Toobtain an accurate estimate of theaverage power dissipation of the driv-ers, this sharing of the quiescentcurrent should be taken into account.This will prevent overdesign of thethermal management area of con-cern. The IQ term in Equation 14should be the only current that con-tinues to flow through the amplifier atthe load current level of IPRI(RMS). Thediverted quiescent current is includedin the IPRI(RMS) term.
Unfortunately, this curve of quies-cent operating current vs load currentis not found on typical data sheets.Some characterization of the chosenamplifier should be done. The designof amplifier power output stages isvaried and has a direct effect on thediversion of the total supplied operat-ing quiescent current.Power dissipated in the line driveramplifiers:
PAMPLIFIERS = PIN – POUT (15)= eAMPLIFIERS(RMS) • [PAR • IQ + IPRI(RMS)• (PAR –1)] + (VHR + VEXTRA) •(IQ + IPRI(RMS))
The power dissipated in the driverpackage is important to consider whenaddressing heat management issues.
To minimize power dissipation, thedriver should be powered from a powersupply with voltages set to theminimum required. Most imp-lementations, however, use existingpower supply voltages, typically ±15V,±12V or just the 12V rail for the linedriver/receiver. Figure 4 provides anindication of the actual power dissi-pation in the line-driver amplifierpackage with commonly availablesupply voltages and a range of trans-former turns ratios. This is a practicalexample where values have beenassumed for the amplifier headroomand quiescent current and sometransformer power loss. The lowerpower upstream modems require lessoperating current, which helps tominimize the package power dissipa-tion. If the turns ratio is too low forthe given supply voltage, the lines onthe graph terminate because the sup-ply voltage is not large enough toprevent clipping of the DMT signalpeaks.
As previously stated, the powerdissipation in the driver is an impor-tant concern as it generates heat inthe system. For each of the ADSLstandards, a certain minimumamount of power dissipation is
required. Three factors that add tothis power dissipation are the ampli-fier headroom voltage, the amplifierquiescent operating current and thepower loss of the line-coupling trans-former. Attention to these three factorswhen selecting an amplifier and trans-former can optimize the overall powerdissipation. Analysis of the sensitivi-ties of the amplifier power dissipation(see Equation 15) for each of thesethree terms is summarized in Table 2.This shows the effect on total packagedissipation for each factor taken indi-vidually with the other two factors setto zero. The term n is the transformerturns ratio.
The factors in Table 2 provide arough indication of the additionalpower dissipation from these threesystem variables. The combined effecton power dissipation from IQ, VHR andPLOSS must still be determined fromEquation 15.
Optimizing PowerDissipation, AdjustableQuiescent Current andShutdownSeveral high speed power amplifiersfrom Linear Technology provide theability to externally set the operatingquiescent current. For the design ofany of the DSL standards, this allowsfor fine tuning the amplifier’s
PMIN • 1.023 • ))PLOSS(dBm)
0.1dBm– 1
Table 2. Additional power dissipation factors
dradnatSetaRlluFLSDA
maertsnwoDetiL.G
maertsnwoD
etaRlluFetiL.Gdna
maertspUlanoitiddA
noitapissiDrewoPmuminiMrewoP
,noitapissiDPMIN
Wm068 Wm763 Wm271
reifilpmAtnecseiuQI,tnerruC Q
n/Wm5.33 n/Wm41.22 n/Wm51 IfoAm1reP ,Q P SSID =I(•)ROTCAF( Q )Am1/
reifilpmAlatoTmoordaeHV,egatloV HR
Wm6.13•n Wm9.02•n Wm1.41•n VfoV1reP ,RH P SSID =V(•)ROTCAF( HR )V1/
remrofsnarT,ssoLnoitresnI
PLOSS mBdni%3.2 %3.2 %3.2
PfomBd1.0reP ,SSOLP SSID =
Linear Technology Magazine • May 200030
DESIGN IDEAS
operating point for minimum powerdissipation and adequate distortionperformance. There is a direct trade-of f between the two, however.Designing for very low quiescent cur-rent significantly reduces the powerdissipation, but obtaining the lowestdistortion performance requiresadditional biasing current for theinternal amplifier circuitry. Figure 5illustrates the adjustability of theoperating current for the LT1795. Aninternal current source is pro-grammed via a single external resistor.The current through this source ismirrored and scaled up to become thebiasing current for the two amplifi-ers. Also shown in Figure 5 is theeffect of adjusting the operating cur-rent on distortion. The spectrumanalyzer plots show the intermodula-tion components from twenty carriertones (from 200kHz to 500kHz). Withtoo low of an operating current, thesignal on the line is far too distortedand interference with other channelsis inevitable. However turning up thecurrent drops all of the distortionproducts into the noise floor. Thisadjustment should be made duringthe evaluation of the driver underactual transmission conditions andoptimized for the highest data ratesobtainable.
The best power and thermal man-agement technique in multiple-portsystems or energy efficient stand-alone modem designs is to shut offthe driver when the line is inactive.The digital circuitry always knows
when there is no data transmissionactivity and can issue a signal to thedriver to shut down operation. Manydrivers accept this control signal andcompletely power down the internalcircuitry. The LT1795, for example,can be shut down to consume lessthan 200µA of current when notrequired to transmit data. When com-manded to power up, the driverrequires only a few microseconds toreestablish full performance, aninsignificant time when compared toa typical communication training-upinterval. When powered down, how-ever, the output stage of the amplifierloses all bias and enters a highimpedance state. This essentiallyopens the connection to the trans-former back-termination resistors. Asthese resistors are often used to sensethe received signal from the line, nosignal can be developed across themif they are left floating.
Figure 6 illustrates a power savingfunction, called partial shutdown, that
keeps the amplifier slightly biasedand thus allows the modem tocontinue to monitor the line for trans-mission signals to be received. Here,two resistors are carefully chosen tocontrol the amount of operating qui-escent current as well as to retain asmall amount of “keep-alive” currentwhen shut down. Resistor scaling canaccommodate a direct connection toan I/O pin from the DSP processorwith any logic voltage level. Shuttingdown to a quiescent current level of2mA keeps the output stage activeand terminates the received signalsensing resistors, resulting in a betterthan 10-to-1 reduction in idle-channelpower consumption and dissipation.
Thermal ManagementDepending on the ADSL standardbeing applied, the power supplies andthe transformer turns ratio used, thedriver amplifier package will dissi-pate somewhere between 500mW and2W. The average power dissipation
VCC
SHUTDOWN
BIAS CURRENTFOR AMPLIFIER A
BIAS CURRENTFOR AMPLIFIER B
RADJ
IADJ
I = 115 • IADJ
(VCC – 1.3V)
SHDNREF
VCC
IBIASIBIAS
ISUPPLY(ON) = 115 • ION
SHDNREF
R1
R2IONIOFF
≈ 1.4V0V
3V OR 5V
20mA
2mA
3V
0V
V LOG
ICI S
UPPL
Y
TIME
ON
OFF
Figure 5a. Proper adjustment of the operatingcurrent minimizes spectral components,adjusting the sujpply current
Figure 5c. Spectrum of 20 carrier toneswith IQ of 2.2mA/amplifier
Figure 6. How to reduce driver supply current in an idle channelwhile maintaining the receiver function
Figure 5b. Spectrum of 20 carrier toneswith IQ of 12mA/amplifier
Linear Technology Magazine • May 2000 31
DESIGN IDEAS
times the overall thermal resistancefrom the junction of the driver to theambient air will determine the rise inoperating junction temperature abovethe maximum ambient temperature.Most power amplifiers have a built inthermal protection mechanism thatwill disable the output stage when thejunction temperature exceeds typi-cally 160°C. Should this temperatureever be reached, the amplifier willprotect itself, but data transmissionerrors will abound and most likelyresult in a data transmission discon-nect. Designing a heat-spreadingsystem to limit the driver junctiontemperature to less than 125°C at thehighest expected ambient tempera-ture will ensure continuous operation.
Fortunately, the power dissipationlevels are not so high that externalheat sinks are necessarily required,so heat spreading can usually bemanaged through planes of PCB cop-per foil. In addition, the packaging ofmost power amplifiers uses thermalconduction enhancements, such asfused or exposed lead frames. Fusedlead frames have several package pinsconnected directly to the metal padwhere the IC is attached. This pro-
vides a continuous path for heat trans-fer from the junction of the IC, out ofthe plastic encapsulation, to pins thatare directly connected to PCB copperplanes. An exposed lead frame doesnot plastic encapsulate the under-side metal where the IC is attached.This provides a metal pad that can beconnected directly to PCB copper fordirect transfer of heat from the ICmounting junction heat source to theambient air. An exposed lead frameallows for very small packages, suchas that used for the LT1795CFE, a20-pin TSSOP, to have thermal con-ductivity characteristics similar tomuch larger sized packages. Verysmall packages with good thermalconductivity can result in very densemultiport ADSL systems for centraloffice applications.
The best way to spread the heatgenerated by the driver is to use asmany planes of copper as are avail-able and to “stitch” them togetherthrough small vias from the topside ofthe board to the bottom, as shown inFigure 7. These vias should be smallenough in diameter (15 mils or less)that they are completely filled withsolder during the plating process. Thisprovides a continuous thermal con-ductivity path from the top of theboard to the bottom for the mostexposure to the ambient environment.There are no fixed rules for determin-ing the lateral area of the copperplanes on the PCB, other than “biggeris better,” and 2oz copper is a thickerand therefore better thermal conduc-tor than 1oz copper. Figure 7 alsoprovides an indication of the improve-ment in the heat spreading thermalresistance from junction to case withvarious amounts of copper foil areaon the top and bottom sides of a PCB.
As most of the heat is dissipated inthe area immediately surrounding thedriver amplifier package, there comesa point of diminishing returns wheremore copper area does not providemuch additional benefit. This can beseen in the plot of thermal resistancein Figure 7 where, beyond a total PCBarea of 1in2, further reduction in ther-mal resistance is minimal. One wordof caution regarding PCB planes forheat spreading is that the fiberglassmaterial (typically FR-4) is a fairlygood thermal insulator. Any compo-nent interconnect traces that cutthrough the plane of copper signifi-cantly reduce the effectiveness of thelateral area. Interconnect tracesshould be made on the inner layers ofmultilayer boards to minimize thedistance between components. Thecomplex interconnect of the logic cir-cuits used in DSL modems usuallyrequires a multilayer PC board thatcan be put to good use in the linedriver area.
Another measure that can be takenis to provide some forced airflow cool-ing. A linear flow of air across thedriver package can significantlyreduce the effective thermal resis-tance from junction to ambient (θJA)of the heat-spreading system. A re-duction of 2°C/W to 3°C/W for each100lfpm (linear feet per minute) can
DRIVER PACKAGE
TOP LAYER COPPERINTERMEDIATE COPPER LAYERSBOTTOM LAYER COPPER
13MIL VIAS THAT FILL DURING THE PLATING PROCESS
0.7"
0.75
"
0.5"
0.75
"
TOP BOTTOM
TOTAL PCB FOIL AREA—TOP AND BOTTOM SIDES (IN2)
θ JA
°C/W
40
38
36
34
32
300.5 0.7 0.9 1.4 1.8 2.3
LT1795CFE20-PIN TSSOP PACKAGE WITH EXPOSED LEAD FRAME FOR DIRECT METALLIC CONTACT TO PCB FOIL
+
+
+0.1µF
0.1µF
0.1µF
0.1µF
10µF
10µF
10µF
DRIVERV+ PIN
DRIVERV– PIN
VCC
VEE
Figure 7a. Using PCB copper foil for heat sinking
Figure 7b. Improving Heat dissipationwith increased copper foil area
Figure 8. Recommended powersupply bypassing for any design
Linear Technology Magazine • May 200032
DESIGN IDEAS
be achieved. This is particularlyimportant in a multiport systemhoused in an enclosed case.
A Gallery of DesignRecommendationsThis section will provide examples ofdriver and receiver circuits for each ofthe ADSL standards. These circuitsprovide a good starting point for imple-menting the line interface functionsfor a DSL modem. The circuits weredesigned with all of the consider-ations mentioned so far, but othersystem variables, such as availablesupply voltages or AFE output andinput dynamic range, could mandatesome modifications. The total voltagegain of each line-driver design, fromthe differential input voltage to theactual voltage output to the phoneline, has been scaled to a value thatrequires less than 3VP-P from the AFEproviding the transmitted signal. Thegain of the amplifier stage is adjustedto take into account the signal boostof the transformer used as well as thesignal loss through the back-termi-nation resistors.
Common to all of the designs is agood power supply bypassingapproach. This is shown in Figure 8.A large- and a small-valued bypasscapacitor at the points where the sup-plies connect to the board providedecoupling of noise and ripple over awide frequency range. Additional highfrequency decoupling at the driverand receiver supply pins is recom-mended. Another large-valued bypasscapacitor connected directly betweenthe supply pins of the driver helps toreduce the 2nd harmonic componentof ripple on the supply lines. Thiscomponent comes from the peak cur-rent demands from each supply,which occur twice for each input sig-nal cycle due to the differentialamplifier topology (each amplifiersources and sinks the peak currentonce each signal cycle).
The Differential ReceiverNot all DSL modems will require areceiver circuit. Some analog frontend ICs have sophisticated circuitryfor a very wide dynamic input range
to directly pick the small receivedsignals out of the noise floor afterpassing through the receive/echo fil-ter. Other designs may use a secondtransformer to process the differen-tial received signal directly to the
filter/AFE. Many designs still preferto sense the differential signal acrossthe termination resistors and providegain to the received signal before pass-ing it through the filter to the AFE.This basic differential receiver circuit
–
+
–
+
100ΩPHONE LINE
1:n
0.1µF
0.1µF
eRCV
eRX
DB
A C
RECEIVEAND
ECHOFILTER
+DRIVER
–DRIVER
RBT+
RBT–
RC
RD
RB
RA
RF1
RF2
VEE
VCC
RCV–
RCV+
AFERECEIVEINPUTS
RB = RARC = RD
AV = RF1 OR RF2
RC OR RD
3
2
5
6
1
7
8
4
CF1
CF2
–
+
12V
RF1 1k
RF2 1k
RG1187Ω
eIN+
eIN–
RBT1 12.4Ω
RBT2 12.4Ω–
+
100ΩPHONE LINE
RG2187Ω
T11:2
A11/2 LT1886
A21/2 LT1886
RC2267ΩCC2
47pF
RC1267ΩCC1
47pF
+
C11µF
C21µF
1µF
10k
10k
0.1µF
0.1µF
RIN120k
RIN220k
12V
DB
A C
0.1µF
0.1µF
T1: COILCRAFT x8390A (847) 639-6400OR MIDCOM 50215 (605) 886-4385
DRIVER CHARACTERISTICS
POWER CONSUMPTION: 470mWPOWER DISSIPATION: 425mWPEAK DRIVER CURRENT: 159mA
RECEIVER COMPONENTS(SEE FIGURE 9)RA, RB: 2kRC, RD: 1kRF1, RF2: 4.02kCF1, CF2: 72pF (G.Lite)
36pF (Full Rate)RECEIVER GAINeRX
eRCV= 1
eLINEeIN
= 6AV =
eIN
eLINE
3
2
8
1
6
5
7
4
Figure 9. Basic differential receiver (4-wire to 2-wire)
Figure 10. Full Rate or G.Lite upstream (CPE) driver
Linear Technology Magazine • May 2000 33
DESIGN IDEAS
is shown in Figure 9. Each receiveramplifier is a summing stage thatsums the received signal and theattenuated transmitted signal seenat the primary of the transformerwith a weighted, opposite-phasetransmitted signal. This weightedsumming of the transmitted signalideally cancels the 180° out-of-phase
signals, leaving only the received sig-nal at the differential amplifieroutputs. This is called local echo can-cellation. In a standard line-driverdesign, the transmit signals at nodesA and B in Figure 9 are twice themagnitude of the signals at nodes Cand D. To cancel these signals in thereceiver requires resistors RA and RB
be set to exactly twice the value ofresistors RC and RD.
The gain of the receiver is simplythe inverting gain of the received sig-nal path, RF1/RC and RF2/RD. In thedriver design examples to follow, thereceiver input resistors connect tothe driver at nodes A through D. Therecommended component values forthe receiver provide for unity gainfrom the received signal appearing atthe line to the differential receiveroutput. This takes into account theattenuation of the line-coupling trans-former. A small feedback capacitor isalso shown that reduces the gain at afrequency just above the receivedsignal bandwidth, which variesdepending on the application.
ADSL Full Rate or G.LiteUpstream (CPE) Line DriverThis driver (Figure 10) is the lowestpowered of the ADSL standards, con-suming less than 500mW. The lowerline power, 13dBm, and resultinglower peak current requirement allowsthe use of the LT1886, which is a highspeed 200mA dual amplifier. The useof a 2:1 transformer turns ratio allowsthis driver to be powered from a single12V power supply.
In order to obtain the highest open-loop gain and bandwidth to minimizedistortion, the LT1886 is decompen-sated and is only stable withclosed-loop gains of ten or greater. Inthis design the signal gain of eachamplifier is only 6.35. To remain stablewith this low value of gain requiresthe addition of gain-compensationcomponents RC1, CC1, RC2 and CC2.These components, which come intoplay only at frequencies greater than15MHz, parallel the gain-settingresistances, RG1 and RG2, to make thefeedback factor of each amplifier avalue of 0.9, which is the same ashaving a closed-loop gain of ten; thus,stability is ensured.
The LT1886 is a 700MHz gain band-width amplifier. The combination ofgain at such high frequencies and notbeing unity gain stable requires thatthe gain-setting resistors be returnedto a low impedance at all frequencies.For this reason, the two gain setting
–
+
12V
RF1 1k
RF2 1k
RG133Ω
eIN+
eIN–
RBT1 34.8Ω
RBT2 34.8Ω–
+
100ΩPHONE LINE
T11:1.2
A11/2 LT1795-
CFE
A21/2 LT1795-
CFE
RIN110k
RIN210k
DB
A C
0.1µF
0.1µF
T1: COILCRAFT x8502-A (847) 639-6400
DRIVER CHARACTERISTICS
POWER CONSUMPTION: 880mWPOWER DISSIPATION: 785mWPEAK DRIVER CURRENT: 141mA
RECEIVER COMPONENTS(SEE FIGURE 9)RA, RB: 2kRC, RD: 1kRF1, RF2: 2.37kCF1, CF2: 500pF
RECEIVER GAINeRX
eRCV= 1
eLINEeIN
= 9AV =
eIN
eLINE
4
3
14
18
8
7
13
20
–12V
110
11
6
159k
5
17
–
+
12V
RF1 1k
RF2 1k
RG169Ω
eIN+
eIN–
RBT1 12.4Ω
RBT2 12.4Ω–
+
100ΩPHONE LINE
T11:2
A11/2 LT1795-
CFE
A21/2 LT1795-
CFE
RIN110k
RIN210k
DB
A C
0.1µF
0.1µF
DRIVER CHARACTERISTICS
POWER CONSUMPTION: 1.88WPOWER DISSIPATION: 1.66WPEAK DRIVER CURRENT: 355mA
RECEIVER COMPONENTS(SEE FIGURE 9)RA, RB: 2kRC, RD: 1kRF1, RF2: 4.02kCF1, CF2: 270pF
RECEIVER GAINeRX
eRCV= 1
eLINEeIN
= 12AV =
eIN
eLINE
4
3
14
18
8
7
13
20
–12V
110
11
6
97.6k
5
17
T1: COILCRAFT x8390A (847) 639-6400OR MIDCOM 50215 (605) 886-4385
Figure 11. ADSL G.Lite downstream (CO) line driver
Figure 12 ADSL Full Rate downstream (CO) line driver
Linear Technology Magazine • May 200034
DESIGN IDEAS
resistors are connected to groundrather than using a single resistorconnected to the other amplifier’sinverting input. Capacitors C1 andC2 are included to prevent applyinggain to the DC offset voltages of theamplifiers. The different values of feed-back capacitors for the receiveramplifier account for the frequencyspectrum of the downstream infor-mation from the CO modem in eitherthe Full Rate (1104kHz) or G.Lite(552kHz) implementation.
ADSL G.Lite Downstream (CO)Line DriverThis moderate power (16.4dBm) driverrequires less than 1W and is shown inFigure 11. This design is biased from±12V supplies and uses a transformerwith a turns ratio of only 1:1.2.Although the peak current is only140mA, the LT1886 cannot be useddue to its limited operating supplyvoltage of 13.2V total. Instead theLT1795CFE, which is in a very smallTSSOP power package, is used. Thissmall package is ideal for central office,multiple DSL port designs for com-pacting a high number of drivers on asingle PC card.
ADSL Full Rate Downstream(CO) Line DriverFigure 12 is the highest powered DSLline driver application, used in cen-tral office applications to obtain up to8Mbps data rates throughout theInternet. This design uses standardback termination and can be poweredfrom ±12V supplies by using a 2:1turns ratio transformer. This resultsin a fairly high, 355mA peak outputcurrent demand from the amplifiers.The LT1795, with a 500mA outputcurrent rating, once again is capableof the task.
Reduced Power DissipationADSL Full Rate Downstream(CO) Line DriverTo address the power consumptionand dissipation issues for Full RateADSL drivers, a slightly modifiedtopology can be used, as shown inFigure 13. Recognizing that one-halfof the power provided by the amplifi-ers is lost in the transformerback-termination resistors, an obvi-ous approach to reduce power is tosimply reduce the value of theseresistors. Doing so, however, modifiesthe output impedance of the modemas seen from the phone line and also
–
+
12V
RF1 1k
RF2 1k
RG226Ω
eIN+
eIN–
RBT1 13.3Ω
RBT2 13.3Ω–
+
100ΩPHONE LINE
T11:1.5
A11/2 LT1795-
CFE
A21/2 LT1795-
CFE
RIN110k
RIN210k
DB
A C
0.1µF
0.1µF
T1: COILCRAFT x8505-A (847) 639-6400
DRIVER CHARACTERISTICS
POWER CONSUMPTION: 1.5WPOWER DISSIPATION: 1.33WPEAK DRIVER CURRENT: 266mA
RECEIVER COMPONENTS(SEE FIGURE 9)RA, RB: 1.62kRC, RD: 1.02kRF1, RF2: 5.11kCF1, CF2: 240pF
RECEIVER GAINeRX
eRCV= 1
eLINEeIN
= 12AV =
eIN
eLINE
4
3
14
18
8
7
13
20
–12V
110
11
6
97.6k
5
17
RP1 2.49k
RP2 2.49k
reduces the amount of received sig-nal developed across these sensingresistors. Although it is powered by±12V supplies, the circuit of Figure13 achieves 300mW of power savings.The driver current is substantiallyreduced by using a transformer turnsratio of only 1.5:1. Normally, thiswould require a higher supply voltageof ±14V and RBT resistors of 22.2Ω.However, although the RBT resistorsare reduced to 13.3Ω, the circuit stillmaintain the proper line-impedancetermination of 100Ω and operatesfrom ±12V supplies. It is not suitablefor every application, however,because it still reduces the amount ofreceived signal. It is most applicablefor systems that use a sensitivereceiver AFE that can still detect thereduced received signal.
The approach is termed active ter-mination. A small amount of positivefeedback in each amplifier is obtainedfrom the opposite amplifier output.This feedback makes the effectiveoutput impedance seen looking intothe circuit at nodes C and D theproper value even though the RBTresistor has been reduced by 40%from what is should be. The designequations for this topology are asfollows.
Instead of using the standard valueof RBT resistance, it can be reduced toany value desired, with attendantreceived-signal loss. A factor called Kcan be used to define the new RBTresistance:
RBT = (16)K • ZLINE
2 • n2
With standard termination and a1:1.5 turns ratio transformer, thevalue of RBT should be 22.2Ω. In thedesign of Figure 13, this resistor isreduced by 40% to 13.3Ω, thereforethe factor K = 0.6.
The normal forward path circuitgain from the noninverting input ofeach amplifier to the output nodes Aand B is a term called G where G = 1+RF/RG.
The gain of the positive feedbacksignal path for each side (from node Dto A and from node C to B, is called Pwhere P = RF/RP.
Figure 13. Reduced power dissipation ADSL Full Rate downstream (C) line driver
Linear Technology Magazine • May 2000 35
DESIGN IDEAS
srevirDeniL
traP 5971TL 7021TL 6881TL 7941TL 6021TL 0121TL
lauD/elgniS lauD lauD lauD lauD elgniS elgniS
tnerruCtuptuO Am005 Am052 Am002 Am521 Am052 A1.1
egatloVylppuS V03otV01 V03otV01 V31otV5 V03otV5 V03otV01 V03otV01htdiwdnaBniaG
tcudorP zHM05 zHM06 zHM57 zHM05 zHM06 zHM53
etaRwelS sµ/V009 sµ/V009 sµ/V002 sµ/V009 sµ/V009 sµ/V009
IQ reifilpmA/ Am03otAm1 Am03otAm1 Am7 Am01 Am03otAm1 Am05otAm1
V TAS+ V2.1 V2.1 V57.0 V2.1 V2.1 V2.1
V TAS– V2.1 V2.1 V9.0 V51.1 V2.1 V52.1
R TAS+ .1 2Ω .3 2Ω Ω1.3 14Ω .3 2Ω .0 9Ω
R TAS– 2Ω .5 3Ω .2 3Ω 10Ω .5 3Ω .1 7Ω
srevieceRreifilpmA-lauD
traP 5531TL 8531TL 1631TL 4631TL 3181TL 3521TL
egatloVylppuS V03otV5 V03otV5 V03otV5 V03otV5 V21otV5 V42otV01
htdiwdnaBniaG zHM21 zHM52 zHM05 zHM07 zHM001 zHM09
etaRwelS sµ/V004 sµ/V006 sµ/V008 sµ/V0001 sµ/V057 sµ/V052
egatloVesioN zH√/Vn01 zH√/Vn8 zH√/Vn9 zH√/Vn9 zH√/Vn8 zH√/Vn3
IQ reifilpmA/ Am52.1 Am5.2 Am5 Am5.7 Am3 Am6
Table 3. Driver and receiver amplifier characteristics
Using these abbreviations:For proper impedance matching: P =1 – K.
To obtain a desired voltage gainfrom the AFE output to the line, AV,the term G is set to:
G = AV • ePRI
eLINE• (1 + K – P) – P (17)
where ePRI and eLINE are the voltagesat the transformer primary and onthe line, determined by taking into
account the turns ratio and trans-former insertion loss.
The use of a high performanceamplifier such as the LT1795 doesnot result in any degradation of dis-tortion performance when modifyingthe closed-loop gain by positive feed-back. Significant power savings canbe obtained but the design may not besuitable for all applications as previ-ously mentioned.
ConclusionFollowing the design proceduresdescribed in this article should makethe design and implementation easyand accurate. At the very least, it willensure that power and heat issuesreceive proper consideration.
Linear Technology offers a varietyof high speed, low distortion poweramplifiers and low noise dual ampli-fiers that can be used to implementthe driver/receiver functions of theDSL modem (see Table 3).
For more information on parts featured in this issue, seehttp://www.linear-tech.com/go/ltmag
For more information on parts featured in this issue, seehttp://www.linear-tech.com/go/ltmag
Authors can be contactedat (408) 432-1900
Linear Technology Magazine • May 200036
DESIGN IDEAS
Measure Resistances Easily, withoutReference Resistor or Current Source
by Glen Brisebois
Measuring the resistance of adevice, for example a thermistor, usu-ally requires biasing it with a precisioncurrent source or combining it withseveral other precision resistors in abridge. The circuit of Figure 1 showshow to use the new LT1168 instru-mentation amplifier to achieve aprecision resistance-to-voltage con-version as simply as possible.Normally, the resistor across pins 1and 8 is the gain-set resistor and thevoltage across pins 3 and 2 is thevariable to be measured. In this case,
however, the 1.25V reference estab-lishes a fixed input voltage so that thevariable to be measured is now theresistance. The equation for VOUT vsRT is VOUT = 1.25V • 49.4kΩ/RT. Giventhe limitation on output swing (withthe supply voltages shown), the small-est measurable resistance is about4.5k. The highest resolvable resis-tance is limited to about 200M by the300µV output offset voltage of theLT1168. The 0.05% accuracy of theLT1634 is not an issue here, becauseit is subtracted at the Ref pin of the
LT1168 and only contributes to gainerror. Figure 2 shows output voltagevs temperature for 10k and 100k (at25°C) thermistors from two manufac-turers. Thermistors are difficult tolinearize, so although the output isstill not linear with temperature, itcan, at least, be read directly by anADC and compared against a lookuptable. The circuit has good noiseimmunity but does not toleratecapacitance at pin 1 or 8 and so is notideal for resistive devices placedremotely from the LT1168.
–
+
15V
–15V
–15V
LT1634-1.25
RT LT1168
2
1
3
45REF
6
7
8VOUT = 1.25 •
49.4kΩRT
22k
14
12
10
8
6
4
2
0–40 –20 0 20 40 60 80 100 120
TEMPERATURE (°C)
OUTP
UT V
OLTA
GE (V
)
YSI #44011
YSI #44006
THERMO-METRICSDC95G104Z
THERMOMETRICSDC95F103W
Figure 1. Simple resistance-to-voltage converter Figure 2. Output voltage vs temperature for thermistors fromtwo manufacturers. Curves are approximations to aid design—contact manufacturers for exact lookup tables:YSI (800) 765-4974; Thermometrics (732) 287-2870.
http://www.linear-tech.com/ezone/zone.htmlArticles, Design Ideas, Tips from the Lab…
Linear Technology Magazine • May 2000 37
NEW DEVICE CAMEOS
LTC1726: A Micropower,Precision Triple SupplyMonitor Adds AdjustableReset and Watchdog TimerFunctionsIn multiple-processor systems, suchas network servers and routers, 5V,3.3V and 2.5V supplies are very com-mon and are typically designed to±5% tolerances. The LTC1726 solvesthe system-level problem of providingearly warning of supply voltage fail-ure to system control logic. Also, itincorporates additional features sothat system designers can implementa variable reset time-out and/or vari-able watchdog time-out. In theLTC1726, both the reset and thewatchdog time-out periods are fullyadjustable using external capacitors.
In the systems mentioned above,PC board area is at a premium. Toaddress this system constraint, theLTC1726 is available in either SO-8or MSOP-8 packaging. For monitor-ing three supplies, two resistors andtwo capacitors are the only externalcomponents required by the LTC1726;Like the LTC1727 and the LTC1728(see “New Device Cameos” in LinearTechnology IX:2, June 1999), theLTC1726 is available in two versions:the LTC1726-5 is designed to moni-tor 5V, 3.3V and one user-definedsupply voltage. The LTC1726-2.5 isdesigned to monitor 3.3V, 2.5V andone user-defined supply voltage. Theadjustable supply voltage monitor isa high impedance input with a 1Vthreshold that allows the systemdesigner to program the monitoredvoltage using two external resistors.Both versions operate at a supplycurrent of 16µA and offer tight ±1.5%threshold accuracy over temperatureand glitch immunity that ensuresreliable reset operation without falsetriggering.
The LTC1726 is powered fromeither of the two preset supply voltagemonitor inputs, allowing the resetoutput to be active low, independentof supply voltage sequencing. In
addition, the reset output is guaran-teed to be active low for supply voltagesdown to 1V.
LTC1751-5 Regulated ChargePump Delivers 5V at 100mAwithout InductorsCharge pump–based power suppliesoffer the advantages of low solutioncost, simplicity and small size. Untilnow, however, their output currentcapability has been very limited. Whileretaining the best features of its pre-decessors, the LTC1751-5 delivers100mA (500mW) to the load from anMS8 package while stepping up 3V toa regulated 5V. Additionally, itprovides shutdown capability,a power-good feature and pro-grammable soft-start timing. TheLTC1751-5 also has built-in thermalshutdown circuitry that allows it tosurvive a continuous short circuit toground at its output.
The quiescent supply current ofthe LTC1751-5 is only 20µA. This lowsupply current ensures very low powerconsumption in light load applica-tions, resulting in extended batterylife. Furthermore, because it usesBurst Mode operation, its efficiencywith a 3V input is 82.4%, which isvery close to the theoretical maxi-mum charge pump efficiency of 83.3%.In shutdown the supply current isless than 1µA.
The PGOOD pin can be used toalert a microcontroller when the out-put voltage of the LTC1751-5 hasreached its final value. This is useful,for example, if the LTC1751-5 is pro-viding power to a peripheral devicethat will communicate with the micro-controller. Furthermore, in the eventof an undervoltage fault at the VOUTpin, the PGOOD pin reports the fault.Once the fault is removed, PGOODreturns to the “all-clear” state.
An optional soft-start circuit allowsthe LTC1751-5’s output voltage to beincreased as slowly as necessary. Aslow rise time on VOUT will preventunnecessary start-up and loading
problems by limiting the current intothe output capacitor. A small exter-nal capacitor on the SS pin programsthe rise time to an appropriate rate. Ifstart-up current isn’t an issue, the SScapacitor can be omitted.
With no inductors and only threeto four small capacitors, theLTC1751-5 regulated charge pumpdelivers significant power and func-tionality from a very small footprint.
LT1962 300mA Low Noise,Micropower Low DropoutRegulator Saves Current inBattery-Powered Applicationsand Operates with SmallCeramic Capacitors
The LT1962 is a low noise, lowdropout linear regulator that is ratedfor 300mA of output current at adropout voltage of 300mV and comesin an 8-lead MSOP package. Theregulator is designed for use in bat-tery-powered systems with 30µAquiescent current and less than 0.1µAsupply current in shutdown. TheLT1962 can operate with as little as3.3µF of output capacitance. Anycapacitor, including small ceramiccapacitors, can be used on the outputwithout the need for additional seriesresistance, as is commonly requiredwith other regulators. Quiescent cur-rent is well controlled; it does not risein dropout, as is the case with manycompeting devices.
The LT1962 features low noiseoperation. With the addition of anexternal 0.01µF bypass capacitor,output voltage noise over the 10Hz-to-100kHz bandwidth is reduced to20µVRMS. This is the lowest outputvoltage noise of any linear regulatorcurrently available. The LT1962 iscapable of operating over a wide 1.8Vto 20V supply range. Internalprotection circuitry includes reverse-battery protection, current limiting,thermal limiting and reverse-currentprotection.
The LT1962 is available in fixedoutput voltages of 2.5V, 3V, 3.3V and5V, or as an adjustable device with anoutput voltage range of 1.22V to 20V.The LT1962 is packaged in the space-saving 8-lead MSOP package with a
New Device Cameos
Linear Technology Magazine • May 200038
NEW DEVICE CAMEOS
fused leadframe for improved ther-mal resistance.
LT1963 1.5A Low Noise, LowDropout Regulator ProvidesFast Transient ResponseThe LT1963 is a low noise, low drop-out linear regulator rated for 1.5A ofoutput current at a dropout voltage of350mV. The regulator is designed foruse in applications where fast tran-sient response is necessary, such aspowering ASICs or FPGAs. The LT1963can operate with as little as 10µF ofoutput capacitance. Small ceramiccapacitors can be used without theneed for additional series resistance,as is commonly required with otherregulators. The 1mA quiescent cur-rent drops to less than 1µA inshutdown. Quiescent current is wellcontrolled; it does not rise in dropout.
The LT1963 features low noiseoperation. Output voltage noise overthe 10Hz-to-100kHz bandwidth isunder 40µVRMS. The LT1963 is alsocapable of operating over a wide 1.9Vto 20V supply range. Internalprotection circuitry includes reverse-battery protection, current limiting,thermal limiting and reverse-currentprotection.
The LT1963 is available in fixedoutput voltages of 1.8V, 2.5V and3.3V or as an adjustable device withan output voltage range of 1.21V to20V. The LT1963 is available in an8-lead SO package, with full featuresand a fused leadframe for lower ther-mal resistance; it is also available ina 3-lead SOT-223 without shutdownand remote sensing, for a small size/
low thermal resistance combination.Surface mount 5-lead DD and throughhole 5-lead TO-220 packages are avail-able for applications requiring higherpower dissipation.
LTC1706-82 Programs Powerfor Next Generation PentiumProcessors Down to 1.1Vwith ±0.25% AccuracyThe LTC1706-82 is a 5-bit desktopVID voltage programmer that pro-grams the output of a whole family ofLTC controller-based DC/DC convert-ers to supply a precise input supplyvoltage to the next generation of Pen-tium® microprocessors. The Pentiumprocessor’s supply voltage require-ments change with the various clockspeed options. To meet this need, theLTC1706-82 uses its five program-mable VID inputs to program a ±0.25%accurate output voltage from 1.10Vto 1.85V in 25mV steps. This is fullycompliant with the Intel Pentium Pro-cessor VID Specification (VRM 9.0).
The LTC1706-82 comes in the small10-lead MSOP package. It consumespractically zero current (only deviceleakage) when all five inputs are high;each grounded VID input adds only68µA of input current in a 3.3V sys-tem. For extremely high currentapplications (up to 200A), such ashigh-end servers and workstations,just one LTC1706-82 can program upto six LTC1629 PolyPhase, highefficiency, step-down DC/DC control-lers to complete an extremely compact,powerful, programmable power sup-ply that uses only surface mountcomponents.
In addition to the LTC1629, theLTC1706-82 also works equally wellwith the LTC1735, LTC1702,LTC1628 and other LTC DC/DC con-verters with onboard 0.8V references.
LTC1779 Step-Down DC/DCConverter Delivers 250mAfrom Tiny 6-Lead SOT-23The LTC1779 is a constant frequency,current mode, step-down DC/DC con-verter capable of 250mA outputcurrent operation without an exter-nal power switch or sense resistor. Itsconstant 550kHz operating frequencyallows the use of a small externalinductor. This feature, coupled withthe part’s tiny SOT-23 package,results in a very small footprintsolution.
The LTC1779 operates over a wideVIN range of 2.5V to 9.8V and is capableof efficiencies up to 90% while main-taining excellent AC and DC load andline regulation. The device, whichboasts ±2.5% output voltage accu-racy, consumes only 130µA ofquiescent current in normal opera-tion and a mere 8µA in shutdown.Under light load conditions, the deviceoptimizes efficiency using Burst Modeoperation. To maximize the life of thebattery source, the internal powerswitch is turned on continuously indropout (100% duty cycle).Information furnished by Linear Technology Corporationis believed to be accurate and reliable. However, noresponsibility is assumed for its use. Linear TechnologyCorporation makes no representation that the intercon-nection of its circuits as described herein will not infringeon existing patent rights.
For further information on anyof the devices mentioned in thisissue of Linear Technology, usethe reader service card or callthe LTC literature servicenumber:
1-800-4-LINEAR
Ask for the pertinent data sheetsand Application Notes.
Linear Technology Magazine • May 2000 39
DESIGN TOOLS
Technical Books1990 Linear Databook, Vol I —This 1440 page collec-tion of data sheets covers op amps, voltage regulators,references, comparators, filters, PWMs, data conver-sion and interface products (bipolar and CMOS), in bothcommercial and military grades. The catalog featureswell over 300 devices. $10.00
1992 Linear Databook, Vol II — This 1248 page supple-ment to the 1990 Linear Databook is a collection of allproducts introduced in 1991 and 1992. The catalogcontains full data sheets for over 140 devices. The 1992Linear Databook, Vol II is a companion to the 1990Linear Databook, which should not be discarded.
$10.00
1994 Linear Databook, Vol III —This 1826 page supple-ment to the 1990 and 1992 Linear Databooks is acollection of all products introduced since 1992. A totalof 152 product data sheets are included with updatedselection guides. The 1994 Linear Databook Vol III is acompanion to the 1990 and 1992 Linear Databooks,which should not be discarded. $10.00
1995 Linear Databook, Vol IV —This 1152 page supple-ment to the 1990, 1992 and 1994 Linear Databooks is acollection of all products introduced since 1994. A totalof 80 product data sheets are included with updatedselection guides. The 1995 Linear Databook Vol IV is acompanion to the 1990, 1992 and 1994 Linear Databooks,which should not be discarded. $10.00
1996 Linear Databook, Vol V —This 1152 page supple-ment to the 1990, 1992, 1994 and 1995 Linear Databooksis a collection of all products introduced since 1995. Atotal of 65 product data sheets are included with updatedselection guides. The 1996 Linear Databook Vol V is acompanion to the 1990, 1992, 1994 and 1995 LinearDatabooks, which should not be discarded. $10.00
1997 Linear Databook, Vol VI —This 1360 page supple-ment to the 1990, 1992, 1994, 1995 and 1996 LinearDatabooks is a collection of all products introducedsince 1996. A total of 79 product data sheets areincluded with updated selection guides. The 1997 LinearDatabook Vol VI is a companion to the 1990, 1992,1994, 1995 and 1996 Linear Databooks, which shouldnot be discarded. $10.00
1999 Linear Data Book, Vol VII — This 1968 pagesupplement to the 1990, 1992, 1994, 1995, 1996 and1997 Linear Databooks is a collection of all product datasheets introduced since 1997. A total of 120 productdata sheets are included, with updated selection guides.The 1999 Linear Databook is a companion to the previ-ous Linear Databooks, which should not be discarded.
$10.00
1990 Linear Applications Handbook, Volume I —928 pages full of application ideas covered in depth by40 Application Notes and 33 Design Notes. This catalogcovers a broad range of “real world” linear circuitry. Inaddition to detailed, systems-oriented circuits, this hand-book contains broad tutorial content together with liberaluse of schematics and scope photography. A specialfeature in this edition includes a 22-page section onSPICE macromodels. $20.00
1993 Linear Applications Handbook, Volume II —Continues the stream of “real world” linear circuitryinitiated by the 1990 Handbook. Similar in scope to the1990 edition, the new book covers Application Notes 40through 54 and Design Notes 33 through 69. References
and articles from non-LTC publications that we havefound useful are also included. $20.00
1997 Linear Applications Handbook, Volume III —This 976 page handbook maintains the practical outlookand tutorial nature of previous efforts, while broadeningtopic selection. This new book includes ApplicationNotes 55 through 69 and Design Notes 70 through 144.Subjects include switching regulators, measurementand control circuits, filters, video designs, interface,data converters, power products, battery chargers andCCFL inverters. An extensive subject index referencescircuits in LTC data sheets, design notes, applicationnotes and Linear Technology magazines. $20.00
1998 Data Converter Handbook — This impressive 1360page handbook includes all of the data sheets, applica-tion notes and design notes for Linear Technology’sfamily of high performance data converter products.Products include A/D converters (ADCs), D/A convert-ers (DACs) and multiplexers—including the fastestmonolithic 16-bit ADC, the 3Msps, 12-bit ADC with thebest dynamic performance and the first dual 12-bit DACin an SO-8 package. Also included are selection guidesfor references, op amps and filters and a glossary of dataconverter terms. $10.00
Interface Product Handbook — This 424 page hand-book features LTC’s complete line of line driver andreceiver products for RS232, RS485, RS423, RS422,V.35 and AppleTalk® applications. Linear’s particularexpertise in this area involves low power consumption,high numbers of drivers and receivers in one package,mixed RS232 and RS485 devices, 10kV ESD protectionof RS232 devices and surface mount packages.
Available at no charge
Power Management Solutions Brochure — This 96page collection of circuits contains real-life solutions forcommon power supply design problems. There are over70 circuits, including descriptions, graphs and perfor-mance specifications. Topics covered include batterychargers, desktop PC power supplies, notebook PCpower supplies, portable electronics power supplies,distributed power supplies, telecommunications andisolated power supplies, off-line power supplies andpower management circuits. Selection guides are pro-vided for each section and a variety of helpful designtools are also listed for quick reference.
Available at no charge.
Data Conversion Solutions Brochure — This 64 pagecollection of data conversion circuits, products andselection guides serves as excellent reference for thedata acquisition system designer. Over 60 products areshowcased, solving problems in low power, small sizeand high performance data conversion applications—with performance graphs and specifications. Topicscovered include ADCs, DACs, voltage references andanalog multiplexers. A complete glossary defines dataconversion specifications; a list of selected applicationand design notes is also included.
Available at no charge
Telecommunications Solutions Brochure —This 76page collection of application circuits and selectionguides covers a wide variety of products targeted fortelecommunications. Circuits solve real life problemsfor central office switching, cellular phones, high speedmodems, base station, plus special sections covering–48V and Hot SwapTM applications. Many applications
highlight new products such as Hot Swap controllers,power products, high speed amplifiers, A/D converters,interface transceivers and filters. Includes a telecom-munications glossary, serial interface standards, protocolinformation and a complete list of key application notesand design notes. Available at no charge.
Applications on DiskFilterCAD™ 2.0 CD-ROM — This CD is a powerful filterdesign tool that supports all of Linear Technology’s highperformance switched capacitor filters. Included is Fil-terView™, a document navigator that allows you toquickly find Linear Technology monolithic filter datasheets, the FilterCAD manual, application notes, designnotes and Linear Technology magazine articles. It doesnot have to be installed to run FilterCAD. It is notnecessary to use FilterView to view the documents, asthey are standard .PDF files, readable with any versionof Adobe Acrobat™. FilterCAD runs on Windows® 3.1 orWindows 95. FilterView requires Windows 95. TheFilterCAD program itself is also available on the web andwill be included on the new LinearView™ CD.
Available at no charge.
Noise Disk — This IBM-PC (or compatible) programallows the user to calculate circuit noise using LTC opamps, determine the best LTC op amp for a low noiseapplication, display the noise data for LTC op amps,calculate resistor noise and calculate noise using specsfor any op amp. Available at no charge
SPICE Macromodel Disk — This IBM-PC (or compat-ible) high density diskette contains the library of LTC opamp SPICE macromodels. The models can be used withany version of SPICE for general analog circuit simula-tions. The diskette also contains working circuit examplesusing the models and a demonstration copy of PSPICE™by MicroSim. Available at no charge
SwitcherCAD™ — The SwitcherCAD program is a pow-erful PC software tool that aids in the design andoptimization of switching regulators. The program cancut days off the design cycle by selecting topologies,calculating operating points and specifying componentvalues and manufacturer’s part numbers. 144 pagemanual included. $20.00
SwitcherCAD supports the following parts: LT1070 se-ries: LT1070, LT1071, LT1072, LT1074 and LT1076.LT1082. LT1170 series: LT1170, LT1171, LT1172 andLT1176. It also supports: LT1268, LT1269 and LT1507.LT1270 series: LT1270 and LT1271. LT1371 series:LT1371, LT1372, LT1373, LT1375, LT1376 and LT1377.
Micropower SwitcherCAD™ — The MicropowerSCADprogram is a powerful tool for designing DC/DC con-verters based on Linear Technology’s micropowerswitching regulator ICs. Given basic design parameters,MicropowerSCAD selects a circuit topology and offersyou a selection of appropriate Linear Technology switch-ing regulator ICs. MicropowerSCAD also performs circuitsimulations to select the other components which sur-round the DC/DC converter. In the case of a batterysupply, MicropowerSCAD can perform a battery lifesimulation. 44 page manual included. $20.00
MicropowerSCAD supports the following LTC micro-power DC/DC converters: LT1073, LT1107, LT1108,LT1109, LT1109A, LT1110, LT1111, LT1173, LTC1174,LT1300, LT1301 and LT1303.
continued on page 40
DESIGN TOOLS
Linear Technology Magazine • May 2000© 2000 Linear Technology Corporation/Printed in U.S.A./40K
LINEAR TECHNOLOGY CORPORATION1630 McCarthy BoulevardMilpitas, CA 95035-7417(408) 432-1900 FAX (408) 434-0507www.linear-tech.comFor Literature Only: 1-800-4-LINEAR
Acrobat is a trademark of Adobe Systems, Inc.; Windowsis a registered trademark of Microsoft Corp.; AppleTalk isa registered trademark of Apple Computer, Inc. Pentiumis a registered trademark of Intel Corp.; PSPICE is atrademark of MicroSim Corp.
CD-ROM CatalogLinearView — LinearView™ CD-ROM version 4.0 isLinear Technology’s latest interactive CD-ROM. It allowsyou to instantly access thousands of pages of productand applications information, covering LinearTechnology’s complete line of high performance analogproducts, with easy-to-use search tools.
The LinearView CD-ROM includes the complete productspecifications from Linear Technology’s Databook library(Volumes I–VII) and the complete Applications Hand-book collection (Volumes I–III). Our extensive collectionof Design Notes and the complete collection of LinearTechnology magazine are also included.
A powerful search engine built into the LinearView CD-ROM enables you to select parts by various criteria,such as device parameters, keywords or part numbers.All product categories are represented: data conversion,references, amplifiers, power products, filters and inter-face circuits. Up-to-date versions of Linear Technology’s
DESIGN TOOLS, continued from page 39 software design tools, SwitcherCAD, Micropower Switch-erCAD, FilterCAD, Noise Disk and Spice Macromodellibrary, are also included. Everything you need to knowabout Linear Technology’s products and applications isreadily accessible via LinearView. LinearView runs underWindows 95 or later.
Available at no charge.
World Wide Web SiteLinear Technology Corporation’s customers can nowquickly and conveniently find and retrieve the latesttechnical information covering the Company’s productson LTC’s Internet web site. Located at www.linear-tech.com, this site allows anyone with Internet accessand a web browser to search through all of LTC’stechnical publications, including data sheets, applica-tion notes, design notes, Linear Technology magazineissues and other LTC publications, to find informationon LTC parts and applications circuits. Other areaswithin the site include help, news and information aboutLinear Technology and its sales offices.
Other web sites usually require the visitor to downloadlarge document files to see if they contain the desiredinformation. This is cumbersome and inconvenient. Tosave you time and ensure that you receive the correctinformation the first time, the first page of each datasheet, application note and Linear Technology maga-zine is recreated in a fast, download-friendly format.This allows you to determine whether the document iswhat you need, before downloading the entire file.
The site is searchable by criteria such as part numbers,functions, topics and applications. The search is per-formed on a user-defined combination of data sheets,application notes, design notes and Linear Technologymagazine articles. Any data sheet, application note,design note or magazine article can be downloaded orfaxed back. (Files are downloaded in Adobe Acrobat PDFformat; you will need a copy of Acrobat Reader to viewor print them. The site includes a link from which youcan download this program.)
InternationalSales OfficesFRANCELinear Technology S.A.R.L.Immeuble “Le Quartz”58 Chemin de la Justice92290 Chatenay MalabryFrancePhone: 33-1-41079555FAX: 33-1-46314613
GERMANYLinear Technology GmbHOskar-Messter-Str. 24D-85737 IsmaningGermanyPhone: 49-89-962455-0FAX: 49-89-963147
HONG KONGLinear Technology Corp. Ltd.Unit 2109, Metroplaza Tower 2223 Hing Fong RoadKwai Fong, N.T., Hong KongPhone: 852-2428-0303FAX: 852-2348-0885
JAPANLinear Technology KK5F NAO Bldg.1-14 Shin-Ogawa-cho Shinjuku-kuTokyo, 162JapanPhone: 81-3-3267-7891FAX: 81-3-3267-8510
KOREALinear Technology Korea Co., LtdYundang Building #1002Samsung-Dong 144-23Kangnam-Ku, Seoul 135-090KoreaPhone: 82-2-792-1617FAX: 82-2-792-1619
SINGAPORELinear Technology Pte. Ltd.507 Yishun Industrial Park ASingapore 768734Phone: 65-753-2692FAX: 65-752-0108
SWEDENLinear Technology ABSollentunavägen 63S-191 40 SollentunaSwedenPhone: 46-8-623-1600FAX: 46-8-623-1650
TAIWANLinear Technology CorporationRm. 602, No. 46, Sec. 2Chung Shan N. Rd.Taipei, Taiwan, R.O.C.Phone: 886-2-2521-7575FAX: 886-2-2562-2285
UNITED KINGDOMLinear Technology (UK) Ltd.The Coliseum, Riverside WayCamberley, Surrey GU15 3YLUnited KingdomPhone: 44-1276-677676FAX: 44-1276-64851
World HeadquartersLinear Technology Corporation1630 McCarthy BoulevardMilpitas, CA 95035-7417Phone: (408) 432-1900FAX: (408) 434-0507
U.S. AreaSales OfficesNORTHEAST REGIONLinear Technology Corporation3220 Tillman Drive, Suite 120Bensalem, PA 19020Phone: (215) 638-9667FAX: (215) 638-9764
Linear Technology Corporation15 Research PlaceNorth Chelmsford, MA 01863Phone: (978) 656-4750FAX: (978) 656-4760
NORTHWEST REGIONLinear Technology Corporation720 Sycamore DriveMilpitas, CA 95035Phone: (408) 428-2050FAX: (408) 432-6331
SOUTHEAST REGIONLinear Technology Corporation17000 Dallas Parkway, Suite 219Dallas, TX 75248Phone: (972) 733-3071FAX: (972) 380-5138
Linear Technology Corporation9430 Research Blvd.Echelon IV Suite 400Austin, TX 78759Phone: (512) 343-3679FAX: (512) 343-3680
Linear Technology Corporation1080 W. Sam Houston Pkwy., Suite 225Houston, TX 77043Phone: (713) 463-5001FAX: (713) 463-5009
Linear Technology Corporation5510 Six Forks Road, Suite 102Raleigh, NC 27609Phone: (919) 870-5106FAX: (919) 870-8831
CENTRAL REGIONLinear Technology Corporation2010 E. Algonquin Road, Suite 209Schaumburg, IL 60173Phone: (847) 925-0860FAX: (847) 925-0878
Linear Technology CorporationKenosha, WI 53144Phone: (414) 859-1900FAX: (414) 859-1974
SOUTHWEST REGIONLinear Technology Corporation21243 Ventura Blvd., Suite 208Woodland Hills, CA 91364Phone: (818) 703-0835FAX: (818) 703-0517
Linear Technology Corporation15375 Barranca Parkway, Suite A-213Irvine, CA 92618Phone: (949) 453-4650FAX: (949) 453-4765