a publication for the radio amateur worldwide especially

A Publication for theRadio Amateur Worldwide

Especially Covering VHF,UHF and Microwaves

Volume No.39 . Spring . 2007-Q1 . £5.25

Display unit for the power detector byDJ8ES

Alexander Meier, DG6RBP

VHFCOMMUNICATIONS

Alexander Meier Display unit for the power detector 2 - 10DG6RBP by DJ8ES

Franco Rota Noise source diodes 11 - 18I2FHW

Henning C Weddig Extending the 50MHz converter into 19 - 30DK5LV a transverter 35 - 40

Index of volume 38 (2006) 31 - 33

Gunthard Kraus Practical Project: Durable and reproducible 41 - 55DG8GB patch antenna for the 2.45GHz WLAN

band, Part 1

John Fielding Corrections to transistor multiplier 56 - 58ZS5JF article published in issue 4/2006

Carl Lodström Instrumentation amplifier noise 59 - 60KQ6AX & SM6MOM (additional information for the article

published in issue 4/2006)

EADX 6m contest rules 61

Gunthard Kraus Internet Treasure Trove 62 - 63DG8GB

K M Publications, 63 Ringwood Road Luton, Beds, LU2 7BG, UKTelephone / Fax +44 (0)1582 581051, email : [email protected]

web : http://www.vhfcomm.co.uk

Thank you to everyone who has renewed their subscription for 2007. This issue seesthe return of an article from Franco Rota, he has been working hard to find aninteresting topic and I think he has succeeded. I have also been asked to publish therules for a Spanish 6m contest that will hopefully increase activity on 6m in Spain.73s - Andy

Contents

VHF COMMUNICATIONS 1/2007

1

In the article by Wolfgang Schneideron his power detector, published inissue 3/2006 [1], he mentioned a dis-play unit. This indicates the measuredpower, on five 7 segment displays, indBm or Watts. With very little extra hardware cost anoffset function can be added. This canbe used to offset the frequency re-sponse of the sensor for more precisemeasurements or the use of a couplerfor measuring higher powers.

1.Introduction

Because the 12 Bit A/D converter isintegrated on the detector PCB [1], themeasuring signal can be transferred sim-ply with a multi way cable from thesensor to the display unit. The digitalsignal is fed directly to an AT89C52micro controller from ATMEL, this com-putes the measured power in dBm. Theuse of a micro controller makes it possi-ble to add further, practice oriented fea-tures like: relative measurements (displayof the level relative to a measured value),the conversion and display of the powerin Watts, a PC interface (RS-232 orUSB) and the consideration of an offset.

With the latter, for example, the varia-tions in the level response of the detectorcan be taken into consideration for par-ticularly precise measurement. If an ex-ternal coupler is used to be able tomeasure higher power levels in the Wattrange, the extra attenuation can be takeninto account for the displayed value.Powers up to +60dBm (1000W) can bedisplayed. The circuit to do all this canbe realised using very few components.

2.Description of the circuit

The circuit diagram of the display unit isshown in Fig 1 and is the same as thecircuit of the frequency input module [2]as far as possible. This display unit forthe power detector has ESD protection onthe detector input, several light emittingdiodes and a serial interface. The serialinterface is 0/5v level suitable for USBbut can have an interface converter addedsuch as the MAX232 [4] for a conven-tional RS-232 interface. A direct connec-tion between the display unit and an RS-232 interface of a PC without an inter-face converter is possible.

The technical data of the display unit

Alexander Meier, DG6RBP

Display unit for the powerdetector by DJ8ES


2

Fi

g 1

: Cir

cuit

diag

ram

of t

he d

ispl

ay u

nit.


3

with the power detector is summarised inTable 1.

The main part of the circuit is the microcontroler U1 (AT89C52) from ATMEL.This frequently used and proven control-ler is inexpensive and has an integratedFlash memory to store the software. AnEEPROM (U3) is used to permanently

store static data e.g. sensor calibration.The switches and LEDs needed to oper-ate the unit are mounted on the PCB withthe micro controller. The use of SMDLEDs and light pipes on the front panelincreases the ESD protection.

Port 1 and 3 of the micro controller drivethe five 7 Segment displays to show the

Power Detector (with DG6RBP PCB)1

Frequency response 10MHz to >2.3GHzMeasuring range -50dBm to +10dBmTypical resolution 0.02dB Input impedance 50Ω Typical level error ±1dB PCB size 62 x 18 x 0.6mmDisplay UnitDisplay range -55.0dBm to +59.9dBm

3.00nW to 995W (3 digits) Display resolution 0.1dB (for display in dBm)

0.02dB (for display in watts)2

Measurment interval 4 measurements per secondReference level for relative measuremants Last value used before relative functionAttenuator compensation function -1.9dB to +49.9dB (adjustable)Interface3 Serial, 19200 Baud, 8 data bits, 1 stop

bit, no parity, 0/5V level Power supply 9 - 12V approximately 130mAPCB size 47 x 90 x 1.5mm

Table 1 : Technical data for power detector and display unit.

Fig 2 : Frequencyresponse of thepower detectorusing optimisedPCB.


4

measured values. These displays, D1 toD5, are activated alternately for 1 ms viatheir common anode, this reduces thecurrent consumption. The power input ofthe circuit is thus below 150mA.

The power detector is connected by theplug, J1, directly to the micro controller.All connections are protected againstelectrostatic discharge using diodes. Thesupply voltage for the sensor can be fedvia the plug connector J1. A suitableconnector on the front panel of theprototype wattmeter was a mini DINsocket (with 6 pins) which worked satis-factorily.

The circuit is powered by an external 9 to12V DC supply. The voltage regulationconsists of an internal 5V regulator forthe display unit. The operating voltagefor the sensor is stabilised on the sensorprinted circuit board. [1]

Explanations for the data from Table 1:

1 For the power detector [1] a somewhatsmaller printed circuit board with opti-mised conductors and smaller SMDparts was designed. Also ESD protec-tion was added on the connections. Alarge capacitor (10µF) is used on portCLPF so that various modulation sig-nals e.g. AM Modulation, can bemeasured correctly. Fig 2 shows theimproved frequency response of thesensor. The small size of the 0603SMD components used gives an opti-mised response.

2 The power is measured in dBm with aresolution of 0.02dB and converted todisplay in Watts. The display of powerin Watts always takes place to threedecimal places.

3 The circuit of the display unit can beattached directly to a USB interface[3].

Fig 3 : Layout forthe front side ofthe display unitPCB.

Fig 4 : Layout forthe back side of thedisplay unit PCB.


5

3.Construction

The layout of the double sided printedcircuit board for the display unit is shownin Fig 3 and 4. Because of the small size,only 47mm x 90mm, the small “Ultra-mas” enclosure from Bopla [5] can beused. This plastic housing is well de-signed for measuring instruments withfolding feet to raise the front panel.

The component layout is shown in Fig 5and 6. There are no special requirementfor assembling the PCB. An IC socketshould be used for U1 to make softwareupdates easier. Note that the micro con-troller, connectors and some SMD com-ponents are on the back of the PCB. The

displays, switches, crystal and remainingSMD components are on the front side ofthe PCB.

4.Parts list

U1 AT89C52, DIP, programmedU2 L4931CDT50, DPAKU3 24C01-2.7, SO-8Q1-Q13 BCV47D1-D5 SA39-11 green D6-D7, LED green, SMD 1206D9-D11, D15 D8 SMBJ16D12-D14 CDS3C09GTA

Fig 5 : Componentlayout for the frontside of the displayunit PCB.

Fig 6 : Compo-nent layout forthe back side ofthe display unitPCB.


6

R1-R8 47Ω, SMD 0805 R9-R10, 220Ω, SMD 0805R15-18 R13, R14 4.7KΩ, SMD 0805R12 10K, SMD 1206 R11 8 x 10KΩ, resistor networkC1, C2 33pF, SMD 0805 C5, C7, 100nF, SMD 0805C8 C3, C6 10µF/16V, SMD tantalumC4 47µF/35V, SMD ElectrolyticB1-B54 push switch DTS-64NJ1 LP patch cord. RM 2.54, 5 pinJ2 LP patch cord. RM 2.54, 2 pinJ3 LP patch cord. RM 2.54, 3 pin1 x 40 pin IC bases 401 x PCB, DG6RBP1 x mini DIN socket, with 6 pinsB3 is not fitted it is for future

extensions!

5.Calibration and operation

Before using the display unit it must becalibrated with the power detector. To dothis the “Up” and “Down” switchesshould be pressed while the power isswitched on. “CAL” then appears on thedisplay. Subsequently, the input voltageof the A/D converter in millivolts isdisplayed. Connect a signal generator tothe detector set to 0dBm (1mW) andpress the “Up” key to store the output ofthe detector as a reference voltage. Thisis confirmed with a short flashing of thedisplay. A level of –20dB (10µW) is nowset on the signal generator and the“Down” key is pressed. The micro con-troller calculates the voltage level and theupward gradient of the detector charac-teristic then stores these. This is con-firmed with a short flash of the display.

The reverse sequence using -20dBm firstthen 0dBm is not possible. An error withcalibration is indicated by a “C” on the

Fig 7 : Picture of the front side of the prototype PCB.


7

display e.g. a completely wrong refer-ence voltage or upward gradient, pointsto an error in the power detector. In thecase of an error the calibration must berepeated again after a reset.

The calibration is only valid for thedetector used and should be repeated atregular intervals (e.g. every 2 years). Thesignal generator used must be as accurateas possible, otherwise the measuring ac-curacy of the wattmeter suffers accord-ingly. The actual output level of thesignal generator should not deviate fromthe adjusted level by more than ±0.01dB.The frequency used for calibration can bebetween 10MHz and 2.4GHz but shouldbe the same for both calibration steps.

It is advisable to determine the frequencyresponse of the detector before calibra-tion and to use a calibration frequencythat corresponds to the average value. Inthe frequency response of the sample(Fig 2) this would be approximately650MHz or 1.3GHz. Thus the measuringerror of the power measuring instrumentover the entire frequency range becomesas symmetrical as possible.

After final calibration the power measur-ing instrument can be used immediately.

The display power is indicated in dBmwith a resolution of 0.1dB (the measuredvalue with 0.02dB resolution is takeninternally and rounded on 0.1dB). Press-ing the key “Abs./Rel.” switches to rela-tive mode, the measured value is storedfrom now as a reference level and thedifference between measured value andreference level in dB is indicated. Press-ing the key again changes the displayback to dBm. This function useful fortransmission measurements.

To measure the gain of an amplifier firstthe input power is measured and the“Abs./Rel.” changed to relative mode.Subsequently measures at the output ofthe amplifier. The instrument now indi-cates the gain directly in dB on thedisplay.

Pressing the “dBm/Watt” key the displayis switched between dBm and Watts. Theappropriate unit is indicated with thelight emitting diodes. The measuredvalue in Watts is calculated from themeasured value in dBm. Here the micro

Fig 8 : Picture of the back side of the prototype PCB.


8

controller uses the maximum resolutionof the sensor (0.02dB from the sensorcircuit with the 12 Bit A/D converter).The display resolution for a linear display(Watt) is 0.02dB. That means, for exam-ple, following 955 Watts (29.80dBm) thenext step is 959 Watts (29.82dBm), orfollowing 50.1mW (17.00dBm) the nextstep 50.4mW (17.02dBm).

The keys “UP” and “Down” can be usedto adjust the offset level. If one of thesekeys is pressed briefly, the current offsetin dB is indicated. An “F” in the firstdisplay place marks the offset. Pressingone of the two keys the offset value canbe changed between –1.9 and +49.9dB. Ifthe key is pressed for a longer time, theadjusting speed is increased by largechanges of the offset level. If neither ofthe two keys is pressed for a long time(approximately 2 seconds), the indicatedoffset level is removed and to the displayof the measured value returns. The offsetvalue remains stored, in EEPROM, afterswitching the equipment off. In order toremind the user after switching on theoffset is indicated briefly in the display.

The offset level is added to each meas-urement. If the power detector is oper-ated, for example, at a frequency wherethe deviation in level amounts to -0.3dB(i.e. the result of measurement would betoo small around 0.3dB), an offset can be

set of +0.3dB for precise measurements.The measuring instrument will automati-cally add 0.3dB to make the measure-ment accurate.

A still more interesting application is themeasurement of high powers using acoupler. Since the power detector canonly measure up to approximately10mW, a coupler can be inserted be-tween the transmitters and the antenna ordummy load. The uncoupled power ismeasured and the coupled attenuation isused as an offset levels to display theactual transmitter power.

The indicated value in Watts or dBm isalso constantly sent over the serial inter-face once it is activated. To activate thetransmission the ASCII value “B”, forbeginning, is sent. In order to terminatethe transmission the value “S”, for stop,is sent. Further interface functions are notimplemented yet.

Printed circuit boards as well as printedfront panels will be available from VHFCommunications.

The front and back of the completed PCBare shown in Fig 7 and 8, the optimisedpower detector is shown in Fig 9 Thecomplete power measuring instrument inthe housing and with power detector is

Fig 9 : Picture of the optimised power detector.


9

shown in Fig 10.

6.Literature

[1] Wolfgang Schneider: Power detectorcovering up to 2.7GHz, VHF Communi-cations Magazine, 3/2006 pp179 - 183

[2] Alexander Meier: Frequency inputmodule for 10GHz the ATV transmittermodule, VHF Communications Maga-zine, 4/2005, pp 217 - 221

[3] Alexander Meier: VX-2-Speicherkomfortabel füllen, CQ-DL 10/04,DARC publishing house, Baunatal(2004)

[4] Data sheet MAX-232, Maxim Inte-grated Products, Inc., 120 San Gabrieldrive, Sunnyvale, CA. 94086, USA,http://www.maxim-ic.com

[5] Bopla Enclosures, Phoenix MecanoLtd, Unit 6/7, Faraday Road, Aylesbury,Buckinghamshire, HP19 8TX,http://www.bopla-enclosures.co.uk/products/enc_tech.htm

Fig 10 : Picture of the completed display unit with power detector.


10

1.Noise

1.1 IntroductionAll materials generate noise and the noiseis proportional to its temperature startingfrom 0°K (-273°C). The noise dependson the chaotic movements of the elec-trons, the thermal noise is known aswhite noise (from optical physics) as itfills the whole spectrum.From an electronics point of view thenoise causes big limitations to our de-vices for example amplifiers, instru-ments, radars, receivers, electro-medical,etc… A very simple example is thesensitivity limitation of receivers causedby the noise.Although I have said that noise causesproblems and limitations, I want to ex-plain how in some cases, if it is artifi-cially generated, it can even improve ourelectronic devices (see dithering in Table1) or help to do some tests, a calibratednoise source is a very important tool inour labs.

1.2 Output levelFor noise source applications the outputlevel cannot be indicated as for othersignal generators. Signal generators,transmitters etc… have the output level

indication in mV, dBm, W etc…. If youhave a 100 to 200MHz sweep signalgenerator we say that the output level is,for example -10dBm, the amplitude of -10dBm is swept from 100 to 200MHzbut it is not simultaneously in the wholefrequency range.In the case of noise sources the amplitudeis simultaneously on the entire frequencyrange, this means that the amplitude isdefined in dBm/Hz power spectral den-sity, or in ENR excess noise ratio. ENRmeans the ratio in decibel of the outputnoise between the ON and OFF state ofthe diode, in the OFF state the diode hasonly -174dBm/Hz which is the outputlevel generated by a resistor at 290°K.For example, if you have a power spec-tral density of -142dBm/Hz it means that(174 - 142 = 32) the ENR is 32 dB. If thebandwidth is 10Hz the noise power is -132dBm/10Hz if the bandwidth is1 0 K H z t h e n o i s e p o w e r i s -102dBm/10KHz.

2.Noise generator diode

2.1 Diode selectionThe first noise generators (in the 1940’s)used noble gas such as Argon with

Franco Rota, I2FHW

Noise source diodes


11

15.3dBENR, Neon with 18.5dBENR,Helium with 21dBENR and were born inorder to test the first radar systems.Another system to generate noise is touse hot and cold resistors, mainly used inresearch labs with very high precision.Zener diodes can be used to generatenoise but the output level is not constant,not predictable and used only for HFfrequencies, even some bipolar transis-tors like BFR34 have been used in thepast for amateur applications using thereverse biased base-emitter diode, theoutput level is definitely not constant.For our applications the right selectionsare:

• NS-301 SMD sod323 case, up to3.5GHz

• NS-303 ceramic gold plated case, upto 10GHz

Both types are silicon avalanche diodesthat provide 30-35dBENR with a broad-band spectrum starting from 10Hz. Inthis article I will focus on the 3.5GHztype and in a second article I will alsodescribe the 10GHz type which is morecomplicated.At the beginning I tested the glass casetype but this case was not suitable be-cause the maximum frequency can bearound 1.5 - 2GHz, for the same price wecan have 3.5GHz with a flatter output

Table 1: Some applications regarding the generation of noise, it can improveelectronic devices or help to do some tests on them. Dithering

In an A/D converter for example digitalreceivers, the noise injected improvesthe quantisation error, the sensitivitywill be improved (this method is alsoused in audio and video).

Spectrum Analyser CalibrationWith a calibrated noise source devicesit is very easy to verify the amplitudecalibration of a spectrum analyser, thereal advantage is the RF generationsimultaneously on all the band.

Noise Figure MeasurementTest instruments for noise figure meas-urement in low noise amplifiers, con-verters, receivers, mixers and front-ends.

Gain-bandwidth measurementsA flat noise source can be used as a“tracking generator” combined with aspectrum analyser to ease measure-ments of gain and bandwidth.

Encryption

Audio And Ultrasonic Test

Test On Receiver The noise is useful to measure thesensitivity in some complex receiverslike radars, base stations, radiometersetc… A noise source can substitute for a morecomplex RF generator, moreover it cangenerate noise in a broad band spec-trum simultaneously.

NPR Distortion This is a complex intermodulationmeasurement very often made on mul-tichannels FDM, MMDS, CATV, cellu-lar base stations, etc.. Injecting noise and measuring the dis-tortion with special notch filters is usedto obtain the measurement.

Fading Simulator By modulating an RF signal with noiseit is possible to simulate a signal af-fected by fading, this is very useful inmobile radio testing.

Radio Astronomy

EMI Testing


12

level.The SMD sod323 case has a very lowseries inductance typically 1 - 1.5nHwhich is reasonable for a 3GHz applica-tion. Fig 2 shows the SMD case sod323,the body is about 1.9mm long, it is usefulfor many applications in the lower micro-wave frequency range.

2.2 Schematic diagramFig 3 shows the circuit diagram of a NS-301 noise source diode up to 3.5 GHz.

C1 – dc blocking capacitorThe selection of this capacitor is ex-tremely important to flatten the outputlevel. I spent much time testing several

Fig 1: The samenoise level relatedto 3 differentbandwidths.

Fig 2 : The SMDSOD323 case.


13

types of capacitors, ATC porcelain ca-pacitors have less insertion loss at micro-wave frequencies but they can’t be usedbecause their Q increase the self reso-nance dip.For this purpose I selected a specialcapacitor case, 0805 COG, which can beused up to 12GHz (about 1.5nF), withthis capacitor the minimum frequency isabout 10MHz.In the next article about the 10GHz noisesource diode I will describe these capaci-tors in more details.For 3GHz application the C1 capacitorisn’t a crucial component, case 0805 or0603 and values form 1nF to 10nF aregood anyway.

C2, C3 – bypass capacitorsThese capacitors are not critical; they canbe 1nF and 10nF.

R1 - RF load resistorsThis resistor is the sum of 3 resistors inseries in order to keep the stray capacityas low as possible, the total value can bearound 30 to 40ΩThe manufacturer of noise diodes saysthat the diode impedance is about 20 to40Ω, I noticed that by assigning to R1 alower resistance (20Ω), the output noiselevel is flatter, on the contrary with anincreased resistance (40Ω) the outputnoise level is a little higher.If possible, it is better to solder theresistors without using copper track on

the PCB.

R2 - bias resistorFor the noise diode NS-301 at about5mA, +8/+12 V, the correct value is3.3KΩ if you use the diode for noisefigure measurements with a classic +28Vpulse available from all the noise figuremeters. If the diode is used as a generalpurpose noise generator to test a filter,for example with a spectrum analyser,you can connect directly to a +8/+12Vdcwithout the R2 resistor.

NS - Noise diodeAs described above the NS-301 sod323diode is a good selection for the 3GHzfrequency range, it is important to re-member to keep the pins as short aspossible! The diode must be mountedvery close to the output connector.

P.C. boardThe FR4 fibreglass p.c. board is ok, theinsertion loss is so little that it isnn’tworth a teflon laminate, vice versa it isvery important the noise diode groundconnection that has to be as short aspossible (see the above explanation).I tested several noise source diodes in mylab with sod323 case, Fig 4 shows thebest and the worst result, in the frequencyrange 10MHz to 3GHz with 2dB/step and300MHz/step, the centre reference levelis 15dBENR and the noise source diodeis connected with a 16dB pad attenuator.

Fig 3 : Circuit diagram for a noise source up to 3.5GHz using an NS-301 noisediode.


14

2.3 Output attenuatorThe purpose of this attenuator is twofold, the first one is to obtain the

15dBENR which is the right noise levelaccepted by a lot on noise figure meters.The output noise of the NS-301 diode isabout 30-35dBENR this means that witha 16dB attenuator you can have about15dBENR. Any other attenuation valuescan be used to get other ENR values.The second and most important purposeof this attenuator is to match the outputimpedance to 50Ω. In noise source de-vices used for noise figure measurement,one of the most important condition is tomatch the output impedance as near aspossible the 50Ω resistive load, the easi-est way is to insert an attenuator to theoutput connector.Normally the ultra low noise GaAsFetpreamplifiers have a very bad input re-turn loss, typically a VSWR from 20 to 2(return loss from 1 to 9.5dB), so if wetest this kind of preamplifier with a noisesource with an high return loss the totalerror is unacceptable.Fig 5 shows a simple explanation of themismatch due to the noise figure meas-urement, we can assume that the pream-plifier input return loss is 3.5dB, SWR =5 (it can seem too high but it is a realistic

Fig 4 : The bestand the worst testresults on differentnoise sourcediodes.

Fig 5 : A simple explanation of themismatch due to the noise figuremeasurement.


15

value).If our noise figure meter measures2dBNF and we assume also that the noisesource output return loss is 23dB (SWR1.15), the true noise figure can be be-tween +0.63dB/ - 0.7dB for a 2dB meas-ured value.In conclusion we should keep the SWRof a noise source as low as possible in

order to do more accurate noise figuremeasurements.Fig 6 shows the instability that it is quitegood for amateur applications, for 8hours of continuous operation it is only0.07dB of output level but there is also a0.03dB of testing instrument instability toconsider.

Fig 6 : Instabilityof noise sourcediode output.

Fig 7 : Responsewith a 45dBamplifier.


16

3.General purpose noisegenerator

As shown in Table 1 a diode noise sourcecan be used successfully in a broadbandnoise generator combined with a spec-trum analyser like a “tracking generator”.This is not a true tracking generatorbecause it works in a different way. As Isaid above the tracking generator is like asweep generator so its frequency movesfrom start to end but it is not simultane-ous in all the frequency range.If we combine a broadband noise genera-tor with a spectrum analyser we can do a

measurement of band pass filters, returnloss etc. The signal coming from thenoise generator diode is very low so weneed at least 45dB of amplification,however 65dB is better. The real diffi-culty is to obtain a reasonable flat ampli-fier response. For this purpose I made anampl i f ie r us ing INA03184 andINA10386 MMICs, the result is shown inFig 7 and the total response is given bythe noise source diode combined with the45dB amplifier.Figs 8 and 9 show the 45dB broadbandamplifier from few MHz to 2.5GHz usedas noise amplifier in order to test the2GHz band pass filter. This circuit is notdifficult to build and it can be used in anylab as general purpose broadband ampli-fier.

Fig 8 : Picture ofthe broadbandamplifier.

Fig 9 : Circuitdiagram of thebroadbandamplifier.


17

Fig 10 shows the dynamic range of atypical 2GHz band pass filter with anoise amplification of 45 dB and 65dB.The dynamic range improves with moreamplification, but it is more difficult toachieve a flat output level.Fig 11 shows the equipment setup usedfor the filter measurement.It is demonstrated that with a simplenoise generator and a good amplifier it ispossible to build an instrument very closeto a tracking generator to use with anykind of spectrum analyser. It means thatwe can “upgrade” an old spectrum ana-

lyser, typically the HP 141 series or anyother type, with an option that works likea tracking generator.NS301 noise source specifications are: Frequency range: 10Hz - 3.5GHzOutput level: 30/35dBENR (-144/-139dBm/Hz)Bias: +8/+12V, 5mAIt is available from R F Elettronica - www.rfmicrowave.it

Fig 10 : Dynamicrange of a 2GHzband pass filter.

Fig 11 : Equipment setup used for the filter measurement.


18

Some time ago the author described a50MHz converter. The positive feed-back and requests to extend this into atransverter have persuaded the authormake this possible. The following arti-cle takes the converter already de-scribed and adds the necessary trans-mitter driver.

1.Concept

The basic circuit simulations presented in[12] pointed out the extensions requiredto the 50MHz converters already de-scribed in [11] to realise a transverter.These extensions are:

• Input levels of the transceiver into theconverter (+6dBm single tone; 0dBmfor two tone rejection for a Signal-to-intermodulation ratio of 60dB)

• Power output of the driver(dependant on the output stage used)

• Power output of the output stage(10W = +40dBm for use with adirectional antenna with a gain of6dB)

• Amplification of the driver(dependent on the output stage used)

• Intermodulation products at full out-put (d3 < -60dB)

• Sideband suppression (< -60dB es-pecially at converter sidebands)

• Harmonic suppression (> 60dB)By choosing the input level in considera-tion to the sidebands (harmonics of the22MHz crystal oscillator and other mix-ing products) the signal-to-intermodula-tion ratio can be kept to a minimum.

The specification did not specify theminimum signal-to-intermodulation ra-tios at the transmitter output howeverthird order products should be less than25dB. Because under unfavourable con-ditions, the intermodulation products addtogether, the stages before the outputshould have more than the required 25dBrejection. This must be considered whenchoosing the devices for the outputstages.

The output stage should be protectedagainst bad connections (e.g. open orshort circuit of the antenna connection)effectively and without affecting theALC of the transceiver. This is achievedusing pin diode control using a circuitthat is activated at a fixed value ofVSWR. A manual control of outputpower is required. The pin diode has anestimated insertion loss of 4dB.

Henning C. Weddig, DC5LV

Extending the 50MHz converterinto a Transverter


19

1.1 Signal-to-intermodulation ratio;Input levelIn transmit mode the converter is oper-ated “backwards”, i.e. the input signal(28MHz) of the transceiver is fed into theIF output of the converter, the 50MHzoutput signal is coupled from the RFinput of the converter.

From the Intercept point of the converter(+30dBm) it can be reckoned back thatwith two input signals of 0dBm at the IFoutput, a signal-to-intermodulation ratioof 60dB is obtained. This signal-to-inter-modulation ratio is high enough. Also thestages of the driver should keep to thislevel. Under these conditions the outputsignal at the RF input of the converterwill be -8dBm because the transmissionloss of the converter is 8dB. Thinkingabout the rejection levels, 6db can beadded, i.e. with an input signal (of thetransceiver) of +6dBm the output signalamounts to -2dBm.

1.2 Transmitting power From [1] a transmitting power of 10Wand a 6dBd directional antenna is suffi-cient, in order to keep to the permittedmaximum output of 25W ERP (41WEIRP).

2.Choice of the transmitteroutput stage

The transmitting power of 10W is easilyattainable with one the remaining Mit-subishi [2] M57735 hybrid modules. Thismodule gives approximately 10W(+40dBm PEP) SSB, using an inputpower of 20mW (13dBm). The amplifi-cation of the module is 27dB. Third orderintermodulation products of 32dB PEP

are 6dB more than related to a singletone, thus “only” 26dB. Thus the inter-modulation products of the output stagedetermine the output signal.

With the first harmonic at 25dB a secondattenuation of 30dB is required in orderto reach the > 60dB harmonic suppres-sion. At least a seven pole low pass filteris required (Cutoff 60MHz with 0.05dBripples in the pass band and around 20dBreturn loss).

The intermodulation values of the hybridmodule are not a problem. If the hybridmodule is selected as an output stage, the50MHz signal output of the convertermust be amplified by about 15dB.

An alternative to the hybrid module asoutput stage is to use discrete transistors.Unfortunately the number of manufactur-ers in the RF power transistor market isvery small. Motorola started the manu-facture of transistors in the portable radiomarket but they have given up, so transis-tors like 2N6080 and 2N6081 are nolonger made. Also the silicon transistorsBLY87 and BLY88 from Philips are notrecommended for new developments.These are possibly still available at fleamarkets. The popular types 2SC1970,2SC1971 and 2SC1972 are no longermanufactured by Mitsubishi so there isnot an alternative.

The trend is clearly toward MOS transis-tors. In [3] and [4] a broadband outputstage with two BLF244 MOS transistorsis described. It reaches a third ordersignal-to-intermodulation ratio of 40dBat 10W output with a supply voltage of28V and 125mA quiescent current. Theamplification of the output stage is ap-proximately 15db. A BLF242 could beused as a driver. Unfortunately thesetransistors are expensive. An alternativecould be the MOS transistor RD16HFF1in a TO220 housings from Mitsubishi,see [5]. This Transistor is relatively inex-pensive.


20

3.Mixing products of theTransverters

If a signal of 0dBm at 28MHz is fed intothe IF output the desired 50MHz signalappears at the RF input as well as othermixing products, the frequencies andlevels are listed in Table 1. The mixingproducts should as far as is technicallypossible be suppressed to a value of>60dB. With the values from this tablethe filter required can be measured. Allmixing products that are a small distanceaway from the carrier are in bold. Thefollowing mixing products are notice-able:

• The first harmonic of the quartzoscillator (44MHz; -43dB)

• The mixing product 4*LO-RF(60MHz; -28dB)

• The mixing product 3*RF-LO(62MHz; -42dB)

• The mixing product 2*LO+2*RF(72MHz; -35dB).

Only the first mixing product is levelindependent at –43dB, the filter wouldhave an attenuation of 17dB.

The mixing product at 60MHz is at –28dB. In order to reach 60dB a filter with32dB attenuation is required. The resultsare about 6dB better if the mixing prod-ucts are measured for PEP rather thansingle tone.

A relatively large range low insertionloss and low noise figure 50MHz filter isrequired at the input of the converter as aband pass filter in the transmit path. Aninsertion loss of <3dB is quite accept-able. A three stage band pass filter wasc o m p u t e d w i t h t h e p r o g r a m“nrbnpas.bas”, having the following data:

• Centre frequency 50MHz • Bandwidth 5MHz • Tschebychev characteristics • Pass band ripples 0.1dB

Since the program “nrbnpas.bas” gavenon-standard capacitor values, the filterwas simulated using “RFSim99”. Unfor-tunately this program does not offer

Table 1: List of mixing products.

Frequency Level Mix product Level of mix productin MHz in dBm with no signal in dBm

16 -81 2*LO - RF 73 22 -65 LO 58 28 -59 RF 51 38 -36 3*LO - RF 28 44 -51 2*LO 4350 -8 IF 0 54 -74 5*LO - 2*RF 68 60 -36 4*LO - RF 28 62 -50 3*RF - LO 42 72 -43 2*LO + 2*RF 3582 -57 5*LO - RF 49 88 -78 4*LO 70 84 -52 3*LO + 3*RF 44

100 -71 2*RF + 2*LO 63 106 -77 3*RF + LO 69


21

automatic optimisation. Therefore the op-timisation of transmission loss to < 3dBand return loss -20dB in the pass bandwas carried out with “Genesys” fromEagle Technology. The non-standard ca-pacitors were replaced where possiblewith normal E12 components or usingseveral capacitors together plus changesin the inductor values. This is whatwould happen with a manual alignment.The optimised filter is shown in Fig 1with the response curves in Figs 2 and 3.At 44MHz the attenuation is 32dB and at60MHz 42dB is achieved. Also the inser-tion loss is kept to 3dB. With thesevalues the filter fulfils the requirement tosuppress the sidebands by >60dB.

4.Level plan

The amount of amplification between theoutput of the converter and the transmitamplifier can be determined from theabove. For a single tone of 4mW(+6dBm) with the transmission loss ofthe converter of 8dB this would give a

signal of –2dBm at 50MHz. To drive thehybrid amplifier fully +13dBm is re-quired so an overall gain for the driver of15dB is necessary.

For two tones the levels per tone must belowered around 6dB, i.e. with two tones -8dBm each at the input of the driver.Considering the attenuation of the bandpass filter of 3dB and the attenuation ofthe pin diode of 4dB the gain required is:15dB + 3dB +4dB = 22dB.

It would seem sensible to arrange theband pass filter after the amplifier stagein order to avoid any reaction betweenthe transverter band pass filter and theconverter band pass filter. On the otherhand operating the band pass filter afterthe amplifier with relatively high gain,non-linearity due to the ferrite cores usedin the coils can contribute to intermodu-lation.

The amplifier stage could use an RF2360IC like the one used in the LNA (G =20dB; OPIP3 = +37dBm, IPIP3 =+17dBm; P1dB = +25dBm). It would beoperated 1dB below the compressionpoint (-2dBm + 20dB = +18dBm), how-ever it is 38dB from the third order

Fig 1: The circuit diagram of the optimised band pass filter.


22

intermodulation point. Thus the RF2360would severely degrade the intermodula-tion behaviour of the transverter. Theamplifiers type AH3 and AH101 fromWatkin Johnson are therefore used. Thedata sheets are downloadable from themanufacturer, see [7] and [8]. The rel-evant data from the data sheet is:

AH3

• G = 12dB min., (12.9dB typical) • OPIP3 = 37dBm min., (+41dBm

typical) • P1dB = +27dBm • Noise figure (NF) = 4.5dB (typical) • S11 –10dB, S22 –20dB • Supply: 5V; 120 - 180mA

The value of (S11) is not acceptable, itmust be optimised using a narrow bandfilter to <-20dB.

AH101

• G = 12dB min. (13.5dB typical) • OPIP3 = 43dBm min. (+47dBm typi-

cal)

• P1dB = +21dBm • Noise figure (NF) = 3.5dB (typical) • S11 –20dB; S22 –15dB • Supply: 9V; 170 - 230mA

The input matching for this IC is accept-able but some improvement could bemade to give the ideal value of <-20dB.

Both amplifiers together result in a mini-mum gain of 12dB + 12dB = 24dB,typically 26.4dB which is sufficient toachieve the level plan.

5.Test of filter construction

A test of the three-stage filter in a tinplatehousing with the AH3 amplifier on aprinted circuit board showed that the44MHz signal is reduced still further.Therefore a further three-stage filter isplanned at the output of the second

Fig 2: Frequency response of the optimised band pass filter.


23

amplifier. The additional attenuation ofanother filter can be tolerated, since thereis sufficient gain in reserve.

6.System simulation

The result of the system simulation canbe seen in Fig 4. The power outputs are

Fig 3: Damping factor of the optimised band pass filter.

Fig 4: The results of the simulation.


24

appropriate for a single tone during atwo-tone test. There are approximately2dB in reserve, (G = 16.9dB, 15dB arerequired). The cable attenuation could beovercome if the output stage is at thebase of the antenna like the LNA. Therequired intermodulation values areachieved.

7.Circuit description

The circuit diagram of the transverterdriver is shown in Fig 5 and 6. The pindiode PCB is taken from the circuit in[6]. The attenuation is greater than 70dB,

Fig 5: Circuit diagram of the transverter driver.


25

the control characteristic is howeverstrongly nonlinear. The attenuation startswith less than 8.9V at the output of IC3b,pin 7. At 3.5V the attenuation is 10dBand 1.93V gives 20dB.

The two operational amplifiers in IC3 areused as a buffer, amplifier and an inverterfor the control signal which can besupplied externally. The control level canbe changed with R11. In the prototypethe control voltage is 10.93V with regu-lation starting at 0.8V (at St1).

Tests on the finished circuit showed thatthe regulation start point became sensi-

tive to interference because of the highDC voltage amplification of IC3b. There-fore D4, R10 and R12 were changed onlater units, the values were determinedexperimentally. The control characteristicis shown in Fig 7. It is linear within therange from 0 to 20 dB, it could be mademore linear by adding additional diodesand resistances. Unlike the AGC in areceiver, this application only uses asmall range (0 to 20dB) of the controlcharacteristic therefore this is acceptable.

The AH3 amplifier does not have goodoutput matching (S11; S22 < -20dB) at50MHz, a matching circuit was com-

Fig 6: Circuit diagram of the transverter driver.


26

puted with GENESIS.

Unfortunately the downloadable S pa-rameters from the manufacturer, WatkinJohnsn, only start from 50MHz, thereforeapproximate S parameters were used forGENESYS. The simulation results areshown in Fig 8 and 9. There are discrep-ancies between simulation and measure-ment. Differences can be explained bythe fact that the manufacturer does notprovide the measuring conditions, underwhich the S parameters are provided.

The AH101 has a good input match at50MHz (S11 = -20dB), so that only theoutput needs to be matched. AlthoughS12 is quite small, the input matching

gets worse if the output matching isimproved. Finally no matching was used.The results of measurement are shown inFig 10.

Supply voltages of +5V for the AH3 and+9V for the AH101 are generated usingregulators from the 12V system supplyvoltage. The voltage regulators aremounted directly on the printed circuitboard, in order to dissipate the heat.

The chokes planned (1µH, size 1206,from Stettner) for supply voltage decou-pling have a high DC resistance andtherefore become too hot with the work-ing currents (150mA and 200mA). Fur-ther the voltage drop across these chokes

Fig 7: Theregulation of thePIN diodeattenuator.

Fig 8: Thesimulation resultsfor the AH3amplifier.


27

appeared too high. Therefore chokesfrom Coilcraft, type 1008 were used, theyhave a DC resistance of 1.75 ohms and amaximum permissible direct current of350mA. The voltage drop is V = 1.75 *150mA = 0.265V, in the prototype avoltage drop was measured as 0.45V, it ishowever higher for the lower supplyvoltage of 5V. Therefore a choke with a0805 ferrite core was used in the finalversion, type B82498-B1102 from Epcos.The DC resistance is a maximum of 0.5,with a self resonance of 350MHz whichis far enough away from the workingfrequency, see also [10].

Both amplifiers have a relatively highpower dissipation (AH3: 0.7W; AH1011.8W) so it is not sufficient to use thesurface of the printed circuit board todissipate the heat. In [9] it is pointed outthat for a long life span good heatdissipation must be ensured. That can bedone on the printed circuit board using

plated through holes around the earth pinof the ICs to transfer the heat to a 10mmwide and 20mm high aluminium coolingblock fastened with M2.5 screws to thelower surface of the printed circuit board.

With a coax relay at the input of thedriver the signal from the LNAs is fed tothe mixer on receive. The relay is acti-vated on transmit using the PTT signaland the signal to the driver is mixed upby 28MHz to 50MHz. A further relay atthe output stage feeds the signal to theantenna.

8.Construction and alignment

The main double sided PCB is made

Fig 9: Results ofmeasurements onthe AH3 amplifier.


28

from 1mm thick FR4 material with platedthrough holes. The size of the board waschosen to fit into a standard 55mm x111mm tin plate housing.

Fig 11 and 12 show the layout of theprinted circuit board of the transverter forthe SMD side and the component side.The layouts for the SMD side and com-ponent side are shown in Fig 13 and 14.Photographs of the prototype are shown

in Fig 15 and 16. It can be seen that somechanges were necessary for the finalversion that are shown on the layouts.

Construction is best done in stages. Firstthe pin diode PCB with IC3 and thevoltage regulator IC2 is constructed andtested. Next the two band pass filters canbe constructed and adjusted. The authorused an 8753B network analyser to adjustthe transfer function (S21) and the input

Fig 10: Results ofmeasurements onthe AH101amplifier.

Fig 11: PCB layoutfor the SMD side.


29

reflection (S11).

It is possible to adjust the filters forlowest attenuation in the pass band. As areference the filter characteristics of theprototype are shown in Fig 17 for thefrequency range 20kHz to 100MHz andin Fig 18 for the frequency range 50MHz±5MHz).

Finally the two amplifier stages are con-structed. It should be noted that the earthsurface under the ICs is in good thermalcontact with the PCB because this is theway heat is dissipated. The ICs becomesperceptibly warm despite the aluminiumheat sinks.

After a long test the AH3 amplifier failed

which was discovered by measurementsvia the 1µH drain choke. The failurehappened because the earth surface wasinsufficiently soldered at the edges. Pos-sibly the earth surface should be coatedwith SMD flux before ICs are fitted.Then fit the ICs by soldering the drainand gate connections and finally heat theearth surface by heating up the under sideof the PCB with the soldering iron untilthe solder melts. An alternative would beto feed solder through the plated throughholes form the underside of the PCB.

The AH3 dissipates 750mW (5V x150mA) and the AH101 dissipates 1.8W(9V x 200mA).

Continued on Page 35

Fig 12: PCB layoutfor the componentside.

Fig 13: Component layout for the SMD side of the PCB.


30

Index of Volume 38 (2006)Article Author Edition Pages

6m Band

Ultra linear low noise Dragoslav Dobricic, YU1AW 2006/3 184 - 189preamplifier for 6m

Antenna Technology

Design for a printed circuit Thomas Bergmann, DG8NTB 2006/1 48 - 61board antenna, using a Log and Johannes Schad, DG6NDS Periodic as an example

Reciprocal effects between Prof. Ing. Gerd Janzen, DF6SJ 2006/2 79 - 88antennas and surrounding metal objects, Part 1

Design of a Quad Yagi: Part 1 Johannes Schad, DG6NDS 2006/2 107- 116

Design of a Quad Yagi: Part 2 Johannes Schad, DG6NDS 2006/3 145- 155

Reciprocal effects between Prof. Ing. Gerd Janzen, DF6SJ 2006/3 156 - 168antennas and surrounding metal objects, Part 2

Filters

Design and assembly of a Wolfgang Schneider, DJ8ES 2006/1 2 - 6low pass filters

Practical Project: Stripline Gunthard Kraus, DG8GB 2006/1 7 - 28low pass filter for various frequency ranges. Part 3 cont. from 3/2005

Fundamentals

Correction to the article "An Gunthard Kraus, DG8GB 2006/1 28interesting program: Circuit simulation using PSPICE" in issue 4/2005

Determining S-parameters Gunthard Kraus, DG8GB 2006/2 95 - 106with PSPICE

The Noble Art of piping DC Carl Lodström, SM6MOM 2006/2 117 - 123to the LNA and KQ6AX


31

An interesting program: Gunthard Kraus, DG8GB 2006/3 130 - 144ANSOFT designer SV 2.2

Diode multipliers John Fielding, ZS5JF 2006/3 169 - 178

Transistor frequency John Fielding, ZS5JF 2006/4 224 - 237multipliers

An interesting program: Gunthard Kraus, DG8GB 2006/4 238 - 250ANSOFT designer SV 2.2 Part 2 cont. from 3/2006

Measuring Technology

Selective wattmeter Wolfgang Schneider, DJ8ES 2006/2 72 - 78

1.3GHz prescaler Zeljko Bozic, S52ZB 2006/2 89 - 94

Power detector covering Wolfgang Schneider, DJ8ES 2006/3 179 - 183up to 2.7GHz

Microwave milliwattmeter Bernd Kaa, DG4RBF 2006/4 198 - 211to 18GHz, WM 500 PRO (-55dBm - +30dBm)

Miscellaneous

Internet Treasure Trove Gunthard Kraus, DG8GB 2006/1 29 - 30

12mm DX in New Zealand Ted Barnes, ZL2IP 2006/2 124March 2006




Optical Band

The noble art of measuring Carl Lodström, SM6MOM, 2006/1 35 - 46optical power KQ6AX

Oscillators

Down Converter for YIG Alexander Meier, DG6RBP 2006/2 66 - 71

VCOs as a replacement for Wolfgang Schneider, DJ8ES 2006/4 194 - 197YIG oscillators in the 2 to 4GHz range


32

A complete index for VHF Communications Magazine from 1969 to the currentissue is available on the VHF Communications Magazine web site -http://www.vhfcomm.co.uk. The index can be searched on line or downloaded inpdf of Excel format so that it can be printed or searched on your own PC. If you arenot connected to The Internet you can write to or fax K.M. publications for a printedcopy of the index which will cost £2.50 plus postage.

K.M.Publications, 63 Ringwood Road, Luton, Beds, LU2 7BG, UK.

Fax: +44 (0)1582 581051

Email: [email protected]

Why not keep your magazines tidyVHF Communications Magazine Blue Binders are available to store yourmagazines safe and tidy. Each Blue Binder holds 12 magazines.The magazines are held in the binder using a wire clipped into slots at the top andbottom of the binder and positioned in the centre page of the magazine. Thismeans that individual magazines can be added or removed easily.The Blue Binders are £6.50 each plus postage. Postage is £1.40 in The UK, £1.60for surface mail, £2.00 for air mail to Europe and £3.60 for air mail outsideEurope.You can order Blue Binders using the order form on the web site -www.vhfcomm.co.uk, by fax to +44 (0)1582 581051 or by post to K MPublications, 63 Ringwood Road, Luton, Beds, LU2 7BG, UK


33

Year Q1 Q2 Q3 Q42006 Y Y Y Y2005 Y Y Y Y2004 Y Y Y Y2003 Y Y Y Y2002 Y Y Y Y2001 Y Y Y Y2000 Y Y Y Y1999 Y Y Y P1998 Y Y Y Y1997 Y Y Y Y1996 Y Y Y Y1995 Y Y Y Y1994 Y Y Y Y1993 Y P P P1992 Y Y Y Y1991 P P P P1990 P P L P1989 P P P P1988 P P P L1987 P P P P1986 P Y Y Y1985 Y Y Y Y1984 Y Y Y Y1983 P P Y Y1982 P P P P1981 P L P P1980 P P P P1979 L P P P1978 P P P P1977 P P P P1976 P P P P1975 P P L P1974 L P P P1973 P P P P1972 P P P P1971 P P P P1970 P P P P1969 P P P P

VHF Communications Back IssuesAvailable

Key: Y = Available L = Low quantity P = Photocopy The back issue sets contain issues marked Y. Issues marked L or P must be ordered separately.

Single issues from 1969 to 2005, £1.00 each plus postage

Single issues from 2006 volume, £4.70 each plus postage

Complete 2006 volume, £18.60 plus postage

Back issue set up to 2005 (65 magazines), £45.00 plus postage

Back issue set up to 2006 (69 magazines), £65.00 plus postage

K M Publications, 63 Ringwood Road, Luton, Beds, LU2 7BG, UK.Tel / Fax +44 (0)1582 581051 Web site http://www.vhfcomm.co.uk


34

It is particularly important that there isgood thermal contact of the AH101 to theearth surface of the printed circuit board.

9.Measurement results

The frequency response of the completedriver is sown in Fig 19. The inputadjustment is sufficient in the usablefrequency range. It gets worse up toapproximately –10dB for attenuation val-

ues for the pin diode around 10dB. Theoverall gain amounts to 16dB, which isabout 1dB more than the level plan. At44MHz an attenuation of 60dB is ob-tained. At 60MHz it is around 80dB (themix products from the 50MHz converteron this frequency are attenuated by only29dB according to Table 1).

The intermodulation behaviour with a –7.6dBm two tone signal is shown in Fig20. A signal-to-intermodulation ratio ofthird order products of –54dB is ob-tained. The signal-to-intermodulationratios as function of the input power inthe range -10dBm/Tone to 0dBm/Tone isshown in Fig 21. The -8dBm/Tone

Fig 14: Component layout for the component side of the PCB.

Fig 15: Picture ofthe component sideof the completedPCB.


35

signal-to-intermodulation ratio for thedriver is 60dB which means that theoriginal specification is fulfilled. With

two input signals of 0dBm at the input ofthe transverter, the output obtained is16dBm/Tone = 22dBm PEP. Considering

Fig 17: Filtercharacteristics ofthe prototype forthe frequencyrange 20kHz to100MHz.

Fig 16: Picture of the SMD side of the completed PCB.


36

the attenuation of the band pass filter, thesecond amplifier supplies a power outputof 25dBm PEP and therefore is operated1dB below the compression point. Thedegradation of the signal-to-intermodula-tion ratio is explained because to datasheet shows for the output intercept pointwith two output signals of over +5dBmare measured with a frequency spacing of10MHz.

10.Parts List

IC1 AH3 SOT 89, Watkin JohnsonIC2 7805 TO220, National

SemiconductorsIC3 LM358 SO8 TI, National

SemiconductorsIC4 7809 TO220, National

SemiconductorsIC5 AH101 SOT 89 Watkin

JohnsonD1, D2 HSMP 4810 AgilentD3 LL4148D4 BAS 16 SOT 23 InfineonR1, R4 560Ω SMD 0805R2 330Ω SMD 0805R3, R5 1k5Ω SMD 0805R6 680Ω SMD 0805R7, R15, 0Ω SMD 0805R18 R8, 9, 10kΩ SMD 0805R12, R14, 16, 10kΩ SMD 0805R17R10 22kΩ SMD 0805R11 1KΩ 75 R VishayR13 100kΩ SMD 0805C1, 2, 4, 10nF SMD 0805C5

Fig 18: Filtercharacteristics ofthe prototype forthe frequencyrange 50MHz ±5MHz.


37

Fig 20: Theintermodulationproducts of thetransverter with aninput of -7.6dBm/tone.

Fig 19: Thefrequency responseof the completetransverter.


38

C10, 22, 10nF SMD 0805C25C3, 11, 100nF SMD 0805C12C6 12pF SMD 0805C7, 15, 10µF/16V KemetC24, 33 C8 56pF SMD 0805C9, 19, 100pF SMD 0805C36C26-C28 100pF SMD 0805C14, 21 22pF SMD 0805C29, 35 C16, C31 330pF SMD 0805C17, C32 18pF SMD 0805C20, C34 1pF SMD 0805C23 1000pF SMD 0805C30, C37 10pF SMD 0805L1 - 3, 1µH SMD 1206 StettnerL8, 14 L4 - 6, 605nH, 5036 NeosidL10, 12 L7 150nH SMD 1206 StettnerL9 120nH SMD 1206 StettnerL13, L15 1µH B82498-B1102 EpcosBu1 - 3 SMA right angle socketRel1 RK 12, MatsushitaSt1 - St5 1.3mm solder point

11.References

[1] The Charm of the 6m Band (1),special operating techniques, MartinSteyer DK7ZB, Funkamateur 3/2000, pp299 - 301

[2] M57735 data sheet, Mitsubishi Semi-conductors data book 1989, pp 2-82 to 2-87

[3] A wideband 30W push pull amplifierwith two BLF244 MOS transistor (VDS= 28V); 25 – 110MHz application noteNCO 8701, Philips Semiconductors,www. Semiconductors.Philips.com

[4] Performance of a 30W push pullamplifier for the frequency range 25 -110MHz with 2 BLF244 MOS transis-tors, Application note NCO 8702, PhilipsSemiconductors, www. Semiconductors.Philips.com

[5] RD16HFF1 data sheet, Mitsubishi,www.mitsubishichips.com

Fig 21: The signal-to-intermodulationratios as functionof the input powerin the range -10dBm/Tone to

0dBm/Tone.


39

[6] A Low cost Surface Mount pin diodeAttenuator, Application note 1048, Agi-lent Technologies

[7] Data sheet AH3, Watkin Johnson,October 2003

[8] Data sheet AH101, Watkin Johnson,June 2003

[9] AH1 Thermal Resistance analysis,AH2, AH3, FH1, & FH101 Included bySimilarity, Application note, WatkinJohnson

[10] SMT Inductors SIMID SeriesB82498-B data sheet, Epcos 04/00

[11] A modern 50/28 MHz, HenningChristof Weddig DK5LV, VHF Commu-nications Magazine 2/2004 pp 95 – 115and 4/2004 pp 23 – 248

[12] Design of a low-noise and largesignal 50MHz transverter, HenningChristof Weddig, Conference booklet ofthe Munich amateur radio conference2004, pp 37 - 54


40

It has always been the desire everyradio amateur to increase the range ofhis or her station as simply and inex-pensively as possible. Nothing changes,therefore in this article the same ap-plies to the computer age WLANrange.

1.Preface

Connection of computers into a networkand to The Internet using a wireless“Access Point” usually uses around2.45GHz. Radio amateurs use this tech-niques to make contact and try to in-crease the range within the legal limits.An alternative is to use directional anten-nas that concentrate the radiated power ina certain direction. They can be used forboth the transmit and receive side withinteresting possibilities:

• Higher receiver antenna signal overthe same distance and thus bettertransmission quality (e.g. expressedas SINAD improvement) or a greaterrange for the same received signal.

• No more all round radiation thusreducing unwanted propagation

and/or directions of the signal.

• Mechanisms for practical bug proofpoint-to-point connections.

At 2.45GHz the wavelength is small(122.45mm in air) and the half wave-length, important for many antenna de-signs is six centimetres. Therefore quiteuseful antennas can be designed and thepatch antenna was chosen for this design.

This antenna is:

• Extremely simply and durable, be-cause it consists of just a doublesided printed circuit board.

• It can be designed in such a way thatit is matched to 50Ω therefore it doesnot need an additional matching net-work.

• It shields against backward radiationand therefore offers a higher gain andan excellent front to back ratio whichis half the radiation pattern of adipole.

• It has a maximum gain of 6.5dBiwhich doubles the receive signal inthe forward direction compared to anisotropic antenna. Compared to adipole or vertical antenna there isabout 4.5dBi of gain.

Gunthard Kraus, DG8GB

Practical project: Durable andreproducible antennas for the2.45GHz WLAN band, Part 1


41

If the transmitter uses a similar antennathe overall gain of the connection isapproximately 9dB over the original an-tennas and somewhat more than the“three-way” signal. (WLAN range meas-urements are often quoted for a three-way connection i.e. three connected PCs– ed.). If this thought process is extendedfor the far field, the received signal isgreater in a greater distance therefore the“three-way” range can be increased.

Naturally there are some points to beconsidered:

The patch antenna is narrow band with aradiation pattern in form of half an eightfor a single patch and is ideally operatedon a single frequency. Therefore theperformance at the band edges must beconsidered. During the design thickermaterial and larger width patches can beused to give a wider band. In practice thiswill probably reduce the three-way range.

The requirements of the printed circuitboard are very high. Apart from absolutemechanical stability and small losses atthe operating frequency, small tempera-ture and humidity dependence of allelectrical parameters are demanded. Ad-

ditionally the manufacturing accuracymust be kept within 0.01mm. On thepositive side, if everything is under con-trol, the finished antenna is very durableand can withstand falling to the groundwithout damage. The antenna can bereproduced easily, even for large produc-tion volumes.

Some articles have already appearedabout the function and design of patchantennas for different frequency ranges,e.g. [1], [2] or [3]. Therefore the intro-duction is to be kept very short andaimed directly at the practical project.

Not only will the design procedure andthe necessary steps up to productionstage be considered but an importantcriteria is not to use expensive fullversions of the EM Simulator programs.This discourages anyone from using pi-rate copies of these applications. There-fore one tries to get along with programsthat can be obtained free of charge fromthe Internet, these have things missingthat are replaced with an imaginativeapproach.

Fig 1: Arectangular patchantenna with a 50Ωfeed.


42

2.Defaults and productspecifications

Three different forms of single patchantennas are to be designed, built andtested.

The size of the printed circuit boards is tobe about 80mm x 60mm, fed by an SMAsocket.

The resonant frequency is 2450MHz,with a bandwidth of 60MHz required toincorporate the whole frequency rangefrom 2420 to 2480MHz.

The PCB material used is Rogers R04003with its excellent mechanical stabilityand workability. It has a thickness of1.52mm (= 60 MIL), 35µm copper coat-ing, a dielectric constant of 3.38 as wellas a dissipation factor of 0.001 at 2GHz.

A more detailed look at the finishedproducts:

• A rectangular patch with a 50Ω “In-sert feed” is shown in Fig 1. The startpoint of the feeder on the face of thepatch is chosen to give an inputimpedance of 50Ω. A semiridgidcable with an SMA plug is used as atemporary connection, it is better touse a surface mounted SMA plugsoldered to the lower surface of thePCB with the centre conductor sol-dered to the stripline.

• Fig 2 shows the two other antennadesigns. These are a rectangular anda round Patch. The feed is madewithin the patch face by an SMAsocket, which is soldered on thelower surface. Both versions arehandy and durable but the roundversion might prove better in hardeveryday life because of the missingcorners and edges. This is at the costof a narrower bandwidth. In additionit must be connected to the transmit-ter and receiver with more care inorder to align the polarisation cor-rectly. Even for specialist this is a bitunusual.

Fig 2: On the left is a rectangular patch antenna with an SMA connector forthe feed. On the right is a circular patch antenna with an SMA connector forthe feed.


43

3.Development of the rectangularpatch antenna

To start, the simple rectangular antennais chosen, fed from the lower surface.There is a free design program availableon The Internet called “patch16”. It willbe interesting to test how it compareswith a genuine high speed EM simulatorand to manufacture the first PCB afterthe simulation. This will give an idea ofthe differences that can be expected.

3.1. Design from the program“patch16”This program is loaded, installed andstarted from the Web onto your compu-ter, it runs perfectly under Windows XP.When greeted with the message “Do youneed Instructions” answer “no”. The fol-lowing steps are shown in Fig 3, choose“Auto design mode –A” and use thefollowing values for the questions:

• Resonant frequency: 2.45GHz

• Range: 60MHz

• Dielectric constant: 3.38

• Dissipation factor: 0.001(Loss tangent)

After these values have been entered, thescreen shown in Fig 3 is displayed, butthere is a malicious trap to consider (Fig4). The program has chosen a PCBthickness of 88 MIL (because of thesquare patch format), in order to give thedesired range, however a thickness of 60MIL was required. As long the squareformat is chosen, the program cannotchange the calculated PCB thicknesstherefore another way must be found.

Consider the following:

• A larger range for a patch antennameans a thicker PCB or a biggerpatch width with the same patchlength (additionally a high dielectricconstant helps).

• So the patch width must be increased(with a constant PCB thickness of 60MIL) until the required range of60MHz is reached. If necessary thepatch length must be changed inorder to give again the desired reso-nant frequency of 2450MHz. Thatcan be done as follows:

Note the suggested values for the length(1.24 inch), the width (also 1.24 inch) aswell as the feed point (0.45 inch) andpress the letter “D” on the keyboard.

Fig 3: Theintroduction screenfor "patch16".Select option "A"for Auto-DesignMode.


44

Enter these values including the data ofthe PCB material R04003. Once this isdone the screen will be like Fig 4.Pressing “C” will calculate the patchparameters including the transformationline if required.

Now increase the patch width to get therequired bandwidth of 60MHz. Afteroptimisation the final parameters areshown in Fig 5 with the required resonantfrequency and bandwidth. The final stateafter successful optimisation is to be seenin Fig 5, now tuned for both the resonantfrequency and the range. In addition theradiation resistance for the patch edgeshas been computed. Looking at the mainscreen shown in Fig 6 the PCB thicknessis now 60 MIL. The secret for doing thisis revealed.

If “E” is pressed all of the designparameters are available to be edited asshown in Fig 7 and can be changed using

the small menu. The patch length, patchwidth and feed point can be changed,always return to the main menu and press“C” to calculate the results. Finally theresults shown in Fig 5 can be achieved.This does not take as long as it sounds.

All important data is summarised infollowing table:

• Resonant frequency: 2.450GHz

• Bandwidth: 60MHzapproximately

• Gain: 6.5dBi

• Radiation resistance: 116.88ΩAt patch edge

• Patch length: 1.2595 inches = 31.99mm

• Patch width: 2.2 inches= 55.88mm

Fig 4: The Auto-Design mode givesa PCB that is toothick. Use option"D" to change this.

Fig 5: The resultsof the "C"calculate optionshows the resultsfor the patchantenna.


45

• Distance of the feed: 0.34inchespoint from the front = 8.64mmpatch edge

• Wavelength in air 122.45mmat 2450MHz

This basic information is enough to beable to analyse the design with a goodEM simulator. Currently the best free ofcharge simulator is “Sonnet Lite”.

3.2. Preparation for the antennadesign with Sonnet LiteThe software can be downloaded andinstalled from the Sonnet homepage(www.sonnetusa.com). When the soft-ware is started only six buttons appear onthe task bar. Unfortunately the Farfieldviewer is missing which is a shame. Toovercome this select the Admin menuand choose the “Register Sonnet Lite”option as shown in Fig 8. Follow the

prompts and increase the availablememory from 1 to 16Mb which willincrease the number of available taskssignificantly - and all free of charge!

Now the simulation can be started byselecting “Edit Project” then “New Ge-ometry”. After the editor is loaded, firststore the project in a suitable place with asuitable name. Then carefully carry outthe following steps:

Step 1: In the “Circuit” menu, choose the“units” option. Ensure that millimetresand Gigahertz are selected for changes tophysical values as shown in Fig 9.

Step 2: Again under “Circuit” select the“Dielectric Layers” option shown in Fig10. The PCB for the antenna is on thelower line and above it is “Layer 0”which is used by Sonnet as an air layerwith a thickness of half a wavelength

Fig 6: The PCBthickness is nowcorrect.

Fig 7: The "E"(Edit) option goesto this menu wheremany parameterscan be changed.


46

(61.22mm). First click on the lower lineto highlight the line. Use “Edit” tochange the data for the RO4003 materialas shown in Fig 11. Then select the topline and edit the air layer with thefollowing data (Thickness = 61.22mm,Erel = 1.0, Dielectric loss tangent = 0.0,Dielectric Conductivity = 0.0). Thescreen should then look like Fig 10. Themagnetic characteristics (Mrel = 1.0,Magnetic loss tan = 0.0) take over!

Step 3: Again under “Circuit “ select“Metal Type”, at the start there is onlyone entry for “lossless”. Therefore clickon “ADD” to open the metal editor. Clickon “SELECT metal from LIBRARY”and then “global LIBRARY” whichdisplays a metal selection menu. Choose“copper” to get the screen shown in Fig12 and ensure that the thickness is ad-justed to 0.035mm, then click “OK” toget to the menu shown in Fig 13. Change

Fig 8: Use thismenu to registerSonnet Lite andincrease theavailable memoryfrom 1 to 16Mb.

Fig 9: Use thismenu to change thedimensions to mmand frequency toGHz.


47

the “Metal for New Polygons” to copperand confirm with “OK”. All of the basicadjustments are now made and the an-tenna design can begin.

3.3. Antenna design with Sonnet LiteYou should be familiar with the represen-tation of the antenna outline by coordi-nates related to the lower left hand corneras shown in Fig 14 because this is theconvention that the Sonnet editor uses.The length (31.99mm) and width(55.88mm) are taken from 3.1. The feed

point is on the centre line and 8.64mmfrom the lower patch edge.

Now the most important work starts:

The antenna design must be broken downin small elements for the simulation, thisis limited to the point where the memorylimit of 16Mb is reached.

Experience must be used to decide on thebest compromise, this indicates that thelength and width of the smallest cell (Δxand Δy) are approximately 1% to 2% of

Fig 10: Set thecorrect values forthe PCB material(R04003) and theair layer.

Fig 11: Use the editmenu to set theparameters for theR04003 PCBmaterial.


48

the wavelength. The pain threshold forrapidly rising simulation inaccuracies isup to approximately 5% and down toapproximately 0.4%. In addition themanual suggests that the antenna edgesshould be at least one wavelength fromthe housing wall (Sonnet simulates eve-rything in a rectangular metal box), inorder to keep disturbing influences of thebox wall small. But, the larger the betterhowever the 16Mb memory limit sets theultimate limit.

The air layer above the PCB is half a

wavelength, the cover of the box formsthe free space area where the radiationbegins. That was taken into account inthe Dielectric Layers setup.

The definition of Δy is for the patchlength. It corresponds about the half awavelength (50% of λ) and thereforeonly needs the length to be divided by 50to be 1% of the wavelength (in the PCBmaterial, not in air!):

Δy = 31.99mm/50 = 0.6398mm

Fig 12: Use thismenu to set thecorrect values forthe metalthickness.

Fig 13: Use thismenu to set copperto the "Metal forNew Polygons".


49

If a wavelength in air of 122.45mm isadded to the patch length on both sidesthe calculation becomes:

122.45mm + 122.45mm + 31.99mm =276.89mm

And then:

276.89 mm/0.6398 mm = 433 cells inthe y direction.

The patch width can use the same cellsize, Δx, as Δy. However further thoughtis needed because the next step in theantenna design is to add the feed. This isplaced on the patch face to match to 50Ω.The width of this is determined using asuitable microstrip calculator or micro-

wave CAD program (PUFF, Ansoft de-signer SV, Appcad etc.) using the avail-able PCB data. All programs agree avalue of 3.48mm, since no housing withcovers is required (and PUFF or Appcadare not able to include the option "coverheight" in their simulation).

Therefore strip line width is used to setthe cell size and get the best from theantenna simulation

Thus simply divide the width value of3.48mm into four equal parts to obtain:

Δx = 3.48mm/4 = 0.87mmThat is about 1.36% of the PCB wave-length and thus everything is in the greenrange for the Sonnet simulation. To de-termine the number of cells in the xdirection:

The patch width is 55.88mm. On eachside, add a wavelength of 122.45mm toget:

122.45mm + 122.45mm + 55.88mm =300.7 mm

That results in:

300.78mm/0.87mm = 346 Cells in the xdirection.

Fig 14: You should be familiar withthe way that Sonnet Lite usescoordinates by making a sketchshowing magnitude and coordinates.

Fig 15: These arethe "Box Settings"for Sonnet Lite tooperate with. Thex and y cell detailsare shown as wellas the "Top Metal"set to "FreeSpace"and the"Bottom Metal"set to "Copper".


50

The associated “box” parameters for theSonnet circuit menu are shown in Fig 15.

• The line “Cell Size” is set to Δx andΔy with the “Lock” box checked

• The line “Num. Cells” is set to thecorrect number of cells for x and y

• “Top Metal” is set to “Free Space”

• “Bottom Metal” is set to “Copper”If the cell size Δx = 0.87mm is used thereis a problem because the patch width of55.88mm is not accurately calculated.This is not so tragic because a differencein the patch width up to 1mm does not

make much difference but a difference of0.1mm in the feeder width makes a bigdifference in the feeder impedance.

The simulation can be started by drawingthe antenna but before that select “Meas-uring Tool” on the “View” menu andclick the mouse symbol as shown in Fig16. The cursor pointer changes to a smallcross which is the zero point or lower leftcorner of the patch, it can, for example,be set on the centre of the displaysurface. The patch can now be drawn byclicking on “Draw Rectangle” and pull-ing with the mouse. The sizes of the twovalues Δx and Δy can be seen in“Measuring Tool”, these should be set to;

Δx = 55.68mm and Δy = 31.99mm.The display surface can be zoomed tomake this easier. The patch area shouldbe green, which means copper metallisa-tion. If this was not set in “Metal Type”and “Material for New Polygons” it canbe set by right clicking the mouse on thepatch area and using “Metal Properties”as shown in Fig 17.

Now set a “Via” (plated through holeinside diameter = 1.27mm) at the feed-point (27.84mm, 8.9572mm). The “Via”should be copper and will show in the

Fig16: The easiest way to set the zeropoint is to us the mouse (see text).

Fig 17: All of theinformation aboutthe patch is shownhere.


51

same green colour as the patch area.Sometimes that is not the case so the Viamust be marked by a right hand mouse-click and then the property in the appear-ing menu changed to “Copper” (Fig. 18).At last a Port must be set exactly at theposition of the Via. Then switch Layer to“Ground”, mark there the Port by adouble-click an choose the property“Standard” in the Port Property Menu.

Now the drawing work is completed!(Fig.19).Things are fast from now on. In the“Analysis” menu there is a Setup for thesweep. Fig. 20 shows the attributes nec-essary for “Adaptive Bandsweep” in thefrequency range from 2.4 to 2.5 GHz.Fig. 21 shows the Simulation Resultsfrom clicking “Project”, then “Analyze”after waiting for the computing time (if

Fig 18: Setting aVia, it must be setto Copper.

Fig 19: Switch thelayer to GND andthen the Port Typeto "Standard".


52

the project has been correctly stored).The key to show the results in a diagramcan also be found in Fig. 21 and we seethat the resonant frequency fo 2449 MHz

could not be better, but the S11 curve ismore worrying. A Smithchart can bechosen from Fig 22 which is much moreinteresting and shown in Fig 23. S11 of

Fig 20: Setting theanalysis param-eters.

Fig 21: The results of an analysis.


53

approximately –20dB is a reflection fac-tor of approximately 0.1, BUT the feedpoint is too far from the lower patch edgebecause the loop in the S11 curve doesnot enclose the centre of the Smithchartand that means that the impedance valueis under 50Ω! The simulation is repeatedafter moving the Via to the next smallerdistance of 8.3174mm using the mouse.This gives a good result with S11=-30dB,near the correct matching point at 50Ω(Fig 24) with an acceptable increase of

the resonant frequency of 2505MHz.Unfortunately the feed point can only bemoved in steps of y = 0.6398mm(because we have to stay within the16Mb memory limit) and that is simplytoo rough for a finer approximation. Sowe have to be content with this simula-tion result, and a PCB with the externaldimensions of 80mm x 60mm with thefeed point as shown in Fig 25 is designedand produced.

Fig 22: The resultsof the simulation ofthe design from"patch 16".

Fig 23: The resultsshown as aSmithchart.


54

Because of many reasons e.g. PCB exter-nal dimensions may be smaller thanassumed in the simulation, manufactur-ing inaccuracies and edge under etching,influence of the soldering the feed pointwithin the patch area to the SMA connec-tor, differences in the material data etc.,the result of measurement using a net-work analysers will surely deviate fromthe simulation. Then a second PCB isnecessary with some changes!

To be continued

X.References for Part 1:

[1] Practical Project: A patch antenna for5.8GH, Gunthard Kraus, VHF Communi-cations Magazine, 1/2004, pp 20 - 29

[2] An interesting program, SonnetLite9.51, Gunthard Kraus, VHF Communica-tions Magazine, 3/2004, pp 156 - 178

[3] Modern Patch Antenna Design Part Iand Part II, Gunthard Kraus, VHF Com-munications Magazine, 1/2001 and2/2001, pp 49 – 63 and 66 - 86

Fig 24: Smithchartof the results aftermoving the feedpoint.

Fig 25: Finally the finished PCBdrawn using TARGET3001.Remember to add enough informationto uniquely identify the PCB.


55

1.Error in Fig 10 and 11

Following the publication of the transis-tor multiplier article John noticed thatthere was an error in Fig 10 and 11.

Essentially he made a mistake with thebase bias connection. He drew Fig 10and then modified the file to generate 11so the error was carried over. The baseof TR1 should only connect to the sec-ondary of T1 and not as originally drawn.The corrected Fig 10 and 11 are shownhere.

John Fielding, ZS5JF

Corrections to transistormultiplier article published inissue 4/2006

Fig 10: Revisedversion Push-Pulldoubler.


56

2.Power output of tripler

John has received an enquiry from anItalian reader who had built the 50 to150MHz tripler shown in Fig 6 & 9. Hewas unable to obtain the same poweroutput that John had achieved and waspuzzled as to why.

John dug out the prototype and ran sometests only to find it gave the powerclaimed. It was then that John realisedwhat the problem was. Fig 6 & 9 are thesame circuit but Fig 9 has the extra testpoint resistors. These are not the prob-lem. The output matching is where theloss of power occurs. The schematicsshow a 1pF output coupling capacitor(C7), this assumed the multiplier is feed-ing another transistor as either a bufferamplifier or a second multiplier. The

component values are dimensioned tofeed into an impedance of about 500Ω.If trying to drive 50Ω there will be alarge mismatch. To drive 50Ω requires achange to the output matching and thecapacitor values. The modified versionof Fig 6 is shown here. The values of C6& C7 will depend on the value of L2 andthe operating frequency. Normally C7will require varying in value to give thecorrect impedance match, this dependson the value of L2 and the dynamicimpedance occurring at resonance. Thevalue of 56pF is a good starting pointwhen the inductor is about 100nH with aQ of 100. If the inductor Q is lower thenC7 reduces in value and if L2 has ahigher Q then C7 increases in value. Thematching is a standard "capacitor-tapped"arrangement.

Fig 11: Revisedversion of Push-Pull tripler.


57

3.Input matching

Some confusion arises about the inputdrive power measurement. John statedthat +7dBm was sufficient to drive themultiplier. This of course is measuredfrom a 50Ω drive source. The base of themultiplier looks like about 500Ω, so theinput drive needs to be matched from the50Ω driver. As long as the input basevoltage swing is about 1V p-p or more itwill work OK. To drive from a 50Ωsource requires a step-up transformationto obtain the voltage swing. Either a 4:1or 9:1 transformer or another BJT config-ured as an amplifier will work. In hisprototype the 50MHz crystal oscillatoroutput was about 1V p-p into a highimpedance and this drives an emitterfollower which then drives the multiplierinput. Hence, the input drive is stillabout 1V p-p.

A signal generator when correctly termi-nated (e.g. driving a pure 50Ω load)delivers the indicated voltage, this is thepotential difference (PD) mode. But ifthe signal generator is operated into an

open circuit it delivers an output voltageof twice the PD mode, the electro motiveforce mode (EMF). If the multiplier base is driven from a50Ω signal generator the drive voltageapproaches that of an open circuit modebecause the base looks like a relativelyhigh impedance, about 500Ω. So evaluat-ing the exact drive signal level is verydifficult if a signal generator is used asthe drive source. The reason this occursis because signal generators are NOTconjugately matched devices but have anoutput resistance of 50Ω, usually fur-nished by an attenuator network in serieswith the output and the signal source.

Fig 6a: Version ofthe typicalfrequencymultiplier stagewith 50Ω outputmatching.


58

A minor issue that maybe should beadded to “The Nobel Art of measuringsmall resistances (and voltages)” articlepublished in issue 4/2006. It was aboutthe AD524 as an Instrument with manyuses. I did not mention anything aboutthe input noise. It is low, but it is there!The more resistance of course thestronger the noise.

The measured noise, 3.7mV for example,with 10KΩ to ground on each input isthen 3.7µV on the input as I used the1000X gain selection for each measure-ment.

I took some pictures from the scopewhen I had one resistor from each inputto ground. In no case did I use anyfiltering. At this gain the BW is ~25KHz.If I had selected a lower gain the noisewould have got weaker and the BWincreased. All pictures are taken with1ms/sweep speed.

The pictures are:

Rin (per input Scope gain Noise mVto ground) mV/div RMS

1MΩ 100 36100KΩ 20 10.210KΩ 5 3.71KΩ 5 1.751KΩ 2 1.75

If the inputs are left open it may seemthat even lower noise can be achieved,but it is the amplifier that bottoms outfrom it's own input bias current (±50nA

or less) so at least 100MΩ are needed tokeep the inputs in the Common ModeRange.

This noise is by far weaker than thesignal from an old dynamic microphone(2.5KΩ, built in transformer) for a reel-to-reel tape recorder. Even in the middleof the night I cannot cover the micro-phone and get it so quiet that I can seemuch of the amplifier noise on themicrophone signal! There is at least a 20- 30dB difference.

It was the DC resistance of the trans-former secondary that I measured, it is aninput transformer for a valve amplifier. Imeasured the impedance at 1KHz and itis 25KΩ. Using the AD524 instrumentamplifier is "as good" (if not better) thana valve amplifier. The input resistance isover 1GΩ, so the transformer operatesinto an open circuit. I have looked at thesignal one night with the microphonewrapped in a pair of jeans (does not helpmuch) and at a time when the fridge wasnot running. There still was some 12mVoutput from the amplifier, meaning 12µVinput, and the oscilloscope shows barelya mm wide band of noise from theamplifier on a cm or more peak to peaksignal from noise I cannot hear! So theinput noise on this amplifier is reallysomething else! I have found, at highersound levels, that the signal (from thisparticular old microphone, with trans-former) is roughly 100µV at 40dB soundpressure. This corresponds to 1µV at0dB, which is about where the amplifier

Carl Lodström, KQ6AX and SM6MOM

Instrument amplifier noise(additional information forarticle published in issue 4/2006)


59

input noise is. So one could build a soundmeter that can work down to about 0dBwith a microphone like this and theAD524. That is pretty awesome. One willnot be able to see any such low levels asone may have to be in an abandonedmine or such. Even an analogue meterwould make more noise when it deflects!With the microphone ~2m distant I getgood readings when I breath very quietlythrough my nose!

Ah, well. I figured afterwards that thenoise of an amplifier is a pretty importantparameter for an amplifier that is goingto be used around the house for a lot ofmeasurements.

Fig 1: Noise with 1MΩ input resistors Fig 2: Noise with 100KΩ inputresistors

Fig 3: Noise with 10KΩ input resistors Fig 4: Noise with 1KΩ input resistors

Fig 5: Noise with 1KΩ input resistors


60

OBJECTIVE: To increase the activity and experi-mentation on this magic band, field day operationsas well as CW. In Spain there are still some areaswhere operation on this band is not allowed orhighly restricted. So a list of all the stations takingpart in this contest will be sent to the Spanishtelecommunications ministry in order to avoid suchrestrictions in the excluded areas. We would appreciate that club stations taking partin this contest send us a letter with their logs signedby the association explaining that they agree withour demand. The letter will be attached to thestations list explained above. Please show clear inthe summary sheet if you do not want to beincluded in this list.ENTRANTS: Licensed radio amateurs and SWLfrom all over the world.CATEGORIES:

• Single operator • Multioperator or club stations • Short Wave Listeners (SWL)

Only one transmitter can be used at the same time.Multi operator stations can use two transmitters(one for each mode: SSB and CW) provided theantennas of both transmitters are separated by lessthan 50 meters. All stations must be operated fromthe same location during the contest period. CONTEST PERIOD: From Saturday 10:00 UTC9th June 2007 till Sunday 16:00 UTC 10th June 2007.Single operators must have a minimum restingperiod of 6 continuous hours.CONTACTS: One station can only be workedonce on each mode (SSB and CW). All duplicateQSOs must be clearly stated as DUPE in the logswith no points claimed. Contacts made throughrepeaters or EME are not allowed.MODES OF OPERATION: Contacts can beperformed on SSB and/or CW.EXCHANGE: RST plus full QTH locator (i.e. 599JN11BH). Contact time must be logged but you donot need to exchange it during the QSO.SWL: They must receive and write down in the logthe following data of both stations: Callsign, QTHlocator, date and time. It’s not allowed to log morethan 5 QSOs with the same station.POINTS: One (1) point per kilometre and Two (2)points per kilometre for QSOs with special stations(these stations will be identified during the QSO).POINTS for SWL: One (1) point for each QSO inSSB or CW.MULTIPLIERS: One (1) multiplier for each dif-ferent main grid square of each WW Locatorachieved during the contest (i.e JN11, IN62, etc.).One (1) multiplier for the first contact made with astation in each DXCC and WAE entities.

LOGS: The following formats are accepted:Paper logs: Normalised IARU contest sheets orsimilar and a summary sheet showing essentialinformation and leaving enough space for contestmanager or contest committee notes. Multi operatorstations must be clearly indicated. Final claimedscore must be clearly shown in the upper part of thesummary sheet.Digital logs: Only in Cabrillo format.We would be very grateful if you would send yourlogs in Cabrillo format in order to make thecorrection task easier.Logs must be postmarked not later than 15th August2007 and you can send them to:Paper logs: EADX6M CONTEST

PO Box 68E-08960 Sant Just DesvernBarcelonaSPAIN

Digital logs: [email protected] and DISQUALIFICATION:The contest organisation will take the responsibilityof log verification. Minor errors can be penalisedwith loss of points. A QSO won’t count if anobvious error in QTH locator or a time error ofmore than 10 minutes is found. Claiming points fora dupe contact will be penalised with 10 times itsvalue.The following cases will be deemed cause fordisqualification of the contestant:

• Any violation of contest rules or the IARUband plan.

• Violation of amateur radio regulations in thecountry of the contestant.

• Use of DX Cluster for self announcements orusing it as a personal log.

• Operators of multi operator stationsparticipating with his own callsign at the sametime.

• Giving false data to other contestants or theorganisation.

• Helping out certain contestants at the expenseof the rest of contestants.

PRIZES: 1st, 2nd and 3rd place in each category.Special prize for:

• Achieving most SSB contacts • Achieving most CW contacts • First place of each EA district • First place of each DXCC or WAE entity • Longest distance contact

Contestants assume total acceptance of the aboverules. All decisions and/or actions of the organisa-tion are official and final.

EADX 6m contest


61

MuRata

A very well known Far East companyfamous for coils, capacitors, filter etc.There are not only the product specifica-tions, but also the collection at S-param-eter files to the Download. The documen-tations with the measuring conditions areworth reading with the S-parameter filesfor separate frequency ranges.

Addresses: a) homepage:http://www.murata-europe.com/

b) S-parameter collection:http://www.murata.co.jp/sparameter/index.html

c) Measurement of the S-parameters:http://www.murata.co.jp/sparameter/measure.html

University of Guelph (Canada)

One finds interesting things again andagain at universities - one must searchvery thoroughly with them. On thishomepage there is a tremendous collec-

tion at electronic hobby circuits. Just asinteresting however there are links to adata sheet collections, SMD code, etc.

Address:http://www.uoguelph.ca/~antoon/circ/circuits.htm__

Eesoft - Agilent

This simulator is the “Rolls Royce” ofRF circuit simulation - both from theprice, as well as the extent and thepossibilities. Nevertheless the search isworthwhile because the non Rolls Roycecontent. Here there are introductions orApplication Notes for the simulation ofdifferent RF and microwave circuits.Also most can be used with little troublefor other cars (Pardon: Simulation pro-grams). A really interesting site.

Address:http://eesof.tm.agilent.com/applications/

Gunthard Kraus, DG8GB

Internet Treasure Trove


62

Helix Antenna Design andConstruction detail

Who is interested in this type of antennaor would like to experiment. Here is abeautiful empirical report on this topic.

Address:http://www.jcoppens.com/ant/helix/index.en.php

An experimental 10Mbit/secMicrowave Data Link

This project takes some time to appreci-ate, besides the empirical reports and/orresults of measurement of the completedproject, there are the planning docu-ments, the initial considerations and thePCB layouts etc.

Unbelievably accurately and the last de-tail inside documents; even the views forRF security and the protection of systemsare there.

Address:http://www.qsl.net/ke5fx/uwave.html

Raltron

Some on-line computations among otherthings converting SSB phase noise tojitter are to be found with RALTRON.Also the application for quartz crystaljitter and noise are worth reading.

Address:http://www.raltron.com/cust/tools/default.asp

Notice

Owing to the fact that Internet contentchanges very fast, it is not always possi-ble to list the most recent developments.We therefore apologise for any inconven-ience if Internet addresses are no longeraccessible or have recently been alteredby the operators in question.

We wish to point out that neither thecompiler nor the publisher has any liabil-ity for the correctness of any detailslisted or for the contents of the sitesreferred to!

With over 700 members world-wide, the UK Six Metre Group is the world's largestorganisation devoted to 50MHz. The ambition of the group, through the medium of its 56-page quarterly newsletter 'Six News' and through its web site www.uksmg.com, is to providethe best information available on all aspects of the band: including DX news and reports,beacon news, propagation & technical articles, six-metre equipment reviews, DXpeditionnews and technical articles.Why not join the UKSMG and give us a try? For more information contact the secretary:Dave Toombs, G8FXM, 1 Chalgrove, Halifax Way, Welwyn Garden City AL7 2QJ, UK orvisit the website.

The UK Six Metre Groupwww.uksmg.com


63

AUSTRALIA - Mark Spooner c/o, W.I.A SA/NT Division, GPOBox 1234, Adelaide, SA 5001, Australia Tel/Fax 08 8261 1998BELGIUM - UKW-BERICHTE, POB 80, D-91081 BAIERSDORF,Germany. Tel: 09133-77980. Fax: 09133-779833 .Postgiro Nbg. 30445-858.DENMARK - KM PUBLICATIONS , 63 Ringwood Road,LUTON, LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected] - Christiane Michel F5SM, Les Pillets, 89240 PARLY,France Fax: (33) 03 86 44 08 82 Tel: (33) 03 86 44 06 91 FINLAND - KM PUBLICATIONS , 63 Ringwood Road, LUTON,LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected] - UKW-BERICHTE, POB 80, D-91081BAIERSDORF, Germany. Tel: 09133 7798-0. Fax: 09133 779833.Email: [email protected] Web: www.ukwberichte.comGREECE - KM PUBLICATIONS , 63 Ringwood Road, LUTON,LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected] - KM PUBLICATIONS , 63 Ringwood Road, LUTON,LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected] - R.F. Elettronica di Rota Franco, Via Dante 5 - 20030Senago, MI, Italy. Fax 0299 48 92 76 Tel. 02 99 48 75 15 Email: [email protected] Web: www.rfmicrowave.itNEW ZEALAND - KM PUBLICATIONS , 63 Ringwood Road,LUTON, LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected] - WAVELINE AB, Box 60224, S-216 09 MALMÖ,Sweden. Tel: +46 40 16 42 66. Fax: +46 40 15 05 07.GSM: 0705 16 42 66Email: [email protected] Web: www.algonet.se/~wavelineSOUTH AFRICA - KM PUBLICATIONS , 63 Ringwood Road,LUTON, LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected] & PORTUGAL - JULIO A. PRIETO ALONSO EA4CJ,Donoso Cortes 58 5° -B, MADRID 15, Spain. Tel: 543 83 84SWEDEN - WAVELINE AB, Box 60224, S-216 09 MALMÖ,Sweden. Tel: +46 40 16 42 66. Fax: +46 40 15 05 07 Email: [email protected] Web: www.algonet.se/~wavelineSWITZERLAND - KM PUBLICATIONS , 63 Ringwood Road,LUTON, LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected] KINGDOM - KM PUBLICATIONS , 63 RingwoodRoad, LUTON, LU2 7BG, UK. Tel: +44 1582 581051. Fax: +44 1582 581051. Email: [email protected]. - GENE HARLAN, ATVQ Magazine, 5931 Alma Drive,Rockford, IL 61108, USA. Tel: +1 815 398 2683; Fax: +1 815 398 2688 Email: [email protected] - KM PUBLICATIONS, address as for the U.K.

Web: http://www.vhfcomm.co.uk

KM PUBLICATIONS,63 Ringwood Road, Luton,LU2 7BG, United KingdomTel: +44 (0) 1582 581051Fax: +44 (0) 1582 581051

Email:[email protected]

Andy Barter G8ATDThe international edition of theGerman publication UKW-Berichteis a quarterly amateur radiomagazine, especially catering forthe VHF/UHF/SHF technology.It is owned and published in theUnited Kingdom in Spring,Summer, Autumn and Winter byKM PUBLICATIONS.The 2007 subscription price is£20.85, or national equivalent.Individual copies are available at£5.25, or national equivalent each.Subscriptions should be addressedto the national representative shownin the next column. Orders forindividual copies of the magazine ,back issues, kits, binders, or anyother enquiries should be addresseddirectly to the publishers. NOTICE: No guarantee is giventhat the circuits, plans and PCBdesigns published are free ofintellectual property rights.Commercial supply of these designswithout the agreement of theAuthor and Publisher is notallowed. Users should also takenotice of all relevant laws andregulations when designing,constructing and operating radiodevices.

All rights reserved. Reprints,translations, or extracts only withthe written approval of thepublishersTranslated using Translutiontranslation software:www.translution.comPrinted in the United Kingdom by:Printwize, 9 Stepfield, Witham,Essex, CM8 3BN, UK.

Publishers

Editor

VHFCOMMUNICATIONS

© KM PUBLICATIONS

ISSN 0177-7505

A Publication for the Radio Amateur Worldwide

Especially Covering VHF, UHF and Microwaves

Volume No.39 Spring Edition 2007-Q1

VHFCOMMUNICATIONS

Action packed - you can’t afford to miss it

You Can Subscribe Securely On-line!B1 Arena Business Centre, 9 Nimrod Way, Wimborne, Dorset BH21 7SH.

Tel: 08451 933 599 Fax: 08451 933 602E-mail: [email protected]

Web: www.monitoringmonthly.co.uk

Massive 100 page Issue on sale every four weeks. Available At WHS,Borders & Leading News Agents and on-line - See our website for details.

Radio and Communications

Monitoring MonthlyISSN 1749-7809

£3.60Buy On-line - E-Store Now Openwww.monitoringmonthly.co.uk

See Us At An Amateur Radio Show Soon

Tired of radio magazines that just look backand recycle old material?Then You Need A Fresh Look At Radio WithBRITIAN’S FAVOURITE RADIO MAGAZINE!

Available from all good newsagents...be sure to reserve your copy!

Monitoring MonthlyFOR THE BEST IN HOBBY RADIO

From DC to Light!

You Can Subscribe Securely On-line!

Nice OnePublishing Limited

Back IssuesAll issues ever published are now available either asphotocopies or actual magazines. Issues from 1/1969 to4/2005 are £1.00 each + postage. Issues from 1/2006 to4/2006 are £4.70 each or £18.60 for all 4 issues + postage.See web site or page 34 of issue 1/2007 for back issue listto see which issues are only available as photocopies.There are two back issue sets that contain the available"real" magazines at a reduced price, see web site fordetails.

Blue BindersThese binders hold 12 issues (3 years) and keep yourlibrary of VHF Communications neat and tidy. You willbe able to find the issue that you want easily. Binders are£6.50 each + postage. (UK £1.40, Surface mail £1.60, Airmail to Europe £2.00, Air mail outside Europe £3.60)

PUFF Version 2.1 Microwave CADSoftware

This software is used by many authors of articles in VHFCommunications. It is supplied on 3.5 inch floppy disc orCD with a full English handbook. PUFF is £20.00 +postage. (UK £0.60, Surface mail £1.30, Air mail toEurope £1.50, Air mail outside Europe £2.50)

VHF Communications Web Sitewww.vhfcomm.co.uk

Visit the web site for more information on previousarticles. There is a full index from 1969 to the presentissue, it can be searched on line or downloaded to yourown PC to search at your leisure. If you want topurchase back issues, kits or PUFF there is a secure

order form or full details of how to contact us. The web site also contains a very useful listof site links, and downloads of some previous articles and supporting information.

Compilation CDsTwo CDs containing compilations of VHF Communica-tions magazine articles are available. CD-1 contains 21articles on measuring techniques published over the last8 years. CD-2 contains 32 articles on transmitters,receivers, amplifiers and ancillaries published over thelast 5 years. The articles are in pdf format.Each CD is £10.00 which includes 2nd class postage inThe UK and surface mail overseas. Air mail postage is£0.60 for Europe and £1.00 outside Europe.

K M Publications, 63 Ringwood Road, Luton, Beds, LU2 7BG, UKTel / Fax +44 (0) 1582 581051, Email: [email protected]

a publication for the radio amateur worldwide especially

Documents