A Publication for the Radio Amateur Worldwide Especially ......(Y3Fe5O12). The sphere is approxi-mately 0.5 to 1mm diameter. If this is exposed to an external magnetic field (H 0)

A Publication for theRadio Amateur Worldwide

Especially Covering VHF,UHF and Microwaves

Volume No.39 . Summer . 2007-Q2 . £5.25

A modern precision automatic SINAD meter,distortion meter and true RMS voltmeter

Ralph Rudersdorfer, OE3RAA and Hans-Otto Modler, OE5SMU

VHFCOMMUNICATIONS

Alexander Meier How Yig oscillators work and are driven 66 - 76DG6RBP plus an example YIG oscillator driver

Ralf Rudersdorfer A modern precision automatic SINAD meter, 77 - 101OE3RAA and distortion factor meter and true RMSHans-Otto Modler voltmeter OE5SMU

Konrad Hupfer 400W power amplifier for 2m 102 - 108DJ1EE

Zeljko Bozic A GPS controlled frequency standard 109 - 115S52ZB

Matjaz Vidmar LCD oscilloscope for spectrum analyser. 116 - 125S53MV Plus update on spectrum analyser project

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

There are some very interesting articles in this issue, I now know how a YIGoscillator works! The SINAD meter was published in three consecutive issues of UKWBerichte but I decided to put the whole article in this issue.For the readers who often ask for some 2m projects there is a nice 2m PA, theD1030UK transistor is a bit expensive but full power without a dangerous HT supplyis worth the price. Zeljko Bozic has come up with a neat design for a GPS controlledfrequency standard, a quick check on eBay and I found the GPS units for about £25.Finally a welcome return of the spectrum analyser by Matjaz Vidmar, I found his website while surfing for something else and spotted the LCD oscilloscope and thought itwould be a good idea to print the article. As Matjaz mentions on his web site he canput many more pictures (and in colour) on his site, so please take a look, but noteveryone has access to The Internet even if you do there is no guarantee you will findthe site.73s - Andy

Contents

VHF COMMUNICATIONS 2/2007

65

YIG oscillators have been used forseveral years in expensive measuringinstruments, so they are now availableas used parts at reasonable prices.However the way that these oscillatorswork and why they are suitable for usein RF Wobblers or spectrum analys-ers, because of their excellent charac-teristics, is not well known by radioamateurs. The way that they aredriven is quite different from otheroscillators, the basics are described inthe following article.

1.Characteristics of YIGoscillators

Among the most important characteris-tics of a YIG oscillator, also known as aYTO (Yig Tuned Oscillator), are thelarge tuning range of more than oneoctave and the linear tuning characteris-tic. They are different from varactortuned oscillators that require a speciallycharacterised tuning voltage to give lin-ear tuning. YIG oscillators are tuned witha variable current. To avoid temperatureeffects most YIG oscillators are heatedand held at a constant temperature. YIGoscillators are available for frequencies

from approximately 1.5GHz to over24GHz with power outputs typically+13dBm (20mW). The first harmonic isusually suppressed by approximately 10to 20dB.

2.Operation of YIG oscillators

The circuit diagram of a simple YIGoscillator is shown in Fig 1. The interest-ing part is the YIG sphere in the emittercircuit of transistor T1. But how does aYIG resonator functions?A YIG resonator consists of a coil with asmall black sphere sitting in the centre asshown in Fig 2. The abbreviation YIGstands for Yttrium Iron Garnet(Y3Fe5O12). The sphere is approxi-mately 0.5 to 1mm diameter. If this isexposed to an external magnetic field(H0) the spinning electrons in the YIGthat form magnetic dipoles start to alignwith the field and precess around thatfield. This generates a small alternatingfield or RF signal that is coupled to thesampling coil. This resonant frequency(f0) is independent of the YIG spherediameter and can be computed by:

Alexander Meier, DG6RBP

How YIG oscillators work andare driven plus an example YIGoscillator driver

( )aHHf += 00 γ ⎥⎦⎤

⎢⎣⎡⋅=

mAHz/

10352 3γ


66

Where:γ = gyro magnetic relationship, a con-stant dependant on the charge massrelationship of electrons.H0 = Continuous external magneticfield.Ha = Anisotropic field, this is thedifference between the applied fieldand the very small (nearly negligible)field internal to the YIG. This istemperature dependent but can beminimised by an adjustment of theYIG sphere.

If this YIG arrangement, with its periodi-cal characteristics, is used in the circuitof an oscillator (Fig 1), then the fre-quency can be controlled by an externalmagnetic field H0. This magnetic field isgenerated by a direct current in fieldcoils, this is the tuning current. If thevery small anisotropy field is disre-

garded, then f0 is proportional to H0giving a tuning range linear to the tuningcurrent.

3.YIG connections

YIG oscillators are usually built intoround or square metal housings, see Fig3. The SMA socket is the RF output, theother solder pins are for the supplyvoltage and the tuning current. The con-nections are not standardised thereforecare is needed when buying a used YIGoscillator at a flea market. The followingdata refers to the connections of a typicalYIG oscillator like the one shown in Fig4.

Fig 1: Circuit of aYIG oscillator [1].

Fig 2: Pictureshowing the YIGsphere.


67

• Heating (not always available)

• Operating voltage (positive, negativeor both)

• Tuning coil (Tune), typically20MHz/mA

• Tuning coil (FM), typically300kHz/mA

• RF output, typically +13dBmFrequently there are two connections forthe heater, one should be connected toground and the other to the appropriateDC voltage (usually +24V). Initially ahigh inrush current will flow (often >0.5A), this settles down within a few

seconds to approximately 50 to 100mA.This heats the YIG sphere to 80 - 125°Cand is kept constant at this value. Theheating is necessary to minimise tem-perature affects. Some YIG oscillatorsonly require a positive supply voltage of15V at 100 to 300mA, others require anegative supply and some require both apositive and a negative supply. Alsounusual supply voltages are sometimesrequired e.g. -4.3V or -5.1V. The groundconnection is usually missing, thereforethis must be connected to the housing.Two coils are used for frequency tuning:

• The main coil “Tune” (approximately10Ω) is used for coarse frequencyadjustment. The tuning rate is usually20MHz per mA! Thus a current of(2000MHz/20MHz) = 100mA is nec-essary for an output frequency of2GHz. For 18GHz this is then900mA! This high value of currentcan cause problems with control, ifthis varies by ±2.5mA, then the out-put frequency varies by ± 50MHz!The control current should thereforebe as stable and as low noise aspossible. The tuning linearity is verygood, typically ±0.1%. To achievebetter stabilisation of the output fre-quency a capacitor of approximately500µF can be connected in parallelwith the tuning coil. However theYIG oscillator cannot be tuned so fastand is not suitable for a Wobbler orspectrum analyser and is only suit-able as a CW oscillator.

• The second coil is the “FM coil”(approximately 1Ω) and a tuning rateof typically 300kHz/mA. Thus theYIG oscillator can be frequencymodulated or this coil can be used forsmall adjustments of the output fre-quency. It has a smaller inductancethan the main coil, so that fastertuning is possible. The small FM coilis wound with much thinner wire sothe control circuit should have asafety device to limit the current to<250mA.

Fig 3: A 4 to 8GHz YIG oscillator.

Fig 4: Typical connections for a YIGoscillator.


68

If only the tuning coil is used it does notmatter which connection is positive. Ifboth tuning coils are used the polarity isimportant so that both magnetic fields areadditive. The tuning coils should neverbe used with a constant voltage becausethe coils warm up due to the current flowchanging their resistance, hence the cur-rent changes which changes the outputfrequency.

4.Control of a YIG oscillator

To control the frequency of the YIGoscillator with a voltage, a current sourceis used as shown in Fig 5. The currentflowing in the YIG coil also flowsthrough the resistance “R” to ground.The current produces a proportional volt-age drop across the resistance “R” that isfed to the inverting input of the operationamplifier. This compares the actual cur-rent with the tuning voltage on the non-inverting input and regulates the current.The control current is given by:

It is advisable to use a field effecttransistor instead of a bipolar transistorbecause the base current of a transistorwould flow through resistance “R” caus-ing a small error. The diode in parallelwith the YIG coil is necessary, if thecontrol voltage exhibits fast jumps e.g.with a ramp signal or when abruptswitching occurs, because an inductionvoltage would be developed that coulddestroy the transistor. The resistor “R”should only be wound from Constantinewire. This gives very good temperaturestability and very good frequency stabil-ity. The characteristics of other wirewound resistors are not as good. Theresistance should be fitted so that it iskept at a constant temperature if possible.An alternatively circuit can be used thatgives an inversion from input voltage tooutput current. This type of circuit isfrequently used and has some advantagesbut needs a negative control voltage. Thecircuit is not described here because theprinciple is the same. A low pass filter should be used in frontof the current source. If the YIG oscilla-tor is used for a Wobbler, the filtershould be designed for approximately50kHz so that is does not to deform thesawtooth. For CW operation the filterR

VI tunecoil =

Fig 5: Currentsource for use witha YIG oscillator.


69

should be a much lower frequency, ap-proximately 10Hz to suppress any lowfrequencies e.g. 50Hz mains. Addition-ally a capacitor with a small seriesresistor can be used in parallel with thetuning coil for CW operation. The ca-pacitor should be approximately 500µF.If the YIG oscillator is used for aWobbler, the FM coil can be used tocontrol the frequency in the same way asthe main tuning coil. For small spansweepers, the current in the main coil isset to a fixed current to give the startingfrequency and then the FM coil used tovary the frequency. The FM coil can onlychange the frequency around that set bythe tuning coil. If the FM coil is not usedfor a Wobbler, it is advisable to use a DCcoupled push-pull operational amplifierin the circuit of Fig 5. This will give thebest low frequency response and allowthe frequency to be varied in both direc-tions. Finally it should be emphasised thatabsolutely low noise and stable supplyvoltages are necessary for the controlcircuits! The smallest fluctuations, e.g.by stray effects of transformers make theoutput frequency vary very greatly! Highquality, low noise amplifiers should beused e.g. OP07 or OP27. The wiringbetween the control circuit and the YIGoscillator must be laid out with careavoiding stray effects from transformersand other cables. The wires to the tuningcoils should be twisted. Unfortunately50Hz modulation is nearly alwayspresent on the RF output caused by strayeffects finding their way into the wiringof the control circuitry.

5.A universal driver circuit forYIG oscillators

The circuit diagram of the driver circuit

is shown in Fig 6 and Fig 7. The controlvoltage between 0 and 4V is applied atpin 2 of J1. The trimmer R2 sets thetuning rate for the YTO between 0.3 and2.3V/GHz. For example this would be setto 1.5V/GHz for a YTO with a frequencyrange from 2 to 8GHz. So that theoscillator is set to its lowest frequencywhen the control voltage is 0V, an offsetis applied to the non inverting input ofthe operational amplifier U1. This isproduced by a low noise precision refer-ence (U4) and inverted with the operationamplifier U3. The trimmer R8 is used toadjust the offset voltage, which producesan offset current in the main coil (Tune)of the YIG oscillator from 50 to 650mA.For our example oscillator from 2 to8GHz the offset current is adjusted to100mA.

The low pass filter (R12, C12) has a cutoff frequency of approximately 50kHz.This does not cause distortion even withvery fast sawtooth sweep speeds. ForCW operation the cut off frequency canbe reduced to 15Hz with the switch Q4and the capacitor C11, this will reducethe 50Hz hum.

The filtered control voltage is fed to thecurrent source U2. That control currentfor the YIG coil is proportional to thevoltage drop across the resistor R18. Theoperation amplifier compares the actualvalue with the desired value (controlvoltage) and regulates the current accord-ingly. The resistor R18 should only bewound from Constantine resistance wire.Only this wire will give outstandingtemperature and thus frequency stability.The current in the main coil (Tune) of theYIG oscillator can be measured with avoltmeter across the test connection TP1at 1V per Ampere.

The diode D2 limits the affects of abruptchanges in the coil current due to jumpsin the control voltage e.g. sawtooth usedfor a Wobbler and protects the controlcircuit.


70

F

ig 6

: Par

t 1 o

f cir

cuit

for

YIG

osci

llato

r dr

iver

.


71

Fig 7: Part 2 of circuit for YIG oscillator driver.


72

The reed relay K1 switches the capacitorin parallel with the main coil (Tune) forCW operation. This greatly improves the

frequency stability. The value of thecapacitor and the series resistor R17(optional) can be selected individually.For a first attempt, select C15 of 470µF

Fig 8: PCB layout for component sideof YIG oscillator driver.

Fig 9: PCB layout for SMD side,actual size is 65mm x 80mm.

Fig 10: Component layout forcomponent side of PCB.

Fig 11: Component layout for SMDside of PCB.


73

and R17 of 0Ω.

The control of the FM coil was devel-oped for Wobber applications with smallfrequency changes. It is only possible tochange the frequency in one direction butthis is sufficient for a Wobbler. Thestarting frequency is set by the main coil.The circuit can be used for FM modula-tion by overlaying the modulation volt-age onto a DC voltage of 2V at pin 4 ofJ1.

The FM coil has a smaller tuning rate of300kHz/mA therefore a precision metalfilm resistor is sufficient for R23. Thecurrent in the coil can be measured witha voltmeter across test point TP3 at 1Vper 50mA. The fuse F1 provides protec-tion for the thinner wire of the FM coil inthe event of a fault.

The supply voltage for the oscillator used(+15V/-5V) is fed to the circuitry via alow pass filter (L6 and C22, L3 andC23). The supply voltage for the controlcircuit (±15V) is filtered separately. Theonly other supply required is +24V forthe heater of the YIG oscillator and thecurrent supply for the YIG coils. Thepower supply for the control circuit must

be able to supply sufficient current forthe YIG oscillator. All output voltages ofthe power supply must be protected inorder to save the expensive oscillator inthe event of a fault. It should only be ahigh quality, linear regulated power sup-ply (no switch mode power supplies).

The PCB is a 65mm x 80mm doublesided board with plated through holes,the layout is shown in Figs 8 and 9.

6.Construction

The component layout is shown in Figs10 and 11. First all of the SMD parts arefitted followed by the wired components.The Constantine wire wound resistor R18is wound on a thin aluminium rod cov-ered with insulating tape. The completedresistor is fastened onto the PCB usingtwo-part adhesive. The transistors Q2 andQ3 are fitted to suitably sized heatsinksusing thermal paste. The circuit can betested with resistors in place of the YIGof oscillator coils, 10Ω for the Tune coil

Fig 12: FinishedPCB for the YIGoscillator driver.


74

and 1Ω for the FM coil. With the supplyconnected all of the supply voltage forthe YIG should be tested on connectorJ2. Next the control voltages are con-nected and the current in the 10Ω and 1Ωresistors measured. If everything func-tions corectly, the YIG oscillator can beattached.

For the wiring of the driver circuit to theother building blocks (YIG oscillator,power supply, etc.) a suitable arrange-ment to minimise stray effects (e.g. 50Hzhum) must be adopted. The control volt-age should be connected to the controlcircuit using shielded coaxial cable.

A finished printed circuit board is shownin Fig 12 and a printed circuit board withYIG oscillator and heatsinks, built into amicrowave Wobbler, is shown in Fig 13.

7.Parts list

R17 0Ω, 1206R9 100Ω, 1206R3 680Ω, 1206R7 1.8kΩ, 1206

R21 2.2kΩ, 1206R14, R22 4.7kΩ, 1206R4, R5, R6, R10, 10kΩ, 1206R12, R15, R19R11 27kΩ, 1206R1, R13 100kΩ, 1206R8 1kΩ, trimmer 64YR2, R20 5kΩ, trimmer 64YR18 1Ω, ConstantineR23 20Ω, 1%, metallic

film 1/4W

C12 330pF, 0805C1-C3, C5-C7, 100nF, 0805C13, C14, C16, C18, C19, C21C8, C9 1µF/16V, SMD

tantalumC11 1µF/63V, RM5,

MKSC10 4.7µF/35V, SMD

ElectrolyticC4, C20, C22, C23, 10µF/35V, SMDC25, C26 ElectrolyticC17, C24 100µF/63V, RadialC15 470µF/63V, Radial

L1 10µH, wired, 1 AL3, L4-L6 10µH, SIMID 1210L2 100µH, SIMID

1210

Fig 13: YIG oscillator driver built into a microwave Wobbler.


75

U1, U2, U3, U5 OP27 SMDU4 MAX 6350, SMDD1, D3, D5 LL 4148D2, D4 DL 4001Q1 UC848Q4 BSS138Q2, Q3 IRF510

K1 HE 3621 A, Reed Relay 24V

F1 Fuse F250mA + holder

J1 5 pole plugJ2, J3 8 pole plug1 x receptacle for 5 pole

plug2 x receptacle for 8 pole

plug21 x crimp pins for plugs2 x Test pins1 x Heatsink1 x PCB DG6RBP YIG

control

8.References

[1] Spectrum analyzer from LF to48GHz, Vieland Carsten, Lecture to theWeinheim VHF conference 1987

9.Literature

For further reading about YIG oscillators,the following literature is recommended:

[a] YIG oscillators and their control,Alexander Meier, Lecture on the VHF,Weinheim conference (2003), The cur-rent version can be found on the AMEEngineering web site, www.ame-engineering.de

[b] A simple approach to YIG oscillators,Bernd Kaa, VHF CommunicationsMagazine 4/2004, pp 217 - 224

[c] YIG resonators and filters, JosephHelszajn, John Wiley & Sons Ltd.(1985), ISBN 047190516

[d] Microwave filters, Impedance Match-ing network, and Coupling Structures,Matthaei Young Jones, Artech HouseInc., 1980, ISBN 0890060991

[e] Microwave engineering, Günter Käsand Peter Pauli, Franzis PublishingHouse, Munich 1991, ISBN 3-7723-5594-3

[f] Spectrum analyser from LF to 48GHz,Vieland Carsten, Lecture to the Wein-heim VHF conference 1987


76

In all receiver engineering with similartypes of modulation the data for thereceiver characteristics refer tosignal/signal-to-noise ratio and noisefigure either directly or indirectly. Thequality of the audio signal is evaluatedby the distortion factor or signalbreakdown. To simplify the measure-ment procedures a SINAD meter isused. The equipment presented in thisarticle combines a distortion factormeter with a true RMS voltmeter withover 102dB range, in fact all the capa-bilities of a laboratory instrument (forthe AF side) to make measurementprecisely and efficiently. A CCITTweighting filter permits, if required,correct evaluation according to thestandard.

1.Introduction

The characteristics of a radio receiver areusually measured by two general meas-urement methods. The receiver (RX) isconsidered as a two port device. Thispractice permits the examination of dataduring maintenance or in service withoutthe need to interfere with the test speci-men. The procedure for determining

many fundamental characteristics e.g. thereceiver sensitivity, is shown in the sche-matic diagram, Fig 1.To determine the behaviour in the sce-nario when several transmitters are re-ceived at the antenna socket, the two ormulti-transmitter measuring procedure isused as shown in Fig 2. Typically this isused to examine the large signal charac-teristics for adjacent channel suppression,blocking or intermodulation.Further information on the fundamentalmeasurements is available in [2]. [3]deals with the special case of frequencymodulation F3E.From the block diagrams it can be seenthat the measuring instrument describedin this article needs to be able to evaluatethe AF output signal. Measuring the frequency of the demodu-lated output is only required in somespecial cased. Because even inexpensivedigital meters can measure frequenciesup to 10MHz with a resolution of fourplaces it was a conscious decision not toinclude an AF counter.To make the instrument easy to use itwas decided to use a moving pointermeter. This is particularly useful foralignment work, it is much easier tofollow a moving pointer rather than aflickering number display. This is alsotrue for changes in measurements

A modern precision automaticSINAD meter, distortion factormeter and true RMS voltmeter

Ralf Rudersdorfer, OE3RAA, Hans-Otto Modler, OE5SMU


77

2.Design of the measuredvariables

A signal VGW splits down into the domi-nant mode plus harmonics (they are ameasure of the signal distortion). Addi-tionally to the signal distortion there isalso more or less noise. The dominantmode represents the actual informationsignal. This must be larger by a givenmargin (expressed in dB) than the un-wanted signals otherwise the informationcontent is impaired too much.

2.1. True RMSIf DC voltage is applied to a resistor,then a constant current flows through theresistor and it warms up. The electricalpower causing this heating is determinedby the well-known equation:

Where:P is in Watts (W); V is in volts and R isin ΩIf the current flowing through the resis-tor is alternating rather than direct thenthe value of current flowing is dependanton time. The resistor heats up and coolsdown with time but the temperature will

Fig 1: Main method for testing the characteristics of a radio receiver.

Fig 2: Method ofmeasuring receivercharacteristicsusing two signalsources.

RVP

2

= (1)


78

settle down to an average value. Inelectronics the power dissipation is ofinterest in connection with the thermalload (see typical data sheet: max. energydissipation Ptot =… W), it is thus usuallyonly this average value that is of impor-tance. The power dissipation is derivedfrom the square of the voltage waveform.It is called RMS (Root Mean Square) andis defined as follows. The RMS is thatportion of the alternating current thatwould cause the same thermal effect asan equivalent direct current! This canalso be applied to the voltage. Thus theRMS of an alternating voltage must besmaller than the peak value. Dependenton the shape of the waveform the RMSvalue is related to the peak value, see Fig3. This relationship defines the FormFactor:The Form Factor is periodic (thus alwaysrecurring) and is value of the relation-ship between the peak value and the RMSvalue. For sinusoidal alternating volt-

ages this is 1.41. The display on movingpointer measuring instruments using arectifier is calibrated in RMS only forsinusoidal signals. The same applies tomany simple digital instruments. How-ever if the signal is not sinusoidal then agenuine RMS measuring instrument isneeded (also called true RMS) [6].Because noise is indefinite and con-stantly changing, measuring its RMSvalue gives a measure of the noise per-formance rather than trying to measurethe peak value. From [8] a measuringinstrument must be able to measure whitenoise having a form factor of 5 with a 1%error.

2.2. S/N, (S+N) /NFor a small information signal, noise isthe dominant degradation factor. Therelationship of the noise to a signal of thesame level can be described as signal-to-noise ratio (S/N) of 0dB - both compo-nents have the same signal level. The

Fig 3: Diagramshowing the formfactors fordifferentwaveforms.


79

term Signal-to-Noise-Ratio (SNR) iscommon. In practice it is not possible to determinethe S/N by measuring the signal then thenoise because the noise cannot beswitched off. Thus the signal must bemeasured with the noise to determine theSignal plus Noise to Noise (S+N)/Nwhich is the auxiliary form that sur-rounds us in the natural environment [9].The larger that the signal-to-noise ratiois, the smaller the actual difference. For amoderate signal with (S+N)/N of 7dB theS/N will be 1dB lower.

2.3. SINADFor larger signal-to-noise ratios, otherparts of the signal become more domi-nant than the noise. These can causeharmonics during the demodulation ofthe AF signal that can impair the compre-hensibility or “quality of understanding”.Measuring the noise and all other signalimpurities is the familiar SINAD value.This stands for “Signal, Noise and Dis-tortion to Noise and Distortion”,(S+N+D)/N+D. This mathematical equa-tion (2) is:

Fig 4 shows how a signal can be splitinto it’s component parts by frequency inaccordance with the equation above.

Measurements of receiver sensitivity us-ing a SINAD meter are quite efficient.For example the capture of the noise andAF signal can be automated by measur-ing the noise when the demodulatedinformation fades out. The SINAD indBs can be directly displayed. Impor-tantly a notch filter is used in the SINADmeter to measure only the signal infor-mation at the centre frequency of thefilter to avoid measurement errors. Thisis because most of the information in anAF signal is contained at 1kHz. In thecase of poor signal processing in a radioreceiver the same values are obtained for(S+N) /N and SINAD.For telegraph (A1A modulation – Morse)a full amplitude signal is available whenthe code is transmitted, this is not thecase for other transmission modes e.g.AM (A3E) or Single Sideband (J3E)where the amplitude of the modulationenvelope varies. Since many transmis-sion modes are complex with much morethan a single tone present (e.g. the wholespeech band) a certain minimum signal tonoise ratio is needed for acceptable trans-mission. Therefore the operating sensitiv-ity for a radio receiver is specified for acertain (S+N)/(N) or SINAD at the AFoutput (for a reference power output).Normally 10dBs, 12dBs or 20dBs isacceptable and are often specified intests. A signal-to-noise ratio of 33dBscorresponds to broadcast quality.

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛

++++⋅=

223

22

21

2 ....log20

nOWOWOWOWR

ges

VVVVV

VSINAD

Fig 4: Diagramshowing thecomponent parts ofa signal.


80

The maximum signal-to-noise ratiomarks the highest attainable informationsignal quality. This is usually due to thenoise but SINAD also measures AFharmonics that can be very meaningful.Besides the AF section of a radio re-ceiver developing distortion, noise volt-age can come from the power supply.Single sideband can generate close innoise from the different conversion oscil-lators [5].

2.4. Distortion factorInstead of measuring the maximumsignal-to-noise ratio the distortion factorcan be measured. The distortion cancome from many sources but mainly theaudio stages. For a sine wave test it isassumed that all harmonics except the ofdominant signal cause a decrease of thesignal quality. The mathematical descrip-tion of this is (3):

The distortion factor is the ratio of theRMS value of all the harmonics, exceptthe dominant signal, to the RMS value ofthe whole signal, Vges, expressed as apercentage (Fig 4). This is normallycalled the Total Distortion Factor or THD(Total Harmonic Distortion) [4]. A valueof 10% for k% means that the harmonicsof a signal amount to at the most 10% of

the total signal.For a radio receiver the demodulationdistortion factor is an important param-eter. This is measured for a referencepower output for several defined RFsignals at the antenna socket. The RFsignals are modulated with nominalmodulation. Good designs exhibit de-modulation distortion factors of under1% with large input signals.

2.5.Weighting filterThe frequency range of the measuringinstrument is restricted with a weightingfilter. In the strictest case this is abandpass filter shaped to the response ofthe human hearing as used for telecom-munication transmission. This is usuallythe standard curve described by theCCITT standard P53A (general known asa CCITT filter) as shown in Fig 5. As aresult of using a weighting filter thesignal-to-noise ratio measured will behigher and the distortion factor will belower. This is due to a decrease in thenoise and the missing harmonics that areoutside the filter pass band, thus on paperreceiver sensitivity is better. For radioreceivers with different AF output sig-nals, i.e. special decoders, the results willnot be meaningful. For meaningful com-parisons it is important to specify theconditions of the test.

%100... 22

32

22

1% ⋅

+++=

ges

nOWOWOWOW

VVVVV

k

Fig 5: The responsecurve for a CCITTfilter.


81

3.The concept

3.1. Handling different RMS voltagesThe Analog Devices [14] AD536Ashown in Fig 6 is an integrated circuitcontaining a true RMS to DC converter.The output voltage is proportional to theRMS value of the input voltage. For aForm Factor of 7 the error is less than1%, even with a Form factor of 10 it isstill only 2.5%. This transfer function ofthe IC can be described in principle as:

Where:Vout = output voltage proportional tothe rms value of the input signalM = number of samples over theaveraging periodVin = input signal at any point in time

(tm)

This is the correct mathematical defini-tion of the internal operation. The lengthof time that is taken to measure theaverage can be determined simply by anexternal capacitor. An internal buffer canbe used to drive a moving pointer micro-ammeter with a current proportional tothe RMS value. The scale of this meterwill be linear in RMS volts.A wide band RMS converter is needed tomeasure all of the harmonics in thesignal. From the response of the humanear, 16kHz is the obvious AF range. TheAD536 has an error of less than 1% up to45kHz for an input of 100mV. Withsmaller input voltages it’s range dropsrapidly so that at 10mV input it is 5kHzfor less than 1% error. This means that anamplifier will be required for measuringsmall signals. Using a change of amplifi-cation at the correct point the resolutioncan be maintained by stretching the inputrange. A proven way to stagger thechanges in amplification is to useswitched gains of 1/3.169/10/31.69. Thisgives a downward step of 10dB eachtime the gain is switched.

Fig 6: Data sheetfor the AnalogDevices [14]ADF536A.

( )∑+

⋅∝M

mminOUT tV

MV

1

21 (4)

dB10169.310log20

1069.31log20 =⎟

⎠⎞

⎜⎝⎛⋅=⎟

⎠⎞

⎜⎝⎛⋅


82

This is shown in the block diagram, Fig8, it can be bypassed when the switch isset to VRMS.

3.2. Development to the automaticSINAD meterAs described in the introduction, thesignal to be measured consists of thedominant mode plus signal distortion andthe noise. The signal to be measured(Vges) can contain three signal compo-nents so using equation (2) we candevelop equation (5).

Because all values are RMS they can

only be summed by squaring and thentaking the square root. It can be seen thatthe term on the top of the equation differsfrom the bottom term by the addition ofVGW

2. If the measured signal can alwayshave a constant RMS voltage then theterm on the top of the equation will havea constant value. This can be realised by using an ALC(automatic level control) circuit, thismust handle high Form factors. If theprimary signal VGW is held constant or“locked out” then the RMS value of theremaining signal is the direct relationshipbetween the two RMS signals and can bedisplayed on an output scale. The SINADis higher for smaller values of noise andharmonics so a high SINAD gives asmall pointer movement. The sensitivityfor measuring SINAD values is good andis predominantly determined by the self-noise of the meter circuit.To represent SINAD correctly in dB thelogarithm of the result must be taken andthen multiplied by 20. This is a math-ematical process that is applied to allvalues. This can be achieve using alogarithmic scale on the meter as shownin Fig 7.

3.3. Expansion for distortion factormeasurementLooking at the equation for distortion, (3)can be expanded to give equation (6).

Fig 7: The output meter has a specialscale to show all of the ranges that canbe measured.

22

32

22

12

223

22

21

22

...

...log20

nOWOWOWOWR

nOWOWOWOWRGW

VVVVV

VVVVVVSINAD

++++

+++++⋅=

%100...

...22

32

22

122

223

22

21

% ⋅+++++

+++=

nOWOWOWOWRGW

nOWOWOWOW

VVVVVV

VVVVk

Fing 8: Blockdiagram of theSINAD meter withfacilities to measuredistortion factorand true RMSvoltage.


83

Comparing equations (5) and (6) thereare some clear similarities, however: • The denominator in (6) is the same as

the numerator in (5) • The numerator in (6) is the same as

the denominator in (5) but the termVR

2 is missingTherefore the considerations alreadymade in the preceding section apply. Thesituation is now reversed so that thesmaller the measured RMS value of theharmonics with locked out primary sig-nal, the smaller the distortion factor willbe. This means that good values ofdistortion factor will result in a smallpointer movement on the meter. Thescale for this measurement is linear andshown as the two upper scales in Fig 7.These are also used for the RMS voltagemeasurement.As mentioned in section 2.2, noise cannotbe switched off and it cannot be isolatedfrom the signal to be measured. This is aproblem because it is not in the definitionof distortion factor. Fortunately the har-monics are usually much larger than thenoise so it has an infinitesimal influenceon the measured result. Even if this is notthe case, the measured value is meaning-ful, since it gives information about thesignal quality.

3.4. Concept summaryThe block diagram of the instrument isshown in Fig 8. The signal is amplifiedwith a switchable gain amplifier, thismakes it possible to make good measure-ments of SINAD and distortion factor bystretching the measurment range. Anadditional weighting filter can be used ifrequired.

4.Circuit description

A balanced input is used to prevent humloops and the effects of other inputdisturbances [10].Fig 9: Input circuit of the SINAD

meter.


84

4.1. Circuit diagram: Fig 9The 1mH chokes and capaci-tors at the input prevent anyRF that may be on the inputsignal from getting into themeasuring circuit. The switchS1 can be used to select be-tween a high impedance inputimpedance of approximately50kΩ and a load of 600Ωusual used for audio systems.The two 10µF electrolytic ca-pacitors form a unipolar ca-pacitor giving galvanic isola-tion from the following cir-cuit. The high capacitor valuesare critical for a low fre-quency response. The 470Ωtrimmer potentiometer can beadjusted for accurate symme-try. If necessary the 4.7µFelectrolytic capacitor can bechanged for a somewhatsmaller value.The input of the operationalamplifier is protected fromlarge voltage differences bytwo back-to-back diodes. S2in Fig 9 and 10 is a standardtwo wafer rotary switch withtwelve positions but only ninepositions are used. It is used togradually change the amplifi-cation and thus the measuringrange. Signals under 250µVeffcan actual be measured per-fectly. The largest measuringrange is 32V full scale givinga signal processing range ofmore than 102dB.The 0Ω connections for themeasuring of ranges 1mV to100mV, and 300mV to 10Von the rotary switch S2a arenot on the printed circuitboard, they are soldered di-rectly at the switch. An option is provided tochange the measuring range tothe largest value at any timewhile leaving all the other

Fig 10: Input amplifier circuit of the SINADmeter.


85

F

ig 1

1: N

otch

filte

r an

d A

LC

circ

uits

of t

he S

INA

Dm

eter

.


86

F

ig 1

2: R

ange

sel

ectio

n an

d ou

tput

am

plifi

er c

ircu

it of

the

SIN

AD

met

er.


87

setting the same. This uses the four-polerocker switch S3. This helps to protectthe moving coil meter when measuringunknown signals. S3 can be switched totake the pointer off of the end stop whilethe least sensitive range is selected.

4.2. Circuit diagram Fig 10The non inverting inputs of the followingoperational amplifier chain are connectedto half of the supply voltage by a resistordivider, therefore a single positive supplyis all that is required. The amplificationrange can be set using the 47kΩ trimmerpotentiometer in the feedback path ofIC2. The parallel 10pF capacitor mini-mises the high frequency response. Thesame applies to the other 10pF and 47pFcapacitors. The AC portion of the signalis coupled to the true RMS level controlusing a 10µF electrolytic capacitor.

4.3. Circuit diagram Fig 11The 22kΩ resistor forms a variable volt-age divider with the photoelectric cell(like a light-dependent resistor) in theopto-isolator NSL-32 from Silonex [16].This opto-isolator is ideally suited asdescribed in the product description [17]it says: “The resistive output of this typeof device has very low distortion charac-teristics”. The true RMS to DC conveter,IC4, gives an output proportional to thetrue rms value of the signal from IC3, pin7. It is available at pin 6 across the 1.2kΩresistor. The following circuit formed by thesecond half of the operational amplifier

IC3 works as integrator. The output atpin 1 feeds current to the LED in theopto-isolator as long as the voltage on thenegative input is not equal to the 3V onthe positive input. The illumination of theLED changes the resistance of the photo-electric cell between some hundred ohmsand over 200kΩ. The changes of resist-ance take place slowly with the durationto the final value in the seconds range.This give the ALC it’s control system, itis impossible for it to react to shortvoltage peaks and the response is to themeasured signal.A signal with a constant level of approxi-mately 710mVeff is fed via a 4.7µF elec-trolytic capacitor to the input of the notchfilter. It corresponds to an accurate full-scale image of the measured signal.Measurements confirm that the level iskept extremely constant at this point,even if the measuring range werewrongly selected (see Fig. 16).The 1kHz dominant mode of the measur-ing signal is eliminated by three cascadedband pass filters (Fig 11 and Fig 12).They can be easily adjusted to differentnotch frequencies so that faulty measure-ments are excluded by slightly deviatingfrom the 1kHz frequency. The literature[11], shows that these filters require verylittle calculation. The 100Ω resistors atthe output of the operational amplifiers inthe feedback loop, easily expand themaximum stop band attenuation. Thisgives an accurate band pass filter withsteep sides, Fig 13 shows the actual filter

Fig 13: Frequencyresponse of the1kHz notch filter.


88

F

ig 1

4: T

he C

CIT

Tfil

ter

circ

uit o

f the

SIN

AD

met

er.


89

characteristic. It is characterised by highstop band attenuation and very highquality. Also the first harmonic of themeasured signal is covered. The smallpass band attenuation at 2kHz meets therequirements for SINAD measurements.

4.4. Circuit diagram Fig 12Switch S4 is used to select the differentdistortion factor measuring range andSINAD measuring range by switchingthe amplification of the following invert-ing amplifier. Each of the four trimmerpotentiometers (“30% -DF”, “10% -DF”,“SINAD”, “SINAD +20dB”) are for fineadjustment of each individual measuringrange. The capacitors in series with the twobranches for distortion factor measure-ment limit the frequency response to lowfrequencies. Signals below the first har-monic of the measured signal are withoutmeaning. Therefore the dynamics of thedistortion factor measurement range isincreased slightly. The 82pF capacitor inparallel with the resistors in the feedbackbranch of the operational amplifier limitsthe high frequency response. Switch S5 selects the measured variable,for RMS (VRMS) the signal before theALC is measured. The signal decoupledfrom the DC voltage is the true RMS toDC transducer, IC9, input signal. It di-rectly feeds the moving coil instrumentwith 100µA full-scale.

4.5. Circuit diagram Fig 14If required the CCITT filter can beconnected using switch S6. The circuitwith operational amplifiers IC7 and IC8gives a frequency response in accordancewith CCITT P53A. The trimmer potenti-ometer at the output stages can be used toalign the filter.

4.6. Circuit diagram of the powersupply (Fig 15)The circuit of the power supply is shownin Fig 15 with the PCB layout in Fig 17and the component layout in Fig 18. Theactual stabilised DC voltage required is14V. It is obtained using an economical

Fig 15: The power supply circuit forthe SINAD meter.


90

15V regulator and depending on theoutput voltage either one or two silicondiodes.

5.Construction

5.1. Selection of suitable componentsSince this is a measuring instrument,

resistors with 1% tolerance should beused. Some of the resistors shown in thecircuit diagrams with a value in bracketsare quite rare so the value shown inbrackets can be use with a small loss inaccuracy. Similarly capacitors with closetolerances save problems with alignment.This particularly applies to componentsin filters where foil capacitors or smallvalue ceramic capacitors with NP0 di-electric should be 5% tolerance. If nototherwise stated components should berated for at least 16V. The 500pF capaci-tor on pin 1 of IC7 is made up from a

Fig 16: The signalat the output of theALC on pin 7 ofIC3 with differentinputs. Forexample withsignals between140mV and 3V fedinto the 1Vmeasuring rangethe output isindependent of thesignal form andalways a constantRMS value. Theoutput is always710mV.


91

470pF capacitor with a 27pF capacitor inparallel, these could be SMD 0804 or0603 size.The PCB for the power supply has beendesigned to accept various transformerswith E-I cores from 3 to 3.8VA. Usingdifferent transformers may result in someunused solder connections on the PCB.To ensure that the reading accuracy andmeasurement tolerance are maintained alarge 100µA moving coil meter with 1%to 2% accuracy should be used.The AD536A is available in variouspackages, the PCB shown in Figs 19 and22 are designed for the DIP14 package.The more economical TO100 packagecan be used for the AD536AJH IC4 andIC9, an adaptor PCB, shown in Fig 25,was designed to adapt this package to aDIL footprint. To use the adaptor fitsolder pins for the DIL positions andform the leads of the TO100 package intoan oval shape before fitting to the adap-tor. The rest of the instrument uses thewell-proven RC4558 low noise op ampli-fier from Texas instruments [15].

5.2. Scale layoutThe scale for the meter was producedusing software available from JamesTonne, WB6BLD, [19]. It was written todesign meter scales and has the facilitiesto accommodate various meter scalesizes, there is an application note (Howto do it appnote) on the web site [19] thatdescribes exactly how to re scale a meter.Three indicator scales with different divi-sion are required, a fourth for general useto read off is dB is useful. Two scalesneed linear divisions with an initial valueof 0 and a full scale of 10.2 and 3.2 (forvoltage and distortion factor). Becausethe switching of voltage measurement isin 10dB steps these two ranges have theirfull scale deflection at the same pointerangle. In the menu “Scale Design” of thedesign software the “Linearity Exponent”should be set to 1 for these two scales.The two further scales require a logarith-mic layout with increasing compressiontowards the left end division and thus aever greater compression to the left end.The beginning of the scale represents

Fig 17: PCB layoutof the powersupply for theSINAD meter.

Fig 18: Componentlayout of the powersupply for theSINAD meter.


92

Figs 19 - 21: Layout of PCB 1 for the SINAD meter. The full size is 71.1 x152.5mm.


93

Figs 22 - 24: Layout of PCB 2 for the SINAD meter. The full size is 71.1 x152.5mm.


94

minus infinity.In common with the AF standard, the dbscales use 775mV as the 0dB reference.Measuring –34dB on the 30dB rangereads –4dB which is smaller than 0dB. Amathematical relationship is required torepresent this difference and calculate thelinearity exponent to draw the scale. Inaddition the reference point for full scalemust be maintained.where:

xb = reference factorVBez = reference voltage on that linearvoltage scale, in VVmax = full scale on the linear voltage

scale, in VTo draw the logarithmic scales with thesame full scale deflection angle of thepointer with the correct initial value indB the following formula is needed:where:

ys = linearity exponentxb = reference factor from (7)yl = initial scale value, in dBIn order to avoid the unmanageable mi-nus infinity, for yl, a manageable value isselected. A good compromise is -68dB.Any errors are only small [6]. An exam-ple is:

therefore the linearity exponent becomes:

A linearity exponent of 7.6 is the best touse with the “Meter” software [19] Withdivisions every 0.5dB gives 141 divisionlines (-68dB to 2.5dB).The fourth scale, is SINAD in dB show-ing full dBs bearing in mind that aSINAD of less than 5dB hardly makessense. Using the same linearity exponentof 7.6 the scale in dB has 71 divisions(76dB to 5dB SINAD).The software only permits evenly distrib-uted legends to be printed over the entirescale, for example at each third scaleline. For the logarithmic scales thiscauses the legends to be squeezed up atthe left hand end. To overcome this thefinished scale is printed to a pdf or otherpicture format file and the unwantedlegends deleted. The scale shown in Fig26 was produced by this method.

5.3. Construction and mechanicaldetailsGreat importance was paid to the housingof the prototype particularly for service-ability and ergonomic layout of the con-trols. The complete interior is accessibleafter the cover plate and bottom platehave been removed, see Fig 27. Thisgives the best situation for changes andexperiments. The front and back plateform part of the chassis, closer details areshown in Fig 28 and 29. The front panelwas manufactured from copper clad FR4,which combines the advantages of goodmechanical workability and stability, aswell as the possibility for the front panelto be designed with a panel layout pro-gram. After polishing the copper side iscovered with protective varnish. The con-trols are arranged on the basis of use andtechnical criteria, see Fig 30. The impor-

maxVVx Bez

b = (7)

( )

( ) ⎟⎟⎠

⎞⎜⎜⎝

⎛+⋅

=

1

1

log20log

log

yxyxy

b

bs (8)

7598.002.1775.0

==bx

( )

( ) ( )

97.7

687598.0log2068log

7598.0log=

⎟⎟⎠

⎞⎜⎜⎝

⎛−+⋅

−=

dBdB

ys

Fig 25: Adaptor PCB for TO100 case.PCB size 10.3 x 19mm.


95

tance on good accessibility was attachedto individual buttons and switches. Afurther criterion was that each controlfunction should have its own physicalcontrol.The complete metal housing is grounded,which results in good shielding as well asgood safety practice. The individualprinted circuit boards should be isolatedand have their ground surfaces connectedto the common ground with a short pieceof wire from one point. Thus the inputsare floating and their symmetry is main-tained. The input socket on the frontpanel should be connected to the input onPCB 1 with wires of the same length topreserve the symmetry. The points “C”and “D” on PCB 1 and PCB 2 should beconnected with a short piece of coaxialcable with the screen connected toground at one end. In some cases thewiring from S3 can cause undefinedsignals on the 3mV measuring range. Thesolution to this is to solder a 4.7pF (10pFmaximum) across the connections to S3aon PCB 1. For the remaining signalwiring in the instrument, screened cablesuch as RG174 should be used.The ideal connectors for the symmetrical

input are BNO connectors fromHuber+Suhner [20]. Outwardly BNOplugs resemble BNC plugs with theirbayonet fixing but they have two internalconnections, see Fig 31. Suitable testleads can be made, as shown in Fig 32,that have screened leads almost up to thetip. A bundled test lead was planned forthe prototype including two leads with4mm banana plugs, this can be used inthe laboratory.

6.Switch on, alignment andcalibration

All preset potentiometers should be set totheir central position. Before the twomain PCBs are connected to the powersupply, check that with two diodes con-nected in series with the power supplyoutput as shown in Fig 15, the outputvoltage is 14V with a 80mA load. WithPCB 1 and PCB 2 connected and 14Vsupply voltage the supply current shouldbe between 75mA and 95mA.

Fig 26: The meterscale.


96

6.1. Alignment of the true RMSvoltmeterFront panel Switch positions: Voltmeter,CCITT filter out, 1V or 0dB, 50kΩ symm.Equipment needed: Signal generator(sine wave generator), 1kHz and adjust-able output levelFeed 775mV into the symmetrical inputand adjust the preset potentiometermarked “Verstärkung allg” (generalgain), see Fig 21, to give a reading ofexactly 0dB. If the signal generator useddoes not have an output meter, use asuitable measuring instrument; it doesnot need to be a true RMS instrument. Toalign the circuit for the maximum possi-ble symmetry (common mode rejection)

the same test signal should have one sideconnected to the common ground at theinput socket The preset potentiometermarked “symmetrie” (symmetry), seeFig 21, should be varied until the point ofthe smallest pointer deflection is foundon one of the sensitive voltage measuringranges. The same procedure is recom-mended at 50Hz and 20kHz, then the bestcompromise between the three points canbe found. With this adjustment the com-mon mode rejection should over 45dB. Incritical cases the 4.7µF electrolytic ca-pacitor in the lower input arm, see Fig 9,can be changed for 2.2µF or 1µF. A fineadjustment of the general gain potenti-ometer may be required after this.

Fig 28: Detail of panel assembly. Fig 29: Detail of panel mounting.

Fig 27: Picture ofthe completedprototype with thecover removed.


97

If the recommended components havebeen used, no further adjustments shouldbe needed on the other measurementranges. If an examination of individualranges is required, this is best made usingthe following data with a 1kHz inputsignal:

Measuring range Full scale input30V or +30dB 24.5V10V or +20dB 7.75V3V or +10dB 2.45V0.3V or –10dB 245mV0.1V or –20dB 77.5mV30mV or –30dB 24.5mV10mV or –40dB 7.75mV3mV or –50dB 2.45mV1mV or –60dB 775µV

For each range the meter reading shouldbe within ±4dB of that for 0dB input.The frequency response can be examinedin the same way with an input of 2.45Von the 3V or 10dB range at frequenciesof 30, 300, 10,000 and 20,000Hz. Againthe meter reading should be within ±4dB

of that for 0dB input.

6.2. Alignment of the CCITT filterFront panel switch positions: Voltmeter,CCITT filter, 1V or 0dB, 50kΩ symm.Equipment needed: Signal generator(sine-wave generator) with 800Hz adjust-able output levelFeed 775mV into the symmetrical inputand adjust the preset potentiometermarked “Verstärkung CCITT-Filter”(CCITT filter gain), see Fig 24, to give areading of exactly 0dB.If the recommended components havebeen used, no further adjustments shouldbe needed on the other measurementranges. If an examination of the fre-quency response of the CCITT filter isrequired, this is best made using thefollowing data with a 775mV input sig-nal:

Fig 30: The front panel design for the front panel.


98

Measurement Measurement Readingrange frequency0.1V or -20dB 200Hz -1dB (±2dB)1V or 0dB 400Hz -6.3dB (±1dB)1V or 0dB 1kHz 1dB (±1dB)1V or 0dB 3kHz -5.6dB (±1dB)0.3V or –10dB 4kHz -5 dB (±dB)

The filter is within the permissible toler-ance if these measurements are obtained.They are defined in CCITT P53A (nowalso ITU-T recommendation O.41).

6.3. Alignment of the 1kHz notchfilterFront panel switch positions: SINAD &distortion factor, CCITT filter out, 1V or0dBs, SINAD +20dB, 50kΩ symm.Equipment needed: Signal generator(sine wave generator) with985/1000/1015Hz and adjustable outputlevel, (oscilloscope with low ranges andprobe)Feed 1V at 985Hz into the symmetricinput. Adjust all three preset potentiom-eters in the 985Hz notch (Fig 21) alter-nately in small steps until the voltagemeasured on the pole of S4 is thesmallest possible value. Repeat this pro-cedure with signals of 1000Hz and1015Hz and adjust the relevant threepreset potentiometers in the 1000Hz andthe 1015Hz notch filters until the opti-mum result is obtained. With the filterscorrectly aligned the measured signalshould be less than 10mV, see Fig 33.Use a 1:1 probe on an oscilloscope tomeasure the signal; this will give the bestresolution. If a digital storage oscillo-scope is used, the average of severaltraces can be used to get extra resolution.If an oscilloscope is not available thefilters must be adjusted for minimummeter reading.

6.4. Alignment of distortion factormeasurement and SINAD meterFront panel switch positions: SINAD &distortion factor, CCITT filter out, 3V or+10dBs, measured variable selectorswitch see below, 50 kΩ symm.Equipment needed: two signal generators(sine wave generators) with adjustableoutput level - one with 1kHz and asecond with a single frequency between2kHz and 10 kHz, power combinerFeed a 1V signal at 1000Hz and a250mV signal at the second frequencyinto the symmetric input using a powercombiner. With selector switch set to“SINAD” and use the “SINAD” presetpotentiometer, see Fig 24, to set themeasured value to 12dB SINAD. Set theselector switch to “distortion factor 30%”and use the preset potentiometer “30-Prozent-KF” (30% distortion factor), seeFig 24, to set the measured value to 25%distortion factor. Now reduce the 250mVsignal to 70mV, set the selector switch to“distortion factor 10%” and use the pre-set potentiometer “10-Prozent-KF” (10%distortion factor), see Fig 24, to set themeasured value to 7% distortion factor.Reduce the 70mV to 39mV and set the

Fig 31: A BNOconnector.

Fig 32: Picture of the completedSINAD meter with a set of test leads.


99

selector switch to “SINAD +20dB” anduse the preset potentiometer “SINAD+20dB”, see Fig 24, to set the measuredvalue to 8dB SINAD (actually 28dBSINAD).If the output of the signal generators andthe power combiner used are designedfor 50Ω, then use two 100Ω resistors inparallel to provide the correct load. The

attenuation of the combiner and the lowimpedance must be considered; the volt-age values given always apply directly atthe symmetrical input.

7.Conclusion

When measuring distortion factor or SI-NAD the measured signal should bewithin the limits of the measuring rangebeing used, this should be at least onefifth of the total scale. The completedynamic range can only be used on the1mV measuring range with the CCITTfilter activated.This is a valuable instrument for use inthe workshop to measure radio receivers.The circuits used have been explained ingreat detail. The components requiredcan be sourced easily and there are nocomplications when making the PCBsbecause all wired components are used,this means the meter is easily repro-duced. The achievable results are sum-marised in Table 1 and they speak forthemselves!The book shown in [13] is a rich sourceof suggestions and circuits. Thanks aredue to Markus Pichler and ErwinSchimbäck of “The Linz centre of Com-petence in Mechatronics” for long dis-cussions about constructional and techni-cal detail.

Fig 33: The ikHzsignal after thenotch filter.(1mV/Div.,500μs/Div.)

Ranges:Voltage 1mV/3mV/../10V/30V

(0dB, 775mV) Distortion Factor 10% / 30% SINAD 5dB...>43dB

Filter for Distortion factor and SINAD:Notch filter 985Hz...1015Hz

Measurement error:Voltage <0.4dB + tolerance of

meter Distortion factor <0.4dB + tolerance of

meterSINAD <1dB in the upper part of

scale otherwise <2dB + tolerance of meterError with high form factor: <0.1dB for form factor up to 7 and ,0.3dB up to 11 Own distortion factor: <0.2% Frequency response: <0.4dB between 20kHz and 45Hz for 30V to 0.1V. 20Hz to 22kHz with CCITT filter-3dB frequency: 6HzInput: Symetrical 600Ω or 50kΩCommon mode rejection: >55dB at 50HzTempearture range: 5°C to 40°CPower supply: 230V AC < 1.8VA

Table 1: Performance data.


100

8.References and sources ofsupply

8.1. Literature[1] Complete measuring methods forcollecting important AF parameters inreceiver engineering, Ralf Rudersdorfer,Hans Otto Modler, Conference volumefor amateur radio conference Munich2006, pp 135 - 155[2] The radio measuring regulation forHF receivers, transmitters, transceivers,receiving and sending devices, FrancSichla, Ralf Rudersdorfer, Funk 2/2002,P. 68-71; ISSN 0342-1651[3] The radio measuring regulation forFM receivers, transmitters, transceiver,receiving and transmitting devices, FrancSichla, Ralf Rudersdorfer, Radio 3/2002,pp 68-70, ISSN 0342-1651[4] Measuring technique in the communi-cations technology – Edition 1, UlrichFreyer, Carl Hanser publishing house1983, ISBN 3-446-13703-3[5] Paperback of high-frequency engi-neering – Edition 5, Hans HeinrichMeinke, Friedrich Wilhelm Gundlach,Springer publishing house 1992, ISBN 3-540-54717-7[6] Circuit analysers use and understand-ing, Ralf Rudersdorfer, QSP - The Aus-trian transmitter federation 5/2004 -29.Jhg. /5, pp 72-79[7] The handling of modern multirangeinstruments, Ralf Rudersdorfer, LeisureElectronics 1/2003, P. 16-19, __ISSN0342-1651[8] Noise, Hans Hellmuth Cuno, Script ofthe lecture practical electronics, FH-Re-gensburg 9/2003, P. 6.1-6.11[9] Im Größenrausch der Rauschgrößen,Franc Sichla: Funk 12/2001, P. 59-63;ISSN 0342-1651[10] Electrical measuring techniques, Re-inhard Lerch, 1st Edition, Springer pub-lishing house 1996, ISBN 3-540-59373-X

[11] The large electronics workbook,Dieter Nührmann, 6th Edition, FranzisPublishing House 1994, ISBN 3-7723-6546-9[12] All you need to know about SINAD(and its use with 2023 signal generatormeasurements), Dave Owen, IFR Appli-cation note 4/1999[13] Service manual for Stabilock 4020Series radio, Hrsg.Acterna/Schlumberger ENERTEC,Edition 1980, part 3 display section andoperating sector

8.2. Sources of supply[14] Analog Devices , No 3 HorizonBusiness Village, 1 Brooklands Road,Weybridge, Surrey, KT13 0TJ, web:http://www.analog.com[15] Texas Instruments, 800 PavilionDrive, Northampton Business Drive,Northampton, NN4 7YL, web:http://www.ti.com[16] Silonex Inc., 5200 pc. PatrickStreet, Montreal, Quebec, H4E 4N9,Canada, Tel.: ++1 514/7688000, fax.:++1 514/7688889, Internet:http://www.silonex.com[17] Farnell InOne, Canal Road, Leeds,LS12 2TU, web: http://uk.farnell.com[18] Reichelt Electronics e. Kfr, D-26452 Sande, Tel.: +49 (0) 4422/955-333, fax.: +49 (0) 4422/955 111, web:http://www.reichelt.de[19] Tonne Software, James Tonne,WB6BLD, 118 Garnet Lane, Oak RidgeTN 37830, USA, web:http://www.tonnesoftware.com[20] Huber+Suhner GmbH, flour berryroad 6, D-82019 baptizing churches,Tel.: +49 (0) 89/61201-0, fax.: +49 (0)89/61201-162, Internet:http://www.hubersuhner.de/co-de-tk/de-tk-index[21] proMa Technology GmbH, ImLeibolzgraben 16, D-36132 Eiterfield,Tel.: +49 (0) 6672/898 600, fax.: +49(0) 6672/898 - 688, Internet:http://www.proma-technologie.com


101

Because the 2m band is enjoying anincrease in popularity a power ampli-fier with the maximum output of 400 -450W with a supply voltage of 28 –32V is described in this article. Theactive device was chosen as a proven“VHF workhorse” from the manufac-turer SEMELAB. It is very robust; theload VSWR may vary up to 20:1 thusgiving more scope for output networkoptimisation without destroying thesemiconductor. Full data for the tran-sistor can be found in the datasheet [1].The circuit is not complicated yet hasgood characteristics.

1.The circuit

The critical part of a transistor poweramplifier circuit (Fig 1) is the outputnetwork. A mismatch at the output repre-sents a danger to the semiconductor.Therefore the matching circuits shown inFig. 2 will be described in more detail inthis article.A narrow band solution would be suffi-cient for the 144MHz to 146MHz fre-quency range, it is nevertheless advisableto use a wider band solution in order toallow for adjustment tolerances andalignment sensitivity. The input and out-

put matching circuits NW1 and NW4consists of λ/4 matching transformers(25Ω coax cable), they also provide a50Ω asymmetric to 12.5Ω symmetrictransformation. The networks NW2 andNW3 are very simple L/C circuits withinductances made from individual strip-lines.

1.1. Design of the circuit for networkNW3The λ/4 matching transformer NW4 ismade from 25Ω coaxial cable (flexible orSemirigid cable should be used with anoutside diameter not less than 3mm be-cause of the 400W output power that itmust handle) that transforms the asym-metric 50Ω load RL to 2 x R´L = 2 x6.25Ω = 12.5Ω.To make the computation a bit clearer,Fig 3 shows that the real load is dividedfrom 12.5Ω to 2 x 6.25Ω. The networkNW3 must be designed to match theoptimal load resistance Rópt at half of thetotal output of 200W with the internalvalues of the transistor shown in Fig 4.The value Rópt is shown in the SE-MELAB data sheet. The internal loadresistance R´DS can be determined fromthe effective RF Drain voltage V´RF at200W. The effective RF voltage is closeto the supply less “the bottoming volt-age” VK (assumed to be 3V):

Konrad Hupfer, DJ1EE

400W power amplifier for 2m


102

F

ig 1

: Cir

cuit

of th

e 40

0W2m

pow

er a

mpl

ifier

.

L1, L

2=

3 tu

rns

of 1

mm

diam

-et

er c

oppe

r wire

, 6m

mdi

amet

erL3

= 5

turn

s of

0.5

mm

copp

erw

ire

wou

nd

on

a 5.

6Ω2W

resi

stor


103

In parallel with this is an output capaci-tance Cob = 190pF (see data sheet). Theeffective capacity is:

The inductance of the drain connection istaken from the SEMELAB data and is0.63nH.

1.1.1. Calculating the value of L1 (Fig4)L1 is made from a small low impedancestripline. The transformation procedurecan be seen using a Smith chart. TheSmith chart program by Fritz Dellsperger[2] was used for this process and found tobe extraordinarily helpful for this task.The Smith chart in Fig 5 is standardisedfor the terminal resistance R´2 = 6.25Ω,the internal load resistance R´DS = 1.56Ωand the goal value R´L = 6.25Ω can beseen. The chart is simplified to make it

easier to understand. Point 1 (Z1) on thereal axis represents R´DS = 1.56Ω. Cób =250pF is point 2 on the conductancecircle. The bond inductance is a seriesinductor L = 0.63μH and is representedas Z2 = 1.4 + j0.5Ω giving point 3. This isthe impedance at the drain connector lug;Z3 = 1.4 + j0.1Ω. This “transistor connec-tion resistance” is now transformed withan inductance, formed by striplines, S3and S4, on the circle around the centre ofthe diagram to point 4 giving Z4 = 1.4 +j2.6Ω. This is the intersection with theconductance circle corresponding to thetarget resistance Z5 = 6.25 +j0Ω.The stripline width W was selected as8mm and the substrate thickness, H, usedwas 0.83mm. The characteristic imped-ance Z0 is given by:

This is for a dielectric constant εR = 1. Inreality the characteristic impedance ofthe striplines S3 and S4 are 16.5Ω be-cause the substrate has a dielectric con-stant εR = 3.3. The inductance of thisstripline is 2.8μH and is “quasi-stable”the length of the line is l = 30mm. Theinductance of this very short line isindependent of εR!If a 320pF capacitor is connected frompoint 4 to ground then a real resistance of6.25Ω is achieved (point 5 on the Smithchart). This matches one half of the

Fig 2: Block diagram of the 400W 2m power amplifier.

( ) ( )Ω=

−=

⋅

−=′ 56.1

400328

2

22

o

KDS P

UUR

pFCC obob 2501903.13.1 =⋅=⋅=′

Fig 3: Network NW3.

mmmm

HW

83.08

= gives 28.4Ω


104

power transistor to 6.25Ω. For the push-pull circuit twice this value is used, seeFig 6. Because the “virtual ground” doesnot exist, the two single capacitors in Fig6 are combined into a single capacitor (2x 320pF in series = 160pF).If the output networks are used with thecalculated values the output power willbe approximately 250 to 300W. Thetransistor equivalent circuit used do notexactly correspond to the actual valuespartly because of differences in themounting of the transistor. In order to

achieve the maximum output power of400W with good efficiency (~70%),changes in the striplines S3 and S4 aswell as the capacitor C5 are required.This is shown in Fig 1 by the variablearrow on these components. This tuningwas carried out while watching the effi-ciency and the values shown in Fig 7were the result.The characteristics of the output stagewith a supply voltage VDS of 28V and32V are shown in Fig 8.

Fig 4: The equivalent circuit of 1/2 D1030UK.

Fig 5: Smith chartshowing thetransformation to6.25Ω.


105

1.2.1. Measured performanceThe performance of the amplifier wasmeasured with the equipment shown inFig 13. The results for a supply voltageof 28V are shown in Table 1 and for a32V supply voltage in Table 2.

1.3. Input matchingThe input impedance of ½ D1030UK hasthe typical values shown in Fig 9.R´GS is found, like R´DS, in the data sheet[1]. R´GS can be transformed to 6.25 +j0Ωusing a simple L/C circuit. The inductoris again made from a stripline. Thetransformation to 50Ω is achieved withλ/4 matching transformers made from25Ω coax cable; thin cable is sufficienthere because the power is low. Using thevalues calculated using a Smith chart, theinput matching from 5:1 to 2:1 isachieved. The striplines and capacitorsmust be optimised as shown in Fig 10.Unlike the output the input can betrimmed to give a return of zero.

Fig 7: Dimensionsof striplines S3 andS4 after optimisa-tion.

Fig 8: Output characteristics of theD1030UK.

Fig 6: Circuit showing the two the two halves of the push-pull amplifier.


106

2.Mechanical construction

Fig 12 shows the construction of theamplifier. The 50 x 150mm baseboard isfitted to the large 150 x 120 x 80mmheatsink as shown in Fig 11. The base-board is 1.6mm thick copper clad FR4material; the upper surface is the RF andDC ground. So that fitting the powertransistor does not interrupt this ground,a very thin copper foil (~ 0.1mm) is fittedinto the cutout for the transistor. Thestriplines are individually cut from0.83mm thickness material and solderedto the baseboard ground surface. The DCwiring can take place as desired. Goodheat transfer between the foil, transistorand heatsink is extremely important.Spread thermal compound on the indi-

Fig 9: The input equivalent circuit.

Table 1: Measured parameters at 28V.

UDS = 28V, ID = 1.7A, f = 145MHzCW: Po 400W

PST 8.9WEfficiency 75% U´DS 58V

Harmonics: 2 x f0 -49dB3 x f0 -44dB

SSB power: PEP0 300WPEPST 4.7WFMD3 -29 to 31dBFMD5 -35BEfficiency 49%

Fig 10: Dimensionsof striplines S1 andS2 after optimisa-tion.

Table 2: Measured parameters at 32V.

UDS = 32V, ID ≈ 1.8A, f = 145MHzPo/W PST/W ID/A G/dB Effic%50 0.7 6.8 18.7 24100 1.35 9.2 18.9 200 2.7 12.7 18.7 300 4.3 15.9 18.5 350 5.0 17.0 18.5 400 5.9 18.1 18.3 71450 8.2 19.6 17.4 74SSB power PEPo 350W

IM3 -30dBMaximum drain voltage:(DC + RF) ≈ 65V for Po = 480W


107

vidual surfaces very thinly because onlythe pores in the metal are to be filled!The heatsink should be well cooled withthe aim of a maximum flange tempera-

ture of 60°C with a PEP output of 300W.

3.References

[1] SEMELAB, web:www.semelab.co.uk. The data sheet forthe D1030UK can be downloaded fromhttp://www.semelab.co.uk/pdf/rf/D1030UK.pdf[2] Fritz Dellsperger, Smith chart pro-gram and Smith chart tutorials, email:f r i t z . d e l l s p e r g e r @ i s b . c h , w e b :http://www.fritz.dellsperger.net/

Fig 13: Block diagram of the performance measuring equipment.

Fig 12: Themechanicalconstruction ofthe 400W 2mpower amplifier.

Fig 11: The heatsink and baseboard.


108

1.Introduction

Most of radio amateurs have frequencycounters with reference oscillators thatare not temperature compensated andthey are made with bad quality crystals.A special problem is the ageing of crys-tals that can take up to 10 years. Becauseof this frequency counters can deviate bymore than a hundred kHz at a frequencyof 10GHz and are really only suitable for100MHz or less. A simple and inexpensive solution is aGPS controlled frequency standard, oftenreferred to as a GPSDO (GPS Disci-plined Oscillator). It is a frequency stand-

ard that calibrates itself using GPS.

2.Block diagram

The block diagram of a GPS controlledfrequency standard is shown in Fig 1.

3.Circuit diagram

The circuit diagram of a GPS controlledfrequency standard is shown in Fig 2.The circuit diagram is simple as it can be.

A GPS controlled frequencystandard

Zeljko Bozic, S52ZB

Fig 1: Blockdiagram of theGPS controlledfrequencystandard.


109

F

ig 2

: Cir

cuit

diag

ram

of t

he G

PSco

ntro

lled

freq

uenc

y st

anda

rd.


110

The Navman TU60-D120-041 Jupiter-TGPS receiver [1] has a high accuracy10kHz output synchronised to UTC. A10MHz C-MAC TCVCXO CFPT-9001[2] is divided down to 10kHz and phaselocked to this GPS signal. Both outputs(10MHz and 1PPS) are buffered withinverters.

The PLL filter components are designedfor ωn=0.239 rad/s, ς=0.707 andKo=67Hz/volt. R2 and C2 are not used.The GPS receiver can be controlled viathe RS232 port and a 1PPS signal isavailable on DCD that can be used by acomputer to keep precise UTC time to afew tens of ns.

4.Construction

The frequency standard is constructed ona double sided PCB made from a50.8mm x 55mm piece of 1.6mm thickFR4 laminate. It has plated throughholes, solder resist and lead free tinning.The PCB layouts are shown in Figs 3 and4, the component layout is shown in Fig5 and the parts list in Table 1. The GPS receiver used is a NavmanTU60-D120-041 Jupiter-T which hasboth 1PPS and 10kHz outputs. It wasdesigned especially for precision timingapplications. In 2005 Navman discontin-ued manufacturing the Jupiter-T. Insteadof the Jupiter-T you can use the standard

Fig 3: Top side PCB layout for theGPS controlled frequency standard.

Fig 4: Bottom side PCB layout for theGPS controlled frequency standard.

Fig 5: Component layout for the GPScontrolled frequency standard.


111

Jupiter receivers TU30-D140-xxx andTU30-D410-xxx, which also have a10kHz output. They are similar in sizebut the connector and pin out are differ-ent. These receivers are often available

on eBay. Before purchase it is importantto ensure that the firmware is v1.18onwards. The firmware version is printedon a label and is also emitted at start upin an ASCII plain text NMEA message.The default protocol of the GPS receiveris Motorola UT+ “@@”. Navman binaryprotocol and NMEA protocol are alsoavailable. The GPS antenna is an external activeantenna with 5V supplied through thecoaxial cable. It is important that the GPSantenna has the fullest possible view ofthe sky. Getting the maximum number ofsatellites in view will get the best timingperformance. A better quality oscillator such as anOCXO can be used instead of the C-MAC TCVCXO but the PLL filter com-ponents must be recomputed. The oven-controlled oscillator has considerablybetter performance than a miniatureSMD TCVCXO, but at the cost of greaterpower consumption and increased size.Good but big OCXOs can be found in oldNMT base stations and also on eBay.The 10MHz output is square wave atHCMOS level, an output level of

Fig 6: Picture of the GPS controlledfrequency standard PCB.

Fig 7: Top view ofthe completed GPScontrolledfrequencystandard.


112

+15dBm. The PLL IC is a 74HC7046A[3] [4], it is an improved version of thefamiliar 74HC4046A. The PC3 functionis replaced with an improved lock detec-tor. The serial port driver IC is anSP233A [5], a low cost version of thefamiliar MAX233A.The frequency standard status is shownby two LEDs. The green LED “Heart-beat” flashes at 1PPS and the red LED“Unlock” lights up when the PLL isunlocked. The value of the lock detector

capacitor C1 is dependant on the PLLfilter components. If the PLL filter com-ponents are changed, the value of C1must also be changed. The operating voltage of the frequencystandard is 9 to 35V and the currentconsumption should be around 250mAincluding GPS receiver and GPS activeantenna.The frequency standard is built into acustom made housing 130mm wide,110mm deep and 30mm high made from

Fig 8: Front viewof the completedGPS controlledfrequencystandard.

Fig 9: Rear view ofthe completed GPScontrolledfrequencystandard.


113

1mm thick aluminium plate. Figs 6, 7, 8and 9 show the prototype GPS controlledfrequency standard.

5.Results

First measurements of the frequencystandard show that it has accuracy of 1 x10-10 after 1 hour and about 1 x 10-11 after24 hours. The test equipment was anAgilent PSA E4445A (1Hz resolution)spectrum analyser locked on a 10MHzfrequency standard OSA 5230 from Os-cilloquartz. Measurements with the spec-trum analyser were on the 100th (1GHz),200th (2GHz) and 500th (5GHz) har-monic. More accurate measurementswere with an oscilloscope (Lissajous fig-ure).GPS satellites can be monitored usingone of the NMEA 0183 data programssuch as Tac32 [6] or VisualGPS [7].These programs give a graphically dis-play of specific NMEA 0183 sentences.I wish to thank James Miller (G3RUH)[8] for help and support, Robert Vilhar(S53WW) for measurements and StojanKuret (S51WI) for help in all phases ofproject.

6. References

[1] Navman datasheet: Jupiter-T,web:http://www.jrmiller.demon.co.uk/projects/docs/10039C.PDF[2] C-MAC datasheet: CFPT-9000Series, web:http://fcp.rakon.com/pdf/cfpt-9000.pdf[3] Philips datasheet: 74HC7046A, web:http://www.datasheetcatalog.com/datasheets_pdf/7/4/H/C/74HC7046A.shtml[4] TI Application Note AN8823, web:http://focus.ti.com/lit/an/scha003b/scha003b.pdf[5] Sipex datasheet: SP233A, web:http://www.sipex.com/Files/DataSheets/sp232_312a.pdf[6] Tac32 software from CNS SystemsInc., web:

R1 30kΩ 0805R2 - 0805, not usedR3, R4 560Ω 0805

C1 4n7 0805 C0G (NP0)C2 - C, SMD tantalum 35V, not usedC3 100μF C, SMD tantalum 35VC4, C5, C7, C8, 100n 0805C10, C11, C12, C13C6 4.7μF C, SMD tantalum 35VC9 10μF C, SMD tantalum 35V

D1 LED Red 3mmD2 LED Green 3mmD3 SM4001 MELFU1 74HC7046AD SO-16U2 CFPT-9001 SMD 7x5x2mm, 10- pinU3 74HC04 SO-14U4, U5 74HC390 SO-16U6 SP233ACT SO-20 wideU7 L7805CV TO-220

J1 SIL socket 6-pinJ2 SIL socket 4-pinJ3, J4, J5 SIL socket 3-pinPCB - 50.8 x 55mm, FR4 1.6mmHousing - Aluminium plate 1mm

Table 1: parts list


114

http://www.cnssys.com/cnsclock/Tac32Software.html[7] VisualGPS software from VisualGPS LLC. web:http://www.visualgps.net/VisualGPS/default.htm[8] Simple GPS Stabilised 10 MHz Os-cillator, James Miller, G3RUH,web:http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm[9] A GPS-Based Frequency StandardQST July 1998, Brooks Shera, W5OJM,web:http://www.rt66.com/~shera/QST_GPS.pdf

[10] Time and Frequency MeasurementsUsing the Global Positioning System,Michael A. Lombardi, Lisa M. Nelson,Andrew N. Novick, Victor S. Zhang:,Cal. Lab. Int. J. Metrology, pp. 26 -33,(July-September 2001), web:http://tf.nist.gov/general/pdf/1424.pdf


115

1.Spectrum analyser project 2007update

Since the development of the wide bandVCO almost 10 years ago, the wholespectrum analyser project with all relatedaccessories: tracking generator, harmonicconverter, storage normaliser, markercounter and accessories developed byother experimenters (Darko S57UUD)have been published in many differentplaces: magazines “VHF Communica-tions Magazine issues 4/1998 – 4/1999”,“AMSAT-DL Journal”, “CQ ZRS” andthe book “Beacon 99”. It is reasonablybelieved that the project has been suc-cessfully reproduced in hundreds of units(33 sets of PCBs have been supplied toVHF Communications Magazine readers– Andy). Since the spectrum analyser project isstill very interesting for many radio ama-teurs and other radio and electronic ex-perimenters, I decided to republish myoriginal articles in English on my per-sonal web page [1]. Unlike printed maga-zines, the web allows to publish manycolour pictures of all interesting details ofthe project in addition to PDF, PCB andsoftware files. Last but not least, updatesto the project are really simple on theweb. (This article has been taken from

the web site, if you are interested in theproject of want PCB or software files youshould visit the site – Andy) In this article a simple and inexpensiveLCD oscilloscope to be used as a displayfor the spectrum analyser will be pre-sented. Although a small LCD screen isunable replace a good analogue oscillo-scope, an LCD may be very useful infield measurements under strong daylightconditions, for battery operation or sim-ply when the available oscilloscope isrequired for a different measurement atthe same time. Before describing the LCD oscilloscope,some necessary updates to the spectrumanalyser project will be presented. Asexpected and also according to the feed-back, the most difficult part to reproduceis the wide band VCO. This difficulty isin partly due to component tolerancesand in partly due to assembly (soldering)tolerances. In particular, the BB833 var-actor diodes were found to have ratherwide manufacturing tolerances both inthe capacitance range and in the series(loss) resistance affecting the Q of thevaractor. As already explained in the VCO article,two different printed circuit boards areproposed for the wide band VCO. Start-ing with unknown components, oneshould always assemble the “narrowstripe” wide band VCO first (PCB#1).The “wide stripe” PCB#2 should only be

Matjaz Vidmar, S53MV

LCD oscilloscope for spectrumanalysers. Plus update onspectrum analyser project


116

used if the frequency coverage obtainedwith PCB#1 is insufficient.If output power of the wide band VCOdrops in the middle of its frequencyrange between 2.5GHz and 3GHz or theVCO simply stops oscillating at all inthis frequency range, excessive losses inthe BB833 varactors are suspected. Be-sides poor varactor Q (try getting betterBB833 varactors from a differentsource), the interdigital feedback filtermay be misaligned due to solderingtolerances. In the latter case, a rathersimple solution is to try to trim the lengthof the open end of the central finger of

the interdigital filter.Although the wide band VCO design wasalso published in a well respected andwidely known professional magazine“Microwave Journal” back in 1999 [2], ittook several years for the professionals tofully understand the operation and capa-bilities of this VCO design. Multipleresonator multiple varactor microwaveVCO modules only became availablerecently on the professional market. Fi-nally, there is a valid and inexpensivealternative to the YIG oscillator! Ofcourse, these inexpensive commercialVCO modules can be used in place of the

Fig 1: The spectrum analyser and all of the additional modules.


117

described wide band VCO.After using the spectrum analyser and allrelated accessories shown in Fig 1 formany years, some long-term effects werenoted as well. The worst seems to be thecorrosion caused by outgassing chemi-cals from the antistatic foam used as amicrowave absorber in high frequencyshielded modules. Higher conductivityfoam makes a better microwave absorber,but unfortunately more chemicals causemore corrosion as well! Fortunately, themicrostrip circuits of the spectrum ana-lyser continue to work correctly withlittle if any degradation, although theyreally look ugly, see Fig 2.A less frequent problem is a long termfailure of ATF35176 or similar HEMTdevices. If the voltage between drain andsource is kept too high, the drain currentslowly decays. This decay may be veryslow, just a few percent per week, but itis cumulative, irreversible and continuesdown to zero leading to a total failure ofthe circuit!

In this project, HEMTs are used in thebuffer stages of wide band VCOs and inthe buffer amplifiers inside the trackinggenerator. The described failure can bedetected as a rise of the drain voltage. Ifthe d ra in vo l t age r i s e s above+3V...+3.5V, the decay will speed up andthe HEMT will have to be replaced soon,possibly with better devices.

Some electronic parts became obsolete orhard to get. The most difficult seems tobe the INA10386 MMIC amplifier thathas no direct replacement. Worst of all,defective factory rejects are shipped forsome obsolete parts, like the uA723voltage regulator. Some uA723 regula-tors have a very noisy voltage referencewith very large 1/f “popcorn” noise. Thelatter may be large enough to disturbeven large signal circuits like the videoamplifier inside the spectrum analyzer.

Besides getting better uA723 regulators,there is a simple solution to this problem.The internal 7V voltage reference, avail-able on pin 6, should be filtered beforeuse. If the latter is fed through a resistorto pin 5 (non-inverting input), then asingle electrolytic capacitor from pin 5 toground (pin 7) solves the problem.

2.LCD oscilloscope for spectrumanalyser

LCD modules are probably the mostpopular displays today, ranging fromsimple numerical displays with few sin-gle colour digits to large, graphical, high

Fig 2: Picture of a wideband VCO after 10 year showing the corrosion on themicrostrip circuit.


118

resolution, full colour computer moni-tors. A medium resolution, single colourgraphical LCD module is required in anoscilloscope display for a spectrum ana-lyser. Of course, a simple solution wassought for the spectrum analyser project.

Medium resolution graphical LCD mod-ules may have different interfaces. Thesimplest LCD modules have no built incontroller. These modules only have shiftregisters associated with the columns androws of the display. The user has toprovide a continuous data flow to refreshand multiplex the display content. Thisrequires a powerful microprocessor withlots of memory or in other words acomplicated circuit with many chips.LCD modules with a built in controllerproviding the refresh and multiplex of thedisplay are much simpler to use. Themost popular alphanumeric controller iscertainly the Hitachi HD44780 that evenhas some very limited graphical capabili-ties. Older graphical LCD modules usethe Toshiba T6963 controller with anexternal 8kbyte RAM (usually 6264).These modules also require a negativevoltage supply for the LCD and a high-voltage AC source for the electro-lumi-nescent backlight.Recent graphical LCD modules use the

Samsung KS0107 and KS0108 chips.The KS0107 is the clock generator, scans64 rows and may drive multiple KS0108chips. The KS0108 drives the columnsand includes storage for up to 64x64dots. The microprocessor simply writesinto the RAM inside the KS0108 chips.These modules usually include one ortwo 7660 chips to generate the requirednegative voltage LCD supply of -5V or -10V on-board the LCD module. FinallyLEDs are used for the backlight so thatthe module can be operated from a single+5V supply. A 128x64 graphical LCD module is usedin the described LCD oscilloscope. Thismodule includes one KS0107 and twoKS0108 chips. All three controller chipsare usually bonded directly to theprinted-circuit board and covered withdrops of black resin, see Fig 3. The LCD oscilloscope is built around aPIC 16F876A microcontroller. The latterincludes an A/D converter and steers thegraphical LCD module directly. The cir-cuit diagram of the LCD oscilloscope isshown in Fig.4.The LCD oscilloscope requires two sig-nals from the spectrum analyser; theanalogue video (including blanking) andthe trigger. Due to the limited resolution

Fig 3: The rearview of an LCDdisplay showing thebonded ICscovered with blackresin.


119

of the LCD, the analogue signal is over-sampled. Each column on the LCD iscomputed from eight consecutive sam-ples and the last sample from the previ-ous column. The A/D sampling rate is setso that the whole sweep (128 columns or1024 samples) corresponds to about20ms or in other words the fastest rate ofthe described spectrum analyzer.

Additional inputs to the microcontrollerare provided for two switches and onepush button. One switch is used to turnon the grid while the other selects a 80dB(4V) or 40dB (2V) full scale range. Thescale is adjusted by two trimmers thatdefine the reference voltages of the A/D

converter. The pushbutton activates aMIN/MAX memory that is very usefulwhen observing wide deviation slow FMsignals or infrequent pulsed signals onthe spectrum analyser. All outputs of the PIC 16F876A haveseries 1.8kohm damping resistors to re-duce the amount of radio interferenceradiated by the microcontroller. Thesame resistors are also used to operatethe data bus to the LCD module bi-directionally, simplifying the program-ming of the PIC 16F876A. In particular,the RC0-7 pins are always programmedas outputs. In order to read the status ofthe KS0108 controllers, the DB7 line is

Fig 4: Circuit diagram of the LCD oscilloscope.


120

also fed back to the RB1 input. The display refresh rate is limited bythree factors; the A/D conversion speed,the PIC computing power and the waitcycles required by the KS0108 control-lers. In the LCD oscilloscope, all threefactors are of the same order of magni-tude limiting the refresh rate to about35Hz - 40Hz and thus providing a livepicture of the input signal. A minimumclock frequency of 20MHz is required forthe PIC 16F876A for this purpose.A 24MHz clock for the 16F876A wastherefore selected to allow the spectrumanalyser sweep time to remain at 20ms atall times with some safety margin. This isconsistent with the resolution of thedisplay (128 columns) and available filterbandwidths and spans of the spectrumanalyser. Slower signals can be observedwith the MIN/MAX function, using 256

bytes of the internal RAM inside the16F876A. The minima and maxima aresimply accumulated as long as the MIN-MAX pushbutton is kept depressed. The same 256 bytes of internal RAM arealso used for 128 minima and 128maxima storage during normal operationwithout the MIN/MAX function. TheA/D routine runs under interrupts trig-gered by an internal timer inside the PIC16F876A. The main program is an end-less loop computing and refreshing theLCD content from the intermediate stor-age. Yet another interrupt is used for thetrigger function. The LCD oscilloscope is fully compat-ible with all described accessories; track-ing generator, storage normaliser andmarker counter. The 40dB scale is in factintended for reflection measurementswith the tracking generator and an addi-

Fig 5: Circuit diagram of the power supply for the LCD oscilloscope andspectrum analyser modules.


121

tional, external directional coupler. Theover-sampling of the video signal allowsa proper display even of the 455kHzmarker. The LCD oscilloscope includes its own+5V regulator (78L05). The latter alsoprovides power to the backlight LEDsthrough two 10ohm current limiting re-sistors. Current limiting resistors andother protection components are alsoprovided on all inputs to the module.

3.230V mains power supply

Since the described LCD oscilloscope issmall and simple, the remaining spacebehind the LCD module is used for a230Vac mains power supply, providing12V for the LCD oscilloscope, spectrumanalyser and all accessories. The circuitdiagram of the power supply is shown on

Fig.5.The discrete component regulator hasmany advantages over integrated circuits;very low drop out voltage, groundedheatsink of the power device and last butnot least, no additional protection diodesare required in the case of (parallel) 12Vbattery operation! Since the spectrumanalyzer is sensitive to low frequencymagnetic fields, it is imperative to use atoroidal core transformer in the powersupply.

4. Assembly of the LCDoscilloscope

The first thing to do is to check theavailability and size of suitable LCDmodules with the KS0107/KS0108 con-troller chips. Do not forget to check the

Fig 6: PCB layout fro the LCDoscilloscope.

Fig 7: PCB layout for the powersupply.


122

presence of the 7660 voltage generator aswell as the type of backlight available.Last but not least, check the connectionsof the available LCD module! The printed circuit board (Fig 6) wasdeveloped for Wintek LCD modules.Other manufacturers may use a differentpin out requiring the swapping of somewires. In particular, the Vdd (+5V) andVss (GND) connections may beswapped! CS1 and CS2 may be activelow or active high, since each KS0108actually has three such inputs (two in-

verted) and just one is brought out to theconnector. The LCD oscilloscope is built on twoprinted circuit boards as shown in Fig.6and Fig 7. The PIC 16F876A and associ-ated components are installed on an80mm x 60mm single sided printed cir-cuit board. The power supply regulator isbuilt on an 80mm x 40mm single sidedprinted circuit board. Both circuit boardsare etched on 1.6mm thick FR4 glassfibre epoxy laminate. Pictures of the

Fig 8: Picture ofthe LCDoscilloscope PCBfitted withcomponents.

Fig 9: Picture of the chassis fitted with the LCD oscilloscope and power supply.


123

completed boards are shown in Figs 8and 9. Figs 10 and 11 show the LCDscreen in operation.The LCD oscilloscope has the samedepth (240mm) as the spectrum analyser[3] or [4]. The width is 110mm while theheight is 70mm considering the size ofthe available LCD module. All modulesare mounted in a single panel . Thebottom of the box is simply a piece of1mm thick aluminium sheet, bent in the

form of a “U”. The cover is a similar “U”made from 0.6mm thick aluminiumsheet.As well as the DIN connector, twoadditional BNC connectors for the video+ blanking and trigger signals are pro-vided on the back panel to connect anadditional, high quality, analogue oscillo-scope. Although the LCD oscilloscope isprotected from over voltages on all inputseven in the power down state, the LCD

Fig 10: The LCD oscilloscope showing the PCB and display.

Fig 11: Picture ofthe LCDoscilloscope in use.


124

oscilloscope should be powered at alltimes to avoid loading the outputs of thespectrum analyser. As well as the LCD module, twoswitches (grid on/off and scale40dB/80dB), the MINMAX pushbuttonand the LCD contrast potentiometer arelocated on the front panel. The lattercould be omitted, since LCD modulesfrom many different manufacturers werefound to operate best at the maximumvoltage. This means that the Vo input issimply tied to the Vee output through the82kohm resistor on the printed circuitboard and no additional potentiometer isrequired. The LCD oscilloscope will display cor-rectly input voltages between about+0.7V and +5.5V due to the emitterfollower in front of the A/D converter.Therefore the trimmers inside the videoamplifier of the spectrum analyser mayneed to be readjusted first, using a reli-able analogue oscilloscope. The trimmersinside the storage normaliser and markercounter may need some minor readjust-ments as well. Finally, the 80dB scale of the LCDoscilloscope can be adjusted with the twotrimmers providing the reference volt-ages to the A/D converter. Please notethat the operation of the LCD oscillo-scope is slightly different between the80dB mode and the 40dB mode. In the

80dB mode, the trace is always visibleand saturates on the bottom or top ofscreen. In the 40dB mode, the trace runsoff of the screen and only the central partof the original 80dB scale is displayed.

5.References

[1] LCD oscilloscope for spectrum ana-lyser, Matjaz Vidmar web pages, web:http://lea.hamradio.si/~s53mv/spectana/lcd.html[2] A Wideband, Varactor-tuned Micros-trip VCO, Matjaz Vidmar, MicrowaveJournal, 6/1999 pp 80-86. Also – Wide-band and low noise microwave VCO,Matjaz Vidmar, VHF CommunicationsMagazine, 4/1998, pp194 – 209. [3] Spectrum analyser 0...1750MHz,Matjaz Vidmar, VHF CommunicationsMagazine, 1/1999, pp 2 - 30. [4] Matjaz Vidmar: "Spektrum-Analyzervon 0 bis 1750MHz, Teil 1: Aufbau derBaugruppen", pages 18-30/4-99,"Spektrum-Analyzer von 0 b is1750MHz, Teil 2: Anzeigebaugruppenund Abgleich", pages 18-29/1-00,AMSAT-DL Journal.

With over 700 members world-wide, the UK Six Metre Group is the world'slargest organisation devoted to 50MHz. The ambition of the group, through themedium of its 56-page quarterly newsletter 'Six News' and through its web sitewww.uksmg.com, is to provide the best information available on all aspects of theband: including DX news and reports, beacon news, propagation & technicalarticles, six-metre equipment reviews, DXpedition news and technical articles.Why not join the UKSMG and give us a try? For more information contact thesecretary: Dave Toombs, G8FXM, 1 Chalgrove, Halifax Way, Welwyn GardenCity AL7 2QJ, UK or visit the website.

The UK Six Metre Groupwww.uksmg.com


125

MicroStripes

This is not individual strip line simulatorbut a matured 3D simulation softwarepackage for draft layout and simulationof microwave modules and antennas

Address: http://www.microstripes.com

Microwaves101.com

An interesting and humorous collectioninformation about “Microwaves”. Theoperators of the site stress that not every-thing is fully developed, but it can beused! The site is well designed withmany alphabetically sorted articles, top-ics, Application Notes and software. It isa wealth of information.

Address:http://www.microwaves101.com

Azimuth

This is a young American companyoffering software test suites for the wire-less communications.

Address:http://www.azimuthsystems.com

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!

Gunthard Kraus, DG8GB

Internet Treasure Trove


126

12GHz Divide by 1000 prescaler by Alex-ander Meier, DG6RBP from issue 4/2003Ready built in diecast housing £160.00with SMA connectorsAlso availabe an 18GHz £230.00version of the divide by 1000 prescaler

Ready built units, Kits and PCBs availablefor recent projects

Microwave milliwattmeter by Bern Kaa,DG4RBF from issue 4/2006PCB £27.50Micro-controller with software £95.00

Display unit for the power detector byDJ8ES, Alexander Meier by DG6RBPfrom issue 1/2007PCB £25.00Micro-controller with software £65.00

Spectrum Analyser by Matjaz Vidmar,S53MVThis project was first published in issue4/1998 to 4/1999 (now included on CD1,see back cover) The LCD oscilloscope is inthis issue.Complete set of PCBs for all modulesincluding the LCD oscilloscope £70.00

To order use either the web site www.vhfcomm.co.uk or contact K M Publications: Tel +44 (0)1582 581051, email [email protected]


127

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 Summer Edition 2007-Q2

VHFCOMMUNICATIONS

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]

Documents

A Publication for the Radio Amateur Worldwide Especially ......(Y3Fe5O12). The sphere is approxi-mately 0.5 to 1mm diameter. If this is exposed to an external magnetic field (H 0)