Noise measure

7/27/2019 Noise measure

1/8

May/June 2007 39

2258 June RdCourtenay, BC [email protected]

An Automatic Noise-FigureMeter

Jim Koehler, VE5FP

Here is a project that will automatically measure the noisefigure of your preamplifier projects. It has an operating rangefrom 3 MHz to over 400 MHz.

Introduction

Anyone interested in building low-noise

pre-amplifiers will want to measure the noisefigure of the device. Commercial noise-figuremeters, often called Precision AutomaticNoise Figure Indicators (PANFI), exist, butthey are expensive even on the surplus orused market because of the demand for them.There have been a few published constructionarticles in the ham literature, but they areeither overly complex or dont compare tocommercial test instruments.1 This article de-scribes the construction of a simple version,suitable for home construction, but whichhas many features of the best commercialinstruments. See Photo A.

Theory

A good background to noise-figure mea-surement is given in a Hewlett-Packard (nowAgilent Technologies) Application Note: AN57-1. This note is available on the AgilentWeb site, and is worth reading.2 Noise figure,in an amplifier, is a measure of how muchexcess noise an amplifier adds to any incom-ing signal. The lower the noise figure, the lessnoise it adds. Noise figure can be expressedin a number of ways, but I prefer to think ofit in terms of temperature. Every passive re-sistive electronic component generates some

noise because of the random motion of chargecarriers in it. The greater the temperature,the greater the motion of the charges, andtherefore, the greater the random noise powergenerated by the component. The noise powergenerated by any component depends only onits temperature and the bandwidth that you are

considering; it is given by the equation:

P = kTB [Eq 1]

where k is Boltzmanns constant, which isequal to 1.381023 joules / kelvin,B is thebandwidth in Hz and Tis the temperature inkelvins. A perfect amplifier would add nonoise to incoming signals, but real ones do.We can specify this excess noise in termsof the temperature of a resistor at the tem-perature that would give that same amountof noise. This temperature, called the excessnoise temperature, Te, is a measure of howgood the amplifier is; the lower the excessnoise temperature, the better the amplifier.The amplifier may also be characterized bya noise figure, which is the ratio of the excessnoise temperature (plus room temperature) toroom temperature.

NF =T +T

T

e r

r

( )

[Eq 2]

where Tr is the room temperature. This ratiois often expressed in dB. An ideal amplifierwould have a Te of zero and therefore a noisefigure of 1.00, or 0.0 dB.

Modern noise-figure meters measure thenoise figure of an amplifier by connecting anoise source, which can be turned off and on,to the amplifier input. A noise source is justthat; a device with an output that is broadbandnoise of a known level, corresponding to aresistor at some temperature. Noise sourcesare characterized by their effective noiseratio, ENR, which is given by:

ENR =T T

T

h r

r

( )

[Eq 3]

1Notes appear on page 46.

Photo A The front panel of the VE5FP automatic noise-figure meter.

VE5FP


2/8

40 May/June 2007

where Th is the hot temperature that a resistorwould have to produce the same noise poweras the noise source when it is switched on.When the noise source is switched off, it pro-duces only the amount of noise that a resistorat room temperature, Tr, would produce.

Now, imagine an amplifier with a noisesource at its input and a power meter at itsoutput to measure the power coming out ofthe amplifier. When the noise source is turned

off, the power going into the amplifier willconsist of the power from the noise source atroom temperature plus the amplifiers excessnoise power. The output power will be thissum multiplied by the amplifiers amplifica-tion factor. When the noise source is turnedon, the input power to the amplifier will bethe noise power of the noise source when itis on, plus, again, the excess noise power dueto the amplifier alone; the output will be thisnew sum again multiplied by the amplifica-tion factor of the amplifier. The ratio of theon-power to the off-power at the output ofthe amplifier is the Y-factor:

Y =T +T

T +T

h e

r e

( )( )

[Eq 4]

Noise-figure meters determine Te by mea-suring the Y-factor and by knowing the valueofTh and Tr . Tr is normally assumed to be290 K, so, rearranging Equation 4, we get;

T =T Y

Ye

h( )

( )

290

1

[Eq 5]

A PANFI is therefore just an instrumentthat has a power detector to accurately mea-sure the output noise power of an amplifierand a means of turning the noise source onand off. It must then do the mathematics todisplay the excess noise temperature eitheras a temperature or as a noise figure. Noisesources are mostly solid state devices andmost are designed to be turned on whenthey are supplied with +28.0 V dc. A moredetailed discussion of how a PANFI worksis given in the Appendix.

Before leaving this topic, let me makea few remarks about the precision of themeasurement. What is being measured isthe power due to wide-band noise. Noise israndom and so there is some uncertainty inmeasuring its power level; the level will fluc-tuate over time. It is a well-known fact thatthe accuracy with which noise power may bemeasured depends on two things: the amountof time over which the power is averaged, ,and the bandwidth of the noise,B, itself; themeasurement error is proportional to

1 B .The fact that the accuracy depends on

the bandwidth may seem a bit strange butif you think of it statistically, the wider thebandwidth, the more samples of the noiseyou are taking, and hence the more accurate

Figure 2 The schematic diagram of the broadband amplifier shows that it uses threeMiniCircuits MAR-3SM monolithic amplifiers.

Figure 1 This block diagram shows the sections of the automatic noise-figure meter.

the average. So, to increase the accuracy, thenoise power must be averaged over longertimes and/or the bandwidth of the measuringinstrument must be increased. CommercialPANFIs may average over periods of a fewseconds and typically have measurementbandwidths of a few MHz.

Older PANFIs usually operated at a fixedfrequency, often 30 MHz. To use them at anyother frequency, it was necessary to convert

the signal to this specific frequency. Manymodern instruments have a much widerfrequency range. The instrument describedhere operates over a frequency range fromabout 3 MHz to over 400 MHz.

Block Diagram

A block diagram of the instrument is shownin Figure 1. There is a wide-band amplifier

with separate inputs and outputs that may beadded to the outputs of the device being testedto increase the signal level. The detector isbased on the Analog Devices AD8307 loga-rithmic detector.3 This wonderful little devicehas a frequency response extending up to500 MHz, and is linear (in logarithmic power)over 8 decades of power. The microprocessorboard controls all aspects of the device opera-tion. In the measurement mode, it turns on the

external noise source by supplying +28 V dc toit and, using its internal A/D converter, mea-sures the signal level detected by the AD8307.It stores this value and then turns off the noisesource and again measures the level detectedby the AD8307. This measurement is repeated50 times over a period of about 1 second. Thenthe noise figure of the device is calculatedand displayed on a 2-line, 16-character liquid


3/8

May/June 2007 41

crystal display (LCD). The A/D converter hasa basic resolution of 1 part in 1024 (10 bits)but each of the 50 measurements is an aver-age of 100 A/D conversions. Because of this,in the absence of noise, the expected overallresolution of each single power determinationwould be about 1 part in 70,000. Ultimately,the final calculated noise figure has an accu-racy that depends mostly on the accuracy ofthe logarithmic response of the AD8307, and

the fact that since the signal being measuredis random noise, it has a limiting accuracy thatdepends on the bandwidth and the averagingtime (in this case, about 1 second).

Detailed Circuit Description

The Broadband Amplifier

The circuit of the broadband amplifier isshown in Figure 2. It uses three MiniCircuitsMAR-3SM monolithic amplifiers.4 The over-all gain is about 37 dB and the circuit, if laidout properly using strip lines of the correctwidth, has a bandwidth from about 3 MHzto close to 2 GHz. Using dead-bug type ofconstruction, the bandwidth will probably bereduced, but since the AD8307s responsestarts to fall off at 500 MHz, this wontmatter. The MAR-3 amplifier is specified tohave a noise figure of about 6 dB in the rangefrom a few MHz to 2 GHz. Each stage drawsabout 35 mA and the total power dissipatedin each stages bias resistance of 200 isabout 14 W. I laid out my amplifier using striplines and surface mount components. The0805 size resistors I used were only rated at0.1 W; therefore I used four 51- resistorsto make up this resistance. Using dead-bugconstruction, you would probably want touse a 200 , W resistor or two 100 , W resistors.

The Detector Circuit

The circuit for the detector is shown inFigure 3. The AD8307 requires a supply volt-age of 5 V but with a current drain of onlya few mA, so a low power 78L05 regulatorwas used. The 51- resistor at the input ofthe AD8307 should be connected betweenpins 1 and 8 with as close to zero lengthleads as you can manage. I used a surfacemount version of the AD8307 and a surfacemount 0805 size resistor located just 1 or2 mm from these leads.

The detector is very sensitive and so thecircuit should be well shielded. I put boththe detector and the wide-band amplifier ina small box made of double-sided circuit-board material and soldered the lid on. SeePhoto B. The detector and amplifier are inseparate compartments in this box. The co-axial inputs to and from the amplifier and tothe detector have the shield braid soldered tothe sides of the box. The dc supply voltage tothe circuit and the detector output are fed intothe box by feed-through capacitors.

VE5FP

Photo B The broadband amplifier is built into a circuit board box mounted inside thebottom of the project case.

Figure 3 An Analog Devices AD8307 logarithmic detector is the heart of the detectorcircuit.


4/8

42 May/June 2007

Photo C This photoshows theassembledSIMM100circuitboard,with theATMega32

Figure 4 An ATMega32 microprocessor on a SIMM100 circuit board provides the brains for the noise-figure meter. This schematicdiagram also shows the power supply, display and control circuitry.

VE5FP

The Microprocessor Module

The microprocessor used was an AtmelAVR-series ATMega32.5 This microprocessorhas eight channels of A/D and 24 other pinsof I/O. I like the AVR-series of microproces-sors because of their low price, their goodperformance and the fact that there are somevery good development tools available forthem. The source code for this project was allwritten for the Gnu C compiler, GCC. There

is a version of this compiler that producescode for the AVR series of microprocessors.A complete package of the compiler and allthe needed utilities is available as a packagecalled WINAVR for installation on Windowscomputers. It is a first-class professional toolset and is completely free! I used a SIMM-100circuit board, which is available from Dontron-ics for about $11 Australian and is a bargain.6Dontronics ships to the USA and the orderingand subsequent shipping is painless. The boardshould be assembled according to the accom-panying directions. See Photo C. The boardwas originally designed for the AT90S8535

microprocessor but it is pin-compatible with

the ATMega32. Use the 16PI version of theATMega32, as it is rated for the industrialtemperature range and for clock frequenciesof up to 16 MHz. It costs just a few pennies

more than the commercial grade. Use a goodquality socket for the microprocessor.

The LCD requires seven I/O pins; thereare two A/D channels used (of eight avail-


5/8

May/June 2007 43

able) and there are two other digital I/O linesused. There is an RS-232 interface circuitbuilt onto this board and it is used to providecommunication to an external computer orterminal, which is needed for calibration ofthe finished unit. I used a 14.7456 MHz crys-tal in the oscillator and, with this clock, theserial RS-232 connection is at 19200 baud.

The Rest of the Circuit

The rest of the circuit is shown inFigure 4. There are two front panel switches.One of these is a spring-loaded toggle switchthat is used to set levels prior to a measure-ment. A push-button switch could alsohave been used here. The other switch is asingle-pole three position switch used to setthe mode of operation. Again I used a toggleswitch. Three-position toggle switches are notcommon and you could as easily use a rotaryswitch instead. There are a large number ofLCDs available. Since they all use the sameprotocol, it doesnt much matter which oneyou choose. I built the instrument in a small

case that had once housed a piece of Tektronixtest gear. Photo D shows the circuit boardsand wiring inside that case. If I were doing itover again, Id put the instrument into a muchlarger box and use an LCD with larger digitsbecause Im getting old and my eyesightisnt what it used to be! The contrast trimmerpotentiometer for the LCD should be adjustedto give the best contrast for the display. Itwill be set so that the voltage at the wiperis close to 0 V. The power supply provides+12 V and +28 V dc. I happened to have anold wall wart on hand, which provided 24 Vac, center-tapped. Any 24 V ac center-tapped

transformer capable of supplying a totalpower of a watt or two is good enough. Thetrimmer potentiometer in the +28 V circuit isused to set the output voltage to exactly 28 V.Do this as accurately as you can.

Calibration and Operating

Calibration of the Power Detector

You need an external terminal or computerto do the calibration of the power detector.The RS-232 connector is designed to connectto a PC and you may use some terminal pro-gram to communicate with the instrument.

The internal instrumental calibration,

which converts from AD8307 output levels toa dBm scale, is stored in the internal EPROMof the microprocessor. The programmed mi-croprocessor has a default calibration based onthe values specified in the AD8307 data sheet.These are fairly accurate for a broad rangeof input frequencies and most users will notwant to bother trying to improve this accuracy.I have made provision for a more accuratecalibration, however, for those who have goodquality signal generators and coaxial attenua-tors and who wish to calibrate the instrumentprecisely at some specific frequency.

Photo D This view shows the top of the project case, with the SIMM100 board, LCDand various wiring of the noise-figure meter.

Photo E Thisphoto shows myhomebrewed noisesources.

VE5FP

VE5FP

If you do have these, then the calibrationprocedure is as follows:

1. Connect a 50 signal generatoroperating at the desired frequency to theSIGNAL INPUT connector on the instru-ment. Set the output level to a fairly

high level around 0 dBm.2. Type the character c (upper or lower

case) on the terminal and the microprocessorwill respond with a line asking for the inputlevel. Type the level being produced by yoursignal generator in dBm. Then, press the SET


6/8

44 May/June 2007

switch on the instrument.3. The instrument will respond by asking

for the next data point.4. Set the generator to a new level about

10 or 20 dB lower. The preferred way to dothis is just to leave the generator setting as itwas before but to add a known attenuationinto the line. Again, type the new signal leveland press the SET switch.

5. Repeat the step above three more timesfor a total of five settings of the signal genera-tor. Choose the five signal levels so that theyare distributed fairly evenly (in dB) between70 dBm and 0 dBm.

The order in which you set the levelsdoesnt matter. Internally, the five valuesare used to calculate the least-square bestfit of a straight line to the data and the twoparameters describing this line are storedin the instrument in EEPROM. These twoparameters are subsequently used to translatefrom AD8307 output to dBm.

While connected to the terminal, you mustalso now tell the instrument the noise level ofyour noise source. Type a single character e(again, upper or lower case) and the instrumentwill respond by asking you to type the ENR,in dB, of your noise source. This value is alsostored internally in EEPROM so it is therewhenever the instrument is turned on again.

While connected to the terminal, you canalso change whether the noise figures aredisplayed as temperatures or in dB. Type ab to select dB or a t to select temperature.The mode is also saved in EEPROM so it willremain in whatever state you left it the nexttime you turn it on. Finally, you may displayall the internal settings of the instrument bytyping the character d.

I have not discussed noise sources heresince they are a subject requiring a separatearticle. Paul Wade has written an excellentarticle about noise sources that tells you howyou may make your own.7 My own noisesource is a homebrew design based on infor-mation from that article. See Photo E.

Operation

Before making any measurements, it isadvisable to check to make sure that thedetector and amplifier are shielded wellenough. With the ON-AUTO-OFF switch inthe ON position, make a note of the indicatedsignal level when you have a termination onthe SIGNAL INPUT connector. Then switch itto the OFF position and again note the signallevel. It should be the same to within 0.1 dBor so. If it isnt, you havent shielded thedetector or amplifier well enough and yourmeasurements will not be accurate.

Assuming all is well, you are ready tomake a measurement. First, connect thenoise source to the instrument as shown inFigure 5A and press the SET switch. Theband-pass filter shown should be appropriateto the device being measured. For example,

Figure 5 This drawing shows the steps involved in making a noise-figuremeasurement. Part A shows the connections to measure the noise source. Part B showsthe device under test connected between the noise source and the noise-figure meter.Part C shows that sometimes an additional amplifier will be needed if the device undertest does not produce a signal with high enough power output.

if the device is a down-converter with anoutput at 144 MHz or a 144 MHz amplifier,the filter should be tuned to 144 MHz. Afterpressing the SET switch, the display will showthe measured levels with the noise source onand off and also will show the excess noisetemperature of the broadband amplifier in theinstrument. This should be something in theneighborhood of 900 K. MiniCircuits speci-fies the MAR-3SM as having a noise figure of6 dB, which corresponds to about that value.Then, connect the device-under-test (DUT) tothe instrument as shown in Figure 5B. With theON-AUTO-OFF switch in theOFF position, note

the signal level measured by the instrument. Itshould be greater than 70 dBm and no morethan about 0 dBm. If it is somewhere in thisrange, you are ready to go. Just put the ON-

AUTO-OFF switch into theAUTO position andit will give you a measure of the excess noisetemperature of the DUT. You may convertthis to dB, if you wish, using Equation 2. Theinstrument will also show a value for the gainof the DUT. The noise temperature displayedhas been corrected for the gain of the DUT andthe noise contribution of the following stage.

If the signal level is lower than 70 dBmwith the configuration shown in Figure 5B, you


7/8

May/June 2007 45

will need to use an additional external amplifierto get the levels up to where the measurementcan be made accurately. This amplifier shouldbe inserted at the output of the DUT and beforethe input to the broadband amplifier. Connectthe circuit as shown in Figure 5C and now lookat the level when the ON-AUTO-OFF switch is inthe OFF position. If it is still too low, you willneed to add another external amplifier. If thesignal level is too high, dont use the broadband

amplifier and/or just add some attenuationbefore the SIGNAL INPUT connector.If the DUT is a down-converter, the gain

shown will be correct if the noise source ENRis the same for both the output frequency of theDUT and the input frequency. That is becausein the SET measurement, the noise level wasmeasured at the output frequency of the DUTwhereas in the final noise measurement, thenoise source is connected to the input of theDUT, which sees it at its input frequency.

If you want to measure the noise tempera-ture of a LNA ahead of a down-converter,you must first SET the instrument with the

down-converter in the circuit, as shown inFigure 6A. Then connect the LNA in frontof the down-converter as shown in Figure 6Band set the switch to theAUTOposition to readthe noise temperature and gain of the LNA. Ifthere is too much overall gain, you may wantto by-pass the broadband amplifier and con-nect the output of the down-converter directlyto the SIGNAL INPUT of the instrument.

Other Considerations

You will need to have a range of band-pass filters for the frequencies that younormally use. I made up some fixed double-

tuned LC filters for 144 and 30 MHz and asmall fixed cavity filter for 432 MHz. I alsomade a tunable cavity filter by modifyinga surplus HP5253B frequency converter.Photo F shows the front panel of this filter.These units plugged into some older HPfrequency counters and they have a verywell-made tunable silver-plated cavity. Theymake very nice 50 MHz to 500 MHz tunable

band-pass filters. One often sees these unitsfor sale at hamfest flea markets for just a fewdollars and they are worth picking up.

One disadvantage of the commercialPANFIs is that it is very difficult to tunesomething where the output is given bychanging figures in a text display. Humansare better at evaluating an output if there issome analog value associated with it, suchas a voltage that can be displayed by a meter.

I added an analog output to the system witha very simple circuit addition. A low-passfilter consisting of a 5.6 k resistor going toa 0.1 F capacitor to ground is connected topin 3, J5 on the SIMM100 board. This pincorresponds to PD5 of the microprocessor. Iadded software to pulse-width modulate thisoutput to produce an analog voltage that isproportional to the measured value of Y. Thevoltage varies from 0 to 5 V as Y changes fromN.00 to N.9999 volts, where N is an integer.For example, if Y is 2.4, the voltage will be0.4 5 = 2.0 V. If Y were to vary from 2.95to 3.05, the voltage would start at 0.95 5 =

4.75 V, increase up to 5 V, jump down to zeroand then go up to 0.05 5 = 0.25 V. Just usingthe non-integer portion of Y is an easy way tomake an expanded-scale voltmeter. To tune asystem for lowest overall noise temperature,you just tune for maximum Y using a voltme-ter on this output. I added a pin jack onto the

front panel on my instrument for this output.

Conclusion

This instrument will make accurate noise-figure measurements if you take just a littlecare in the construction and operation. Thecircuit is very simple and easily reproduced.Analog Devices made it all possible with theAD8307 logarithmic detector. That, and theuse of a microprocessor to massage the data,

results in a first-class laboratory instrument.Both the Agilent Application Note andthe article by Paul Wade discuss someconsiderations to be observed when makingnoise-figure measurements and I recommendthat you read them if youre serious aboutinterpreting the results you get.

The source code and the HEX objectcode for the microprocessor are available onthe ARRL QEXWeb site.8 The author willprogram ATMega32 microprocessors for thecost of return postage.9

Appendix: Measuring Noise Temperature

Consider the basic measuring system,which consists of an amplifier and a detec-tor to measure output power as shown inFigure A1. The system is designed to mea-sure the ratio of the two output powers whenthe external noise source is turned on, tothe output power when the external noisesource is turned off. This ratio is called the

Figure 6 This drawing shows the connections required if a down-converter is neededto bring the output frequency of the device under test into the range of the noise-figuremeter. Part A shows that the noise source and down-converter are measured first, andthen the device under test is added, as shown at Part B.

Photo F I built a tunable cavity filter intoan HP frequency counter box.

VE5FP


8/8

46 May/June 2007

Y-factor. The gain of the system amplifier isGs and the system noise temperature is Ts.The latter, the excess noise of the measuringsystem, can be represented by a resistance attemperatureTs at the input whose noise poweris summed with the external input. When theexternal noise source is turned on, it acts asa resistance at temperature Th and when itis turned off, it acts as a resistance at roomtemperature, Tc.

When the external noise source is turnedon, the power output at the detector will be:

Poh = (Th + Ts)Gs [Eq A1]

When the noise source is turned off, thepower output at the detector will be:

Poc = (Tc + Ts)Gs [Eq A2]

Tc is commonly taken to be 290 K.The ratio of these two powers, Poh /Poc, is

the Y-factor for the system; Ys.This factor is thus:

Y =

P

P =

T +T

T +Tsoh

oc

h s

c s [Eq A3]

We can rearrange the terms of EquationA3 to give the noise temperature of the sys-tem in terms of the Y-factor and the externalnoise temperature:

T =T Y T

Ys

h s c

s

1

[Eq A4]

Knowing Ts from Equation A4, we canalso determine the measuring system gainby substituting it into Equation A1:

G =P

T +Ts

oh

h s

[Eq A5]

Gs will be in units of W / K.To measure the noise temperature of an

amplifier or a converter (or, the noise figure,which may be derived from the measured noisetemperature), the noise temperature and gainof the measuring system are first measured byconnecting the external noise source as shownin Figure A1. This provides the initial calibra-tion of the measuring system. The amplifier orconverter whose noise figure is to be measuredis then inserted between the measuring systemand the external noise source as shown inFigure A2. The Y-factor for the whole ampli-fier/converter measuring system combinationis then measured. From this Y-factor, wecan determine both the overall system noisetemperature as well as the amplifier/converternoise temperature alone. Let the overallY-fac-tor be Y. Then, the overall noise temperatureof the combined system, T, is given by:

T=T Y T

Y

h c

1 [Eq A6]

The true noise temperature of the ampli-fier/converter alone is given by:

T =T T

G

s

[Eq A7]

Therefore, to determine the true noisetemperature of the device, we must know thedevice gain. This can be found by an equa-tion that uses the overall power output when

the external noise source is turned on, Poh,

and when it is turned off, Poc, as well as thesystem gain, Gs:

G =P P

G T T

oh oc

s h c

( )

[Eq A8]

Gs has been measured previously.

In summary, in order to determine the truenoise figure of an amplifier or a converter,you must first measure the measuring systemnoise temperature and gain using the set-upshown in Figure A1, where the external noisesource is connected to the input of the mea-suring system. This is the calibration phase ofthe measurement. Then, you disconnect theexternal noise source from the input and con-nect the output of the amplifier or converterto the input of the measuring system and putthe external noise source onto the input of theamplifier or converter. Then, the instrument

will show the true noise temperature of theamplifier or converter as well as its gain.

Notes1Harke Smits, PAHRK, A Noise/Gain Analyz-

er, QEX, Nov/Dec 1999, pp 5-10. See alsoreferences therein.

2The Agilent Technologies Web site is atwww.home.agilent.com, and the direct linkto the Application Note is http://cp.literature.agilent.com/litweb/pdf/5952-8255E.pdf

3www.analog.com/en/prod/0,,759_847_AD8307,00.html

4www.minicircuits.com5www.atmel.com/dyn/products/product_

card.asp?part_id=2014

6www.dontronics.com/simm100.html7Paul Wade, N1BWT (now W1GHZ), Noise Mea-

surement and Generation, QEX, November,1996, pp 3-12. Also, see references therein.

8The ATMega32 microprocessor source codeand the HEX object code are available on theQEXWeb site. Go to www.arrl.org/qexfilesand look for the file 5x07_Koehler.zip .

9I can only handle the DIP version of the mi-croprocessor. Send it in a padded envelope,along with an addressed return padded en-velope, to J. Koehler, 2258 June Rd, Courte-nay, BC, V9J-1X9, Canada. I cannot use USstamps so please enclose enough IRCs or$2 US to cover the postage.

Figure A1 A Precision Automatic Noise Figure Indicator (PANFI) switches the noisesource on and off, and measures the noise power.

Figure A2 With the device under test placed between the noise source and PANFI, thenoise figure is measured.

Noise measure

Documents

Transcript of Noise measure