Agilent AN 1287-6

Using a Network Analyzer to Characterize High-PowerComponentsApplication Note

DUT

3345

679

1112141415151617171718181919192323 252627

IntroductionDefining high powerWhy high-power measurements can be challengingNetwork analyzer configurations for measuring high-power devicesConfiguration 1Configuration 2Configuration 3Configuration 4Configuration 5Configuration 6Additional configurationsSource levelingSource leveling using power-meter calibrationSource leveling using external levelingCalibration — purpose and typesCalibrating tips for best resultsDynamic accuracyChoosing calibration power levelsCalibrating at one power level versus two power levelsCommon problems of high-power measurementsAmplifiers with AGC loopsOn-wafer devices (pulsed measurements)AppendixNetwork analyzers—definition and capabilitiesTest setsCalibrationSuggested reading

Table of Contents

3

IntroductionThis application note describes linear and nonlinearmeasurements of high-power components and howto use a network analyzer for making them. It coversthe power limitations of a network analyzer, andspecial network-analyzer equipment configurationsfor high-power measurements. How to improve the accuracy of high-power measurements andsolve common problems when making high-powermeasurements are also described.

To get the most from this note, you should have abasic understanding of network analyzers and themeasurements you can make with a network ana-lyzer. For a basic review, please see the Appendixat the end of this note. Additional network analysisliterature and study materials can be orderedthrough Agilent Technologies. A reference list isincluded at the end of the note.

Defining high powerWhat might be considered “high-power device output” (e.g., 30 dBm or 1 Watt) in one applicationcan be insignificant in another application, such asa radar test that uses devices with power levels inthe 60 dBm (1,000 Watt) range. In this note, “highpower” refers to a power level above the compres-sion level and certainly above the damage level of a standard network analyzer. Therefore, a poweramplifier with an output beyond the measurementcapability of a standard network analyzer would beclassified as a high-power device. We extend ourdefinition to also include devices that require adrive level that is higher than a standard networkanalyzer can provide. So a high-power device is one that delivers more power than a standard network analyzer can measure, or requires moreinput power than the analyzer can provide.

4

Why high-power measurements can be challengingTwo main challenges exist when measuring high-power devices:

1. The measurements performed on high-powerdevices can be different than those required tocharacterize lower-power devices. Measurements of high-power devices also can be performed dif-ferently than those made at lower power levels.

Pulsed measurements are a good example.Measurements typically are not pulsed at lowerpower levels since device overheating tends not tobe a problem. High power can heat up a device,affecting its measured characteristics. Many on-wafer measurements, for example, require pulsedRF and pulsed DC bias, which reduces the averagepower dissipation and keeps the temperature ofthe device constant.

2. High-power measurements require special network-analyzer configurations. This can meanadding attenuation or a coupler between the out-put of the device under test (DUT) and the input of the test instrument to protect the receiver. Itcan also mean adding amplification to the stimulus signal if more power is required. Calibration andaccurate measurements become significantly morecomplex as additional equipment is added to thetest setup. In some configurations the additionalhardware can make some types of calibrationimpossible, or limit the number of measurableparameters. For example, reverse S-parameterscannot be measured in some configurations. Theinability to perform certain calibrations can limitthe accuracy of the measurements.

This application note will show configurationsranging from those that are easy to assemble butmay have limited accuracy or measurement capa-bility, to more complex configurations that arevery accurate and can make the same measure-ments as a standard network analyzer.

5

Network analyzer configurations for measuringhigh-power devicesThe high-power test configurations described inthis note are designed to boost the power comingfrom the network analyzer’s source as necessary,and also to protect components such as thereceivers, couplers, and switches inside the network analyzer from excessive power levels.

To select the correct network analyzer configura-tion, you will need to consider the DUT and therequired measurements and accuracy. This sectionshows high-power test configurations including thenecessary hardware, how to set up and calibratethese configurations, and the advantages and limitations of each.

The configurations are ordered by the degree ofmeasurement capability. In general, the increasedcapability results in increased complexity in theconfiguration. The first configuration is simple — it uses a standard network analyzer and doesn’trequire high drive power, which means the networkanalyzer’s source does not need to be boosted. Theother configurations measure devices that requirehigh-power inputs and also have high-power out-puts. Measuring these devices requires amplifyingthe network analyzer’s source signal somewherealong the RF path before it reaches the DUT. Thehigh output power also requires protection for thecouplers, receivers, and switches inside the analyzer.

Available high-power measurements Available calibrations

Forward Forward transmission

Boosted transmission and reflection Forward/ Full Configuration Complexity source only only reverse two-port Response

1 low X X X2 low X X X 3 medium X X X 4 medium X X X 5 high X X X X6 high X X X

6

Configuration 1Configuration summary• Low complexity• No boosted source• Forward and reverse measurements• Full two-port or response calibration

Setup and featuresThe simplest high-power devices to measure arethose that have high gain but don’t require highlevels of drive power, and typically are tested intheir linear region. Since these devices have highgain, the receiver must be protected from the highoutput power. An attenuator or coupler addedbetween the output of the DUT and the analyzer’stest port protects the receiver. If using a coupler,terminate the through arm of the coupler with acharacteristic impedance load. The coupled arm ofthe coupler sends a small portion of the input sig-nal to port 2. For a 20-dB coupler, the signal at thecoupling arm is 20 dB less than the strength of thesignal at the input. Determine the maximum powerout of the DUT, subtract the power level requiredat the test port, and then choose the appropriateattenuator or coupler value. Choose componentsthat are specified to handle your chosen power level.

Configuration 1 makes both forward and reverse,reflection and transmission measurements (S11,S12, S21, and S22) if using an analyzer with an S-parameter test set.

CalibrationThis particular setup allows both forward andreverse measurements so full two-port error-correction, the most accurate calibration, can beperformed. Calibration is performed at test-port 1(or at the end of the cable attached to the port)and at test-port 2 (with the attenuator and any testcable). Include the attenuator or coupler on port 2when performing the calibration to remove anymismatch between the attenuator and the analyzer’stest port. Since calibration is performed with allhardware in place and all error terms corrected,the measurements can be as accurate as the standard analyzer.

Configuration 1

The attenuator on port 2 degrades the uncorrecteddirectivity of port 2 by twice the attenuation value.This limits the stability of the calibration and can make S22 measurements very noisy. If an S22

measurement is needed, calibrate at a higherpower level, lower the power level when makingforward measurements, and then raise the powerlevel for the reverse measurements. There is more information on calibrating and power levels at the end of this note.

A common assumption is that the DUT’s displayedgain should be increased by the amount of theattenuation, but remember that the calibrationincludes the attenuator so the values as displayedare correct.

Network analyzer

DUT Attenuator

Calibration points

1 2

7

Configuration 2Configuration summary• Low complexity• Boosted source• Forward transmission measurement only• Response calibration only

Setup and featuresThe simplest way to increase the power of thestimulus signal is to add a booster amplifier at thetest port of a standard network analyzer. The inputof the booster amplifier connects to the analyzer’ssource port and the output connects to the DUT(see Configuration 2). This configuration booststhe signal level going into the DUT without modify-ing the analyzer’s built-in test set or having to addadditional couplers. The analyzer’s source powerplus the gain of the booster amplifier gives thepower available for testing the DUT. Protect thereceiver by adding an attenuator or couplerbetween the output of the DUT and port 2 of the analyzer, as shown in Configuration 1.

Configuration 2

Configuration 2 is convenient because it consistssimply of a standard network analyzer. However, it has numerous limitations. When boosting the signal at port 1, only high-power transmissionmeasurements in the forward direction (S21) andnonboosted reflection measurements in the reversedirection (S22) are possible. The position of thebooster amplifier, with its high reverse isolation,makes forward reflection (S11) and reverse transmis-sion (S12) measurements impossible. Nonboostedreverse reflection measurements can be made, butwith limitations as described in Configuration 1.

The accuracy of Configuration 2 is limited becausethe network analyzer measures the reference signalbefore the booster amplifier (coupling takes placeinside the analyzer). As a result, any mismatchbetween the booster amplifier and the DUT is notratioed out, and therefore ripple will appear in themeasurement (Figure 1). This ripple can be signifi-cant due to the poor source match of the boosteramplifier. Even the uncorrected source match ofthe network analyzer is likely to be much betterthan the source match of the booster amplifier.Network analyzer

DUTAttenuator

Calibration points

DUTBoosteramplifier

8

The uncertainty associated with the measurementwill depend upon the quality of the match betweenthe booster amplifier and the DUT. In addition, anydrift or power fluctuations associated with thebooster amplifier will appear to be from the DUTsince the reference signal is not measured after thebooster amplifier. This configuration is best foranalyzers that do not allow direct access to the R(reference) channel and when the accuracy of themeasurements is not important. If the analyzerallows access to the R channel and the referencesignal is coupled after the amplifier, a configura-tion that allows more accuracy is possible (seeConfiguration 3).

Figure 1. Ripple is caused by a mismatch between theDUT and the booster amplifier.

CalibrationOnly transmission response calibration is possiblewith this setup. The location of the booster amplifier does not allow the analyzer to maketransmission measurements in the reverse direction,making a full two-port calibration impossible.Perform a response calibration by connecting theoutput of the booster amplifier to the attenuator or coupler on port 2. A basic response calibrationdoes not remove mismatch errors due to the DUT— only the frequency response errors and any mismatch associated with the booster amplifier,the attenuator, and the analyzer. Since mismatch is not corrected for during calibration, even measurements made with a response calibrationare limited in their accuracy.

In this setup you can improve the source match byadding an isolator between the booster amplifierand the DUT, or by adding attenuation. Adding anisolator between the output of the booster amplifierand the input of the DUT will remove the effects ofany mismatch. If the booster amplifier has enoughgain, a 3-dB or 6-dB attenuator at its output can be added to improve its output match. Be sure to include the isolator or any attenuation in the calibration if you include it in the measurementsetup. Also make sure that the isolator or attenuatorcan handle the power level being tested. Instead ofusing a high-power isolator, a circulator with ahigh-power load on its third port can be used.

CH1 S21 &M log MAG 2 dB/ REF 11 dB

CENTER 2.500 000 000 GHz SPAN 4.000 000 000 GHz

Cor

PRm

LINE TYPE

1

18 Dec 1997 16:16:33

1

1_: 13.704 dB

2.512 960 000 GHz

True response

True responsewith

mismatch ripple

9

Configuration 3Configuration summary• Medium complexity• Boosted source• Forward transmission measurement only• Response calibration only

Setup and featuresCompared to Configuration 2, Configuration 3improves measurement uncertainty and flexibilityby measuring the incident signal after the boosteramplifier rather than before. Measuring the inci-dent signal after the booster amplifier ratios outthe effects of any source mismatch.

Boost the analyzer’s source power by connectingthe booster amplifier to the “RF out” port on theanalyzer. An external coupler connected to the output of the booster amplifier directs a portion ofthe boosted source signal to the R channel of theanalyzer. This signal becomes the reference signalused in ratioing (Configuration 3).

Configuration 3

Measuring the reference signal after the amplifierratios out any mismatch ripple between the boosteramplifier and the DUT and any drift or power fluctuations of the booster amplifier. The reverseisolation of the booster amplifier makes forwardreflection measurements and reverse transmissionmeasurements impossible.

In configurations where the user provides an R channel signal, determining the proper level isimportant. Too low a signal level can be affected bynoise and can cause phase-locking problems andinaccurate measurement results. Too high a signallevel can be affected by receiver compression, causing inaccurate results. The receiver can also be damaged if the signal is high enough. To findthe proper signal level for the R channel (a typicalrange might be 0 to –30 dBm) look at the analyzer’sspecifications. Then determine the signal level atthe coupled arm of the coupler, and add attenua-tion to achieve the proper signal level at the R channel.

DUTDUT

Calibration points

Attenuator

R in

Boosteramplifier

Coupler

Network analyzer

1 2

Attenuator

10

Normally, the analyzer attempts to phase-lock near the start frequency of the sweep. When theanalyzer cannot phase-lock at the start frequencythough, the analyzer performs a “pre-tune cali-bration” routine. During pre-tune calibration theanalyzer attempts to phase-lock at a set frequency (100 MHz for the Agilent 8753D family), whichcould be below the start of the analyzer’s sweep.The pre-tune calibration routine would be per-formed, for example, if the analyzer is switched toexternal reference mode without a signal attachedto the R channel. When the external R channel signal is connected the analyzer is attempting tophase-lock at the pretune frequency (100 MHz).With a band-limited device in the R channel path (e.g., a booster amplifier) there may not besufficient signal at this pretune frequency andphase-locking errors can occur on some analyzers.

On the 8753E network analyzer, turning “PLLAuto” off (located in the Service Modes menu) willstop the analyzer from performing the “pre-tunecalibration” routine. Turning “PLL Auto” off rarelyaffects the accuracy of instrument measurements.Another way to prevent this phase-locking error is to use frequency-offset mode in the 8753E. Infrequency-offset mode the analyzer assumes that a band-limited device is being measured, and as aresult the analyzer never goes outside the sweeprange to phase-lock. Setting LO = 0 Hz when in frequency-offset mode allows the analyzer to beused normally.

Band-limited devices in the R channel path maycause phase-locking problems in the Agilent 8720Efamily of network analyzers unless the analyzerhas the high-power option (Option 085) or frequency-offset option (Option 089). With either of theseoptions, phase-locking is not a problem. This isbecause the internal reference switch switches tothe internal signal when phase-locking. After theanalyzer is phase-locked, the reference switchreturns to the external signal and the analyzer is ready to make measurements.

CalibrationCalibrating this configuration is similar to the calibration performed in Configuration 2. Only aresponse calibration is possible since only forwardmeasurements can be made. Perform the responsecalibration by connecting the output of the coupler’sthrough arm to the attenuator on test-port 2 of the analyzer.

11

Configuration 4Configuration summary• Medium complexity• Boosted source• Forward transmission and reflection

measurements only• Response calibration only

Setup and featuresHigh-power reflection and transmission measure-ments in the forward direction can be made onanalyzers that provide access to the A channel andthe R channel. Forward reflection measurementsare possible by adding a reverse coupler betweenthe coupler used for the R channel and the DUT.This reverse coupler, which is connected to the Achannel, allows the forward reflection measurement.

CalibrationAgain only a response calibration is possible sincea reverse measurement cannot be made. When per-forming the calibration, connect the output of thesecond coupler to the attenuator or output couplerso that the calibration includes all hardware.

Configuration 4

8753E with Option 011

DUT

Reversecoupler

Forwardcoupler

Calibration points

Booster amplifier

R A BRF out Attenuator

AttenuatorAttenuator

12

Configuration 5Configuration summary• High complexity• Boosted source• Forward and reverse measurements • Full two-port or response calibration

Setup and featuresThe configurations presented up to now have madeuse of an analyzer’s build-in test set, or includeadditional external hardware to make high-powermeasurements. Another approach is to modify the standard test set to make it more suitable forhigh-power measurements. Typically modificationsto the internal test set are options that Agilent provides at time of purchase.

Configuration 5

High-power optionAn example of a modified test set is the high-poweroption, Option 085, for the Agilent 8720E family of vector network analyzers. This option providesfour features that allow for the measurement ofhigher power levels. These are: 1) access to the RFpath between the source and the transfer switch;2) direct access to the R channel; 3) direct accessto the RF path between the transfer switch and thetest ports; and 4) step attenuators between thecouplers and the samplers on the A and B channels.The 8720E with Option 085 allows input power andoutput power up to +43 dBm at the test ports ofthe analyzer. For more information on the set-up of 8720E, Option 085, refer to the 8720E manualsection Making High Power Measurements withOption 085.

DUTDUT

IsolatorIsolator

Couplerswitchswitch

Coupler 1 2

R in

RF inRF out

Calibration points

Attenuator8720E with Option 085

Coupler

Booster amplifier

13

Forward and reverse measurements by amplifying thesource signalAccess to the RF path between the source and the transfer switch allows the source signal to beamplified and then be switched between port 1 orport 2, allowing forward and reverse high-powermeasurements. The input of the booster amplifierconnects to the “RF out” connector on the analyzer,and the output of the booster amplifier connects to the coupler. The coupled arm of the coupler connects to the R-channel input to provide the reference signal used for ratioing. To insure anoptimum power level at the R receiver, add anyneeded attenuation between the reference couplerand the analyzer. The optimum power range for thereceiver is provided by Agilent. The through arm of the coupler connects to the “RF in” connector on the analyzer. The amplified signal goes throughthe transfer switch, which directs it to either testport. Jumpers between the transfer switch and the test ports give access to the RF signal path.They allow the user to add high-power isolators to protect the transfer switch. Without isolators,signals with too much power can damage the transfer switch. Isolators on both sides of thetransfer switch ensure that after the signal hasbeen measured by the coupler any signal will beterminated, thereby protecting the switch.

Protecting the receiversThe final feature of this high-power option is theinternally controlled step attenuators that protectthe receivers. Located after the couplers andbefore the receivers, these 55-dB step attenuators(with 5-dB increments) reduce the signal to anoptimum level for the receiver. These attenuatorsare controlled from the front panel of the analyzer.

CalibrationA major benefit of this configuration, besides theability to do both forward and reverse measure-ments, is the ability to do full two-port vector errorcorrection. Calibrating in this manner takes intoaccount the effects of the hardware that has beenadded to the setup (isolators, amplifiers, couplers,etc.) and all errors associated with the analyzerand the measurement setup up to the point of calibration. Perform the calibration at the pointwhere the DUT will be connected.

When using this configuration, it is extremelyimportant not to damage any of the internal components because high power levels are insidethe analyzer itself. Since amplification takes placebefore the transfer switch, the power-handlingcapabilities of the switch must be known. Analyzersthat allow signal amplification before the transferswitch must specify how much power the switchcan handle. Agilent specifies maximum power forthe transfer switch for two conditions, when theswitch is switching, and when the switch is notswitching. Typically the nonswitching power-ratingis higher than the switching power rating. Thepower-handling capability of all components in the RF path must be considered when makinghigh-power measurements. When using this configuration, it is important to understand how the analyzer works and the power-handlingcapability of each component in the signal path.

14

Configuration 6Configuration summary• High complexity• Boosted source• Forward transmission and reflection

measurements only • Response calibration only

Setup and featuresConfiguration 6 uses the same high-power test setas Configuration 5, but the hardware is rearrangedso that measurements can be made at even higherpower levels. Configuration 5 is desirable becauseit allows high-power measurements in both direc-tions, but the maximum power level does not reachthe test ports. Due to losses in the test set and thepower limitations of the transfer switch, the powerat the test ports is less than the couplers can handle.To test at power levels up to the level that the couplers can safely handle (+50 dBm), the Agilent 8720E with Option 085 is used in an alternate setup.

Amplification is done after the transfer switch and before the test port to get the maximum power possible at the test ports. This configurationallows higher power levels than possible usingConfiguration 5, but high-power measurements,both transmission and reflection, are possible only in the forward direction.

Configuration 6

CalibrationTwo-port vector error calibration cannot be usedin this configuration. Since the external R channelis being used and is connected only to port 1, the reference signal is not accurate when doing a reverse sweep. This means only a response cali-bration can be done between test-port 1 andtest-port 2. Since full two-port calibration cannotbe done, this configuration should only be used ifmore power is needed than can be provided byConfiguration 5.

Additional configurationsOften Agilent can design test sets to specificallymatch your needs. Contact your local sales representative if you would like to investigate aspecially configured test set. For example, solid-state switches may be substituted for mechanicalswitches in a high-power test set. High-power testsets usually use mechanical switches to handlehigher power levels. If continuous switching isrequired, solid-state switches need to be used.Special configurations can include solid-stateswitches in high-power test sets if needed.

Other special configurations include high-powertest sets that allow through-reflect-line (TRL) calibrations for noncoaxial measurements. Special configurations for applications thatincrease the power range over which a networkanalyzer sweeps such as compression measure-ments, are also available.

DUT

8720E with Option 085

RF out RF in

Attenuator

SwitchCouplerSwitch

Coupler

R in

DUT

1 2

Calibration pointsCoupler

Booster amplifier

15

Source levelingWhen amplifiers operate in their nonlinear region,the measured response may differ from the trueresponse of the amplifier. If the amplifier is operatingin its nonlinear region, variations in the stimulussignal might not be duplicated at the output of theamplifier. In contrast, the response of an amplifiertested in its linear range is not affected by thislevel variation. By ratioing the response to thestimulus, the network analyzer removes the effectsof the stimulus variation and displays the true performance of the amplifier.

However, if the amplifier is operating in its nonlinearrange, at or near saturation, the amplifier does notproduce an output signal with a variation propor-tional to the variation of the signal present at theinput. The ratioing process in the network analyzerthen creates an erroneous display (Figure 2).

To optimize measurements in the nonlinear region, aleveling scheme must be used to produce a stimulussignal with a frequency response that is as flat aspossible.

Figure 2. When a device operates linearly, its trueresponse can be measured. However, a device that oper-ates nonlinearly is affected by nonlevel input signals.

Source leveling using power-meter calibrationUsing power-meter calibration in continuous samplingmode is one way to level the source (Figure 3). Thepower level at each frequency point in the sweep ismeasured with a power meter. The network analyzer,connected to the power meter via GPIB, adjusts itssource power until the power meter measures thedesired power level. Then the measurement ismade. Since the accuracy of the power meter isvery high (uncertainty in the tenths of a dB range),you can have confidence that the power level isaccurate as well as flat.

Figure 3. Power-meter calibration levels the source signal and removes the nonlinear ripple.

8753E with Option 011

DUT

Power meter

GPIB

R A BRF out

Power sensorPower splitter

16

Source leveling using external levelingUsing power-meter calibration for source levelingworks well, but can slow down the measurementstoo much for some applications. A faster method ofsource leveling is external leveling. External levelingmakes use of an external amplitude modulation(AM) input, available on some network analyzers.Adjusting the voltage at the AM input adjusts theanalyzer’s source power. By creating a leveling circuit that is external to the network analyzer,and adjusting the voltage to the external AM input,source power can be leveled on a real-time basis.

The external leveling circuit can be implemented in several ways. One common implementation is tocreate a circuit consisting of a detector diode andan operational amplifier. Changes in the powermeasured by the diode are detected by the opera-tional amplifier circuit. The circuit connected tothe AM input of the analyzer adjusts the sourcepower by adjusting the voltage at the AM input(Figure 4).

Figure 4. This circuit shows a common way to implementexternal leveling.

When using either power-meter calibration orexternal leveling, keep the source-leveling processactive, just as it will be during the measurement.Then perform a response calibration as you wouldif source leveling was not used.

8753E with Option 011

Attenuator

DUT

R A BRF out

Power splitter

Ext AM+–

+

–REF

50 Ω

Loop gain

RF filter and load

Detector

17

Calibration—purpose and typesCalibrating an analyzer eliminates systematicerrors and makes measurements more accurate.This note has shown calibrations for various con-figurations and where in the test setup to performthese calibrations. Of the two types of calibrationdiscussed, full two-port calibration is the mostaccurate because it removes all systematic errorsin the measurement setup, ideally up to where theDUT connects to the analyzer. Response calibrationis less accurate and removes only the frequency-tracking errors. Frequency-tracking errors areassociated with the differences in frequencyresponse between one channel and another. In afull two-port calibration, the analyzer sweeps inboth forward and reverse directions, while in aresponse calibration, the analyzer sweeps only in the forward direction.

Calibration tips for best resultsIn addition to understanding what type of calibra-tion is possible for each configuration and whereto perform the calibration, it is important tounderstand how to get the most out of a calibra-tion. The following topics discuss techniques andconsiderations for achieving the best calibrationresults.

Dynamic accuracyBy examining the dynamic accuracy response of an analyzer, you can optimize the power levels forcalibration and measurement. Dynamic accuracyrefers to the uncertainty associated with calibratingat one signal level and measuring at another. The dynamic accuracy plot for the Agilent 8753Eshows extremely low uncertainty (0.02 to 0.06 dB)if receiver power is between –10 and –50 dBmindependent of the calibration power level (Figure 5).Perform calibration and measurements so that thesignal level at the receiver is in this high-accuracyrange. At lower signal levels, noise is a factor; athigher levels, receiver compression is a factor.

Figure 5. Agilent 8753E dynamic accuracy

–10 dBm

–20 dBm

–30 dBm

Reference power level

Acc

urac

y (d

B) 1

0.1

0.0110 0 –10 –20 –30 –40 –50 –60 –70 –80 –90 –100

Test port power (dBm)

18

Choosing calibration power levelsIf you are measuring an amplifier and the outputpower level is higher than the input power level, at what power level should you calibrate: at thehigher power level, the lower power level, some-where in between the two levels, or at a differentlevel for each port? Calibrating and measuring with the test port powers in the range of lowestuncertainty, –10 to –50 dBm for the Agilent 8753E,results in accurate measurements as shown above.Calibrating at the input (lower) power level canresult in a power level at the output receiver thatis too low, because attenuation is required to pro-tect the analyzer during measurement of the DUT.Calibrating at the output (higher) power levelmight not be possible since the gain of the DUT can result in an output power level that the net-work analyzer’s source together with the boosteramplifier cannot generate. Therefore, you mightneed to perform some calibrations at a power level between the input power level and the output power level. In general, to reduce noise calibrate at the highest possible power below the onset of receiver compression.

Calibrating at one power level versus two power levelsA response calibration is done at only one powerlevel since only one forward sweep is made. Fulltwo-port calibration, on the other hand, can bedone at different power levels since both forwardand reverse sweeps are made. The analyzer cancalibrate one port using one power level and theother port using another power level to more accurately match the power levels present duringthe measurement of a DUT.

As an example, the open, short, and load standardsfor port one and the reverse through measurementsmay be made at the lower power level. The open,short, and load standards for port two and the forward through measurements could then bemade at the higher power level. It is also possibleto perform a full two-port calibration at one powerlevel as in a response calibration. The power levelused would be determined from the dynamic accuracy specification of the analyzer.

A final note on calibration: power-handling capabilities of the calibration standards canbecome an issue when calibrating at high powerlevels. The common standards used for calibrationare the open, short, load, and through. The open,short, and through standards are not a problemsince they do not dissipate any energy. The load,how-ever, does dissipate energy so when you arecalibrating make sure the standards can handle the power level.

19

Common problems of high-power measurementsSome amplifiers contain an automatic-gain control(AGC) loop. AGC loops attempt to keep the outputpower of the amplifier constant by adjusting gainto account for variations at the input of the ampli-fier. Amplifiers with AGC loops can pose a problemwhen measured on some network analyzers, espe-cially at high power levels.

Amplifiers with AGC loopsNetwork analyzers sweep across the selected frequency range while holding the power at thedesired constant power level. At the end of thesweep, some network analyzers might blank, orturn off the source, during the time it takes theanalyzer to reset itself and set up for anothersweep — the retrace time. Turning off the source isa problem for amplifiers containing an AGC loop.

When an amplifier with an AGC loop is measuredand the input signal is turned off at the end of thesweep, the AGC loop of the amplifier compensatesfor the turned-off signal by increasing its gain tokeep the output power level constant. When thesweep begins again, the network analyzer restoresits signal, and there is power again at the input ofthe amplifier, which has ramped-up its gain. Themomentary high output power can cause damageor destroy the amplifier or the analyzer’s receiverif the AGC loop cannot respond quickly enough.

Using a network analyzer that keeps its power constant during retrace will reduce the possibilityof destroying the device or damaging the analyzer.The 8753E keeps power constant except whenswitching frequency bands at 300 kHz and 3 GHz.The 8720E family of network analyzers allows theuser to keep the power constant or to blank duringretrace (the default is to have the power remainconstant). The 8720E family will briefly blank during band changes at 2.55 GHz. The 8722D blanksat 20.05 GHz as well. Be aware that this blankingoccurs if the analyzer sweeps across these bands.

On-wafer devices (pulsed measurements)A problem commonly encountered when meas-uring high-power on-wafer devices is the heatingup of the DUT. Devices on-wafer tend to heat upquickly because they lack sufficient heatsinking.This heating up requires that the temperature of the DUT be controlled in some way since theresponse of a DUT may change as the temperatureof the DUT increases. Two common ways to controlheating up is to pause the network analyzer betweenmeasurements, or pulse the RF and/or DC bias signals so that a constant DUT temperature ismaintained.

20

Presuming that a device will not overheat during a single sweep, pausing between sweeps can keepthe average temperature within bounds. A test-sequencing program that contains a user-definedpause between each measurement will do this auto-matically. More information on test sequencing isavailable in your network analyzer’s user’s guide.Pausing between sweeps dramatically slows downtesting time and might not be practical in some situations. Pulsed measurements can be a bettersolution to the heating problem.

In some cases pulsed RF and DC-bias measure-ments might be needed rather than the usual CWmeasurements made with a network analyzer.Pulsed measurements are used for several reasons.As noted above, pulses can be configured so thatan isothermal measurement is achieved. A pulsedsignal might also be used because it is representativeof the signals that the DUT encounters in actualuse (radar is a good example of this), or if there is interest in the transient response of a devicestimulated with a pulse, etc.

Pulsed measurements typically requires a test setdesigned for this purpose. The Agilent 85108 is anexample of a network analyzer system designed forpulsed measurements.

While not common, it is possible to make pulsed RF measurements with a standard network analyzerunder certain conditions. The ability to make pulsedmeasurements depends on the pulse repetition frequency (PRF) of the pulse used, relative to theIF bandwidth and the sampling rate of the networkanalyzer and the duty cycle of the signal. There are three cases to consider when determining ifpulsed measurements are possible using a networkanalyzer. (Note that we are referring to finding aDUT’s steady-state frequency response under pulseconditions, not a DUT’s transient time-variantresponse to a pulse. A standard network analyzercannot be used to measure the transient time-variantresponse to a pulse.)

1. The first case is when the PRF of the pulse isless than the IF BW of the analyzer. When the PRFis less than the receiver’s IF BW, the pulse side-bands pass through the IF filter and the modulatedRF (the pulse) cannot be measured (Figure 6). Sofor pulsed measurements with a PRF less than theIF BW of the analyzer, a standard network analyzerwill give inaccurate results.

Figure 6. When the PRF is less than the IF bandwidth,pulse sidebands are measured in addition to the centerspectral line.

t t

IF BW

PRF

Pulsed RF IF Filtering Modulated IF

f

PRF

PRF

21

2. The next case is for a PRF greater than the IF BW, but less than the sampling rate of the analyzer’s (ADC) analog-to-digital converter.Pulsed measurements are possible as long as thePRF does not generate a sideband at the samplingrate of the analyzer. For the 8753E and 8720E family, the ADC sampling rate is 16 kHz. If anyPRF sideband occurs at 16 kHz, it will be down-converted and be sampled by the receiver (Figure7). There-fore, signals with a PRF of 4 kHz, 8 kHz,or 16 kHz would not be allowed.

3. The final case is when the PRF is greater thanthe sampling rate of the analyzer’s ADC. In thiscase, the IF filter measures only the center spectralline (carrier) of the pulsed-RF spectrum as if itwere a CW signal in a nonpulsed network analyzerconfiguration (Figure 8). Since this type of measure-ment actually filters off the modulation sidebands,the PRF of the pulse does not affect the measure-ment. Only changes to the duty cycle affect themeasurement.

Figure 7. Pulse sidebands are downconverted and sam-pled by the receiver.

Decreasing the duty cycle of the RF spreads theenergy of this spectral line to the pulsed-RF side-bands. The magnitude of this center spectral line,will be proportional to the magnitude response of the DUT, but will also contain a magnitudedecrease caused by the loss of the energy in thesidebands. The amount of the decrease caused by the energy loss to the sidebands is calculatedusing the formula:

Pulse Desensitization = 20*log(Duty Cycle)

This desensitization factor decreases the dynamicrange of the measurement.

Figure 8. When PRF is greater than the IF bandwidth,pulse sidebands are rejected.

LPF 4 kHz

4 kHz*

1 MHz (from 1st LO)

996 kHz(2nd LO)

984 kHz(from first lower sideband of

a signal with a PRF= 16 kHz)

4 kHz

16 kHz

IF

To IFFilter12 kHz

Sample and hold

*Mixing product from 12 kHz IF and 16 kHz

sampling rate

t t

IF BW

PRF PRF

f

22

When the PRF of the signal is greater than theADC sampling rate, signals with a certain PRFmust be avoided since image frequencies may mixwith the receiver’s LO to produce the IF frequency.For example, the 8753’s first LO is tuned to 1 MHzbelow the RF test signal. For a 200-MHz RF test signal the LO is 199 MHz. Under pulsed conditions,say for a PRF equal to 2 MHz, the first lower side-band will fall at 198 MHz. This 198-MHz signal will mix with the 199-MHz LO to also produce a 1-MHz IF. Therefore, it is important to avoid signalswith a PRF of 2 MHz/N, where N = 1,2,3, ... (see Figure 9).

Figure 9. Any PRF which generates a sideband at the ADC sampling rate will be downconverted and sampled by thereceiver.

The hardware needed to make measurementsunder pulse conditions includes two splitters, apulse generator, and a modulator. Connect a powersplitter to the output of the analyzer’s RF sourceoutput. One arm of the splitter connects to the reference (phaselock) channel since the phase reference channel cannot be pulsed. The other arm of the splitter goes to a pulse modulator. The pulsed RF is sent to a second splitter to makeratioed transmission measurements. Ratioing isnecessary to remove the pulse transient response.The transmission measurement of interest is B/A(Figure 10).

Figure 10. Using a pulse generator and a modulator, pulsemeasurements can be made with a standard networkanalyzer.

4 kHz

4 kHz

199 MHz (1st LO)

1 MHz200 MHz

(RF in)

198 MHz (1st lower sideband

from signal with PRF = 2 MHz)

1 MHz

996 kHz (2nd LO)

To IFFilter

8753E with Option 011

Pulse generator

Modulator

R A BRF out

DUTDUT

23

AppendixNetwork analyzers—definitions and capabilitiesBefore discussing the measurements made with anetwork analyzer, it is important to have an under-standing of a network analyzer block diagram andhow analyzer makes measurements.

Network analyzers provide a wealth of informationabout a device, including its magnitude, phase, andgroup-delay response to a signal. The hardwareinside a network analyzer includes a source forstimulus, signal-separation devices for measuring a portion of the incident signal and for separatingsignals traveling in opposite directions on the sametransmission line, receivers for signal detection,and display/processing circuitry for reviewingresults (Figure 11).

Figure 11. General network analyzer block diagram

Network analyzers measure a portion of the sourcepower to use as a reference signal. The remainderof the signal reaches the DUT where part of thesignal reflects back from the device and part of thesignal transmits through the device. The reflectedsignal creates a standing wave consisting of bothforward and reverse traveling waves. The signal-separation equipment allows for the detection ofeach of these waves separately. After measuringthe reflected or transmitted signal, the analyzerratios it with the reference signal to measure thecharacteristics of the DUT.

Receiver/detector

Processor/display

Reflected(A)

Transmitted(B)

Incident(R)

SignalSeparation

Source

Incident

Reflected

TransmittedDUT

24

A network analyzer consists of either three or four channels (Figure 12). In network analyzer terminology, a channel refers to the hardware usedto detect a signal. There are one or two R channelsin a network analyzer. A portion of the stimulus signal is coupled out and goes to the R channel for ratioing. The A channel denotes the channelassociated with test port 1. The A channel meas-ures the reflected signal when making a forwardmeasurement (stimulus signal is at port 1), andmeasures the transmitted signal when makingreverse measurements (stimulus signal is at port2). The B channel, associated with test port 2,measures the transmitted signal during forwardmeasurements and the reflected signal duringreverse measurements.

Figure 12. Three versus four channels in a test set

Port 1

Transfer switch

Port 2

Source

B

R1

A

R2

Port 1 Port 2

Transfer switch

Source

B

R

A

3 Receivers

4 Receivers

25

Test setsThere are many variations of network analyzers,but one of the most distinguishing factors of a network analyzer is its test set. A test set is a collection of switches and couplers that directs thesource power and separates forward and reversetraveling signals. Some network analyzers, like theAgilent 8720E family of vector network analyzers,have an S-parameter test set with a transfer switchthat directs the source power to either port 1 orport 2 — allowing the analyzer to make forwardand reverse measurements. Other network analyzers,like the Agilent 8712C family of network analyzers,have a transmission/reflection (T/R) test set thatdoes not have a transfer switch and source poweronly goes to port 1 — these test sets allow only forward measurements to be made (Figure 13). In the past, test sets were sometimes not includedin the network analyzer, but almost all of today’snetwork analyzers have a built-in test set.

CalibrationKey to network analyzer measurements is calibra-tion. Calibration does two things. First, it estab-lishes a reference amplitude and reference phaseat a point in the system. Second, it determines theaccuracy of the measurement. Network analyzercalibration corrects for systematic errors (timeinvariant instrument and test setup errors) in themeasurement setup. Ideally, calibration correctsfor all errors up to the point where the DUT will be connected.

Figure 13. Transmission/reflection versus S-parametertest set

Figure 13-top• RF power always comes out of port 1• Port 2 is always receiver• Response, one-port cal availableFigure 13-bottom• RF power comes out of port 1 or port 2• Forward and reverse measurements• Two-port calibration possible

Transmission/reflection test set

Port 1 Port 2

Source

B

R

A

DUTFwd

Port 1 Port 2

Transfer switch

Source

B

R

A

S-parameter test set

DUTFwd Rev

27

Suggested ReadingUnderstanding the Fundamental Principles ofVector Network Analysis, Agilent Application Note 1287-1, literature number 5965-7707E.

Exploring the Architectures of Network Analyzers, Agilent Application Note 1287-2, literature number 5965-7708E.

Applying Error Correction to Network AnalyzerMeasurements, Agilent Application Note 1287-3, literature number 5965-7709E.

Network Analyzer Measurements: Filters andAmplifier Examples, Agilent Application Note 1287-4, literature number 5965-7710E.

Improving Throughput in Network AnalyzerApplications, Agilent Application Note 1287-5, literature number 5966-3317E.

8 Hints for Making Better Network AnalyzerMeasurements, Agilent Application Note 1291-1, literature number 5965-8166E.

Agilent Technologies’ Test and Measurement Support,Services, and AssistanceAgilent Technologies aims to maximize the value youreceive, while minimizing your risk and problems. We strive to ensure that you get the test and measurementcapabilities you paid for and obtain the support you need.Our extensive support resources and services can help you choose the right Agilent products for your applicationsand apply them successfully. Every instrument and systemwe sell has a global warranty. Support is available for atleast five years beyond the production life of the product.Two concepts underlie Agilent’s overall support policy:“Our Promise” and “Your Advantage.”

Our PromiseOur Promise means your Agilent test and measurementequipment will meet its advertised performance and functionality. When you are choosing new equipment, wewill help you with product information, including realisticperformance specifications and practical recommendationsfrom experienced test engineers. When you use Agilentequipment, we can verify that it works properly, help with product operation, and provide basic measurementassistance for the use of specified capabilities, at no extracost upon request. Many self-help tools are available.

Your AdvantageYour Advantage means that Agilent offers a wide range of additional expert test and measurement services, which you can purchase according to your unique technicaland business needs. Solve problems efficiently and gain acompetitive edge by contracting with us for calibration,extra-cost upgrades, out-of-warranty repairs, and onsiteeducation and training, as well as design, system integration,project management, and other professional engineeringservices. Experienced Agilent engineers and techniciansworldwide can help you maximize your productivity, optimizethe return on investment of your Agilent instruments andsystems, and obtain dependable measurement accuracyfor the life of those products.

Agilent T&M Software and ConnectivityAgilent’s Test and Measurement software and connectivityproducts, solutions and developer network allows you totake time out of connecting your instruments to your computer with tools based on PC standards, so you canfocus on your tasks, not on your connections. Visit

www.agilent.com/find/connectivityfor more information.

By internet, phone, or fax, get assistance with all your test & measurement needs

Online Assistance:www.agilent.com/find/assist

Product specifications and descriptions in this document subject to change without notice.

© Agilent Technologies, Inc. 1998, 2000, 2003Printed in USA, June 9, 20035966-3319E

Phone or FaxUnited States:(tel) 800 452 4844Canada:(tel) 877 894 4414(fax) 905 282 6495China:(tel) 800 810 0189(fax) 800 820 2816Europe:(tel) (31 20) 547 2323(fax) (31 20) 547 2390Japan:(tel) (81) 426 56 7832(fax) (81) 426 56 7840

Korea:(tel) (82 2) 2004 5004 (fax) (82 2) 2004 5115Latin America:(tel) (305) 269 7500(fax) (305) 269 7599Taiwan:(tel) 0800 047 866 (fax) 0800 286 331Other Asia PacificCountries:(tel) (65) 6375 8100 (fax) (65) 6836 0252Email:tm_asia@agilent.com

www.agilent.com/find/emailupdatesGet the latest information on the products and applications you select.