How to use the new VNWA Power Sweep Utility Preface: From VNWA experimental version 36.6.9.5, released November 5 2015 and onward, the new VNWA Power Sweep Utility is available. The purpose of the utility is amongst other usages, to characterize, at a fixed frequency, the linearity of an amplifier and e.g. find the 1dB compression point. To some extend the absolute levels in dbm can be measured, provided the test frequency are below 100MHz, as the nominal TX output is -17dBm of the VNWA, and only drops 0.1dB at 50MHz and 0.3dB at 100MHz. Details about this drop of output for the fundamental and the harmonics versus frequencies can be studied in the document http://www.hamcom.dk/vnwa/how to calibrate the VNWA Spectrum Analyzer using the TX output as signal source.pdf where below graph and table for level versus frequency is documented further. An updated report is at the link http://www.hamcom.dk/vnwa/rf Generator-Output-VNWA3-and VNWA2 rev1.pdf These levels measured up to 1300 MHz are based upon VNWA3 setting of the pre-multiplier to x3 and not auto, so the complete setting for VNWA3 is 12 x auto - x3-12 x auto, which is identical to VNWA2 standard settings. The measured data, in the report, can then be used for both VNWA2 and VNWA3. The Power Sweep Utility can be sensitive to spurious signals, feed from the TX port to the mixer, and in particular for high frequencies, so optimal performance is obtained when a bandpass filter is inserted in the TX part before any DUT is exposed to the TX signal. This effect will be demonstrated later in this report at 1120MHz. However at low frequencies e.g. up to 250MHz or even more superb result obtained without filter or just a 500MHz LP filter in the TX path prior any amplifiers. For high frequencies above e.g. above 500MHz where the signal level are low, the use of a 600 or 700MHz high pass filter is needed prior any active devices. The Power Sweep function Utility take advantage of the VNWA TX level setting, which can reduce the nominal level of -17dBm in small steps over a range of 55 to 60dB. Few fact for the Power Sweep Utility are essential to realize and understand, before we go ahead with the usage: - The reflection and Thru Match calibration is harmful to the function so only Thru calibration is allowed. - Every time the Power Sweep Utility is activated/started, you are prompted that the calibration is invalidated and any activated MC (master calibration is turned off) so the sweep results, in this condition, is not absolute or relative but represented by a series of db levels. - When Power Sweep Utility is started the frequency is also automatic set to the center frequency of the present sweep range, with scan set to 0Hz. The test frequency can be changed at your will. - If you, while the Power Level Utility is active, perform a Thru calibration then the sweep result is relative with e.g. 0dB as maximum. Then you are able to calculate the absolute levels, for frequencies below 100MHz quite accurate, and above fairly accurate, pending the data in the referenced document, taken into account that when testing amplifier with gain, you must insert attenuators in the path to the RX port, not to surpass the normal level at the RX port and thus for the mixer circuitry, which by passive DUT s never are above the TX level -17dBm. -When Power Sweep Utility is active, use 100ms per point, and a fair number of points, which might be 10, as the sweep measurements are averaged after every power level sweep. After performing Thru calibration with 100ms per
point you may reduce the time per point to speed up the process and observe the impact if any. The Power Sweep Utility might even be used to investigate the linearity of the RX mixer (if you have the courage) as max allowed input is 0dBm, by using a highly linear amplifier, combined with an step attenuator in the TX tableo RX path. BUT BE VERY CAREFULL, as the VNWA kick out a high power burst to sync, in the beginning of every sweep. To facilitate such linearity measurements, the Power Sweep Utility can besides display the stepped attenuations also display the power level in dbm. You may enter a TX reference level, taking into account the amplification of a tested amplifier and any attenuator added at its output, to prevent the VNWA RX port to be driven above the level where no compression is taking place. You have then the choice to measure the absolute levels on the output of the amplifier as a ramping up curve or the relative or absolute attenuation levels as a ramping down curve. We will see later how that is done. Many initial details to understand, but let us now get started: Above a traditional start condition with sweep from 1 to 500MHz. Note the MC is enabled To start the Power Sweep Utility click on Measure and subsequent Power Sweep and the warning appears once, only until the Utility is closed, then it appears again next time the utility started.
The subsequent Info Tab is providing a very good information about the next steps to perform As further preparations we set the sweep to 10 x 100mS points by choosing Settings and Sweep And we see the center frequency 300.5MHz set to 0 Hz span and time span calculated to 900mS You may at this point change the center frequency by a click on the Center/Span indication and we chose 25MHz as an our example. The next tab is the Attenuation Grid tab, which shows the lines where you enter, in the left column, the wanted Target Attenuations in db, between 60 and 0 db, and the Actual Attenuation/dB is calculated in the right column. By a right click in the Attenuation Grid, you can select to edit an existing list by insert an delete lines, clear the entire list, save the list under a descriptive name, and load a saved list. You may also select the auto-fill list as shown in next screen shot.
The auto-fill list is a convenient starting list to get experienced with the functionality Next tab is the Analyses and Plotting settings The analysis function is standard set to db(s21) and two results can be chosen, Real Result and Complex Results, and for S21 we chose Real Results. The draw guide line to Re() grid may be selected (box ticked as seen) and we will do so already.
Selecting the Sweep Result tab and by a click on Sweep we have the first result where TX and RX directly connected with a test cable. This is the non calibrated condition. If we enable the draw guide line to Re() grid (already done) and right click on the Sweep Results screen, we get some many options. We may fit a straight line to zoomed range with slope= -1 or fit straight line to a zoomed range. Zoom is called box zooming and done by holding down the Ctrl key while drawing a box around required region. We may select more than one zoom step and by selection unzoom we go back to full screen. We first select fixed slope = -1 The save measurement data is possible both as *.dat, *.txt and *.csv for loading into a spreadsheet for further processing. A red thin line is drawn on top of the Sweep Result and extremely precise on top, proving the high performance of the VNWA. Next we will zoom in around the 1dB steps at the top.
Still extremely precise as seen below proving high degree of linearity. Left click with the mouse on the plot area provides a cursor function so a data point can be displayed in the lower part of the plot screen A small experiment with a wider Atten/dB step range and monitor the range for fluctuations in measurements. This test made at 50MHz. By disable the do not normalize. and in Calibrate Level the Display relative attenuation only we get an interesting trace.
Below sweep is in uncalibrated condition and we observe that from TX Level -45dBm and lower the accuracy starts to be poor. Below is a sweep after a S21 only calibration and we observe in the list that attenuations for the entire level range of 55dB does not deviate more than 0.05dB. This indicates a range where accuracy is excellent and stable calibration over the entire range. If we in the Calibrate Level tickmark Display relative attenuation only then the TX level of -17dBm is referenced as 0dB and the range of accurate calibration is easier to monitor
Repeating the test in S21 only calibrated mode at 300MHz the TX level range for accurate calibration is reduced to some 30dB, due to the lower TX signal level at 300MHz by about 18dB, which also is valid for the LO DDS delivered to the mixer and thus the noise floor also is at a higher level. The limitation in range for accurate measurement might be pending harmonics and aliases but for above tests but a 500MHz low pass filter had no effect. For comparison with and without low pass filter we might utilize one or more of the 4 Mem location. Below Mem2 and 4 enabled. Repeating the experiment at 108MHz where TX level still close to -17dB with and without a 200MHz low pass filter, we might see the effect of an low pass filter and below without the low pass filter stored in Mem2 with green trace.
And below shown with 200MHz lowpass filter as the blue trace as being live trace. Even if stored to Mem4 it is now seen as live trace is always on top of the mem trace. The range now slightly or no better with filter. Remember to S21 only calibrate after LP filter inserted. Test at 200MHz with and without 200MHz LP filter show a small improvement With addition of the 200MHz low pass filter, as seen below a 5 db more range was gained So lesson learned is to test you setup before judging results.
Further option to use when right click on the Result screen is the following. Y-offset. The stored data im Mems can be shifted to match the guide line whether it is at a fixed slope or a straight guide line fitted to a segment, which has been box zoomed. The y-offsets can be cleared. Clear mem. Mem1 to 4 can be cleared individually or all in one go. Save measurement data to file. The save measurement data screen is seen below, where a descriptive file name and data extension entered Save memory data to file: The save memory data to file requires a further step as you must select which Mem number to save. Only activated Mems are selectable.
Load data from file to: One or more Mem data stored can be loaded as well stored Measurements data. Mem data can even be loaded to a specific mem number from a different saved Mem number as well from Measurement data. Now to a practical example by finding the 1dB compression level for an amplifier Preparing for the next experiment we perform a Thru calibration (now at 50MHz for practical reasons (as an added amplifier shortly introduced) and by repeating the sweep we normalize the Sweep result to 0 db (Calibrate Level/Display relative attenuation only) Please note in the lower part of the plot screen the fitting result are displayed. Shown are the fit x=0.007067-1*x and for max y-error = + /- 0.036 Thus the fitting can be evaluated Next a demonstration of determination of a 1dB compression point of a Phoenix PA661C amplifier 10-1000MHz with gain 27dB and medium output power of 14.5dBm and a 1dB compression point of 14.5dBm. As the VNWA standard RX input power is -17dBm an SMA inline attenuator of 40dB inserted in the output of the Amplifier to protect the RX port as the level for 1dB compression is approaching -17dBm. A different list file loaded of 0 to 60dB to further see the limits for the dynamic range of the Power Sweep Utility in combination with TX power adjustment limitation. The TX out level of -17dBm and the gain of 27dB can only provide +10 dbm at the output, so a further amplifier added in front of the amplifier under test, with a 1db step attenuator at its input. This amplifier is a version from the LNA4All, really nice and cheap, with 25 to 2500MHz specification.
Below the first attempt is seen with only the PA661C inserted. The Thru calibration made with the TX and RX port directly connected so the 0 db is equal to -17dBm. The measured max output is -11.8dBm and by adding the 40dBm attenuation, we reach 28.2dB which is the amplification in the amplifier at 50MHz. Then we add the LNA4all and 1dB step attenuator in front of the PA661C. At the output of the PA661C is also connected a Minicircuit power splitter to be able to measure the output with a power meter identical to the input to the RX port. We fit a straight line slope -1, for a zoom region being linear. And we see a clear compression and one further zoom attempt is done, to close in one the 1dB steps at top, after the zoom has been unzoomed (which does not change the fitted line)
A 1dB compression found at 4dB attenuation with a level of -8.16dBm at 4dB attenuation. 0dB reference is -17dBm so in addition to -8.16dBm the 40dB attenuator can be added as well as the -3.3dB in the power splitter, summing up to an 1dB compression output from the PA661C of 18.14dBm. The 0dB level measured with the Power meter is - 22.8dBm but 1dB compression occurred with 4dB attenuation and added with afore mentioned 40+3.3dB gives 16.5dBm. A difference of 1.64dBm. Checking the TX out with the power meter shows -17.4dBm so the difference only 1.24dB in reality. I any case the 1dB compression is better than spec by 2 to 3dB.
To take the Power Sweep to the extreme we do the same sweep with 1.33 ms per point. No visual difference but 0.2dB more in the 1dB compression so in fact no difference.
To further stress the Power Sweep Utility I will test it at 1120MHz using a high quality bandpass filter with +/-40MHz bandwidth with steep slopes on both sides. I will also test thee full performance of the VNWA input circuitry consisting of Mixer, LO and internal AD converter. As we do not know the TX level at 1120MHz I will arrange for a method to measure the actual level feed directly at the RX port, and use the alternative presentation with direct power level display and perform Calibrate Level as indicated with the red pointer in the picture above. To avoid any non linearity of other devices the amplifier is placed in from of the filter and only passive attenuator connected in series with the output of the filter if any needed. The setup is the following. From the TX port the 1120MHz signal (which also contains a fundamental frequency besides harmonic and aliases) is fed via a 700MHz SMA inline high pass filter and a 10dB SMA inline attenuator to the two amplifiers in series where the first is the LNA4all and the last one the PA661C amplifier. The output from the last amplifier is feeding the input of the 1120MHz filter and at the output of the filter a further 10dB attenuator terminated with a Minicircuit power splitter ZFSC-2-5 which drives the VNWA RX input and the HP437B power meter. The first test will be with this setup to demonstrate complete uncompressed linearity and determine the TX drive level to enter into the Level calibration prior to remove the 10dB attenuator at the output of the filter to drive the VNWA into compression. Click on Calibrate level and remove the tick mark in Display relative attenuation only and we enter the level at 1120MHz read from the power meter. In this case -15.17dBm slightly higher than the standard level -17dBm. Next close the setting window and select Sweep Results and click on Sweep A clean straight trace seen already fitted with a slope -1 line for the range seen on next page
Line fitted for -45 to -20dBm TX level The compression only 0.1dB at the -17dBm standard level By removing the 10dB attenuator at the filter output and change the calibration level to -5.52dBm (power meter reading) we run a new sweep and fit a straight line to the linear part and observe compression
1dB compression found at -7.53dBm at 1120MHz IMPRESSIVE INDEED
A repetition performed at 50MHz where no filter used and a 1dB/10 db step attenuator inserted from the TX output to the first of the two amplifiers. At first the 1dB/10dB step attenuator set to a TX level of -17dBm and Calibrate level set accordingly. As expected a straight line. Compression at -17dBm 0.1dBm if any at all as fitted line runs in parallel
Step attenuator reduced by 10dB and TX level changed to -7.1dBm according to power meter reading and 1dB compression at -9.86dBm. Tom Baier has done a similar test but used an external sound card as explained below and writes: I have used the power sweep tool to determine the linearity of the VNWA frontend + IF. To ensure that the ADC does not saturate, I have used my E-MU0204 instead of the internal codec. Result at 50MHz: 1dB compression occurs at -3.5dBm. At -17dBm the compression level is 0.04dB, which I considered acceptable at the time of design. Best regards, Tom
Comparison with Tom Baier s results: The internal ADC provides a 1dB compression at -9.9dBm and the External E-MU0204 sound card a 1dB compression at -3.5dBm a difference of 6.4dBm. In both cases a TX level of -12dBm can be considered as linear at 50MHz. The standard -17dBm is a very safe level even up to the highest frequency of the VNWA for such measurements. A final test for the PA661C amplifier using the setup with TX output via 1/10dB step attenuator feeding the LNA4all and then PA661C amplifier. The output then via a 40dB SMA inline attenuator to the power splitter connected to the VNWA RX and the Power meter. The 1/10dB attenuator is then adjusted until compression is observed and by entering the Power meter reading -24dBm +40dB + 3.3dB = +19.3dBm in the Calibrate level then the 1dB compression for the PA661C can be determined without further calculation. 1dB compression at 16.65dB, the same as found before using the VNWA standard level -17dBm without power meter measurements.
A swap of the two amplifiers allows us to find the 1dB compression for the LNA4all amplifier. The TX level adjusted accordingly to -24.8dBm + 40db +3.3dB=+18.5dBm 1dB compression found at +15.7dBm at 50MHz not bad for such a cheap device Final conclusion: The Power Sweep Utility is a superb addition to the VNWA software. For ordinary usage in the HF and VHF bands up to 150MHz there is no need for filters. As demonstrated even at the highest frequencies of the VNWA by using filters and power meter, accurate measurements can be made. At UHF around 430 MHz a bandpass filter is recommended and above 500MHz a high pass filter e.g. 600 or 700MHz is a must if the amplifiers are placed in front of the bandpass filter for the frequency of interest. Kurt Poulsen de OZ7OU 8. November 2015