This issue was resolved. As it turns out, a careful examination of the specifications for the 8902A and other documentation cited below by randymax show the system is not likely to function below -95db above 1.3 Ghz.
Spurious responses from the signal sources appear to be major contributors. Thanks again to randymax, bryan and everyone else.
My 8902A will not measure below -100db above 1.3 Ghz. I am using an Agilent certified 11792A and 11793A down converter. My old 8340B sources are in spec for spurious responses. I am near certain I have the 8902A configured correctly but wonder if anyone knows of any tricks I can try.
Are you autotuning or manually tuning?
Are you making sure you recal/cal the sensor correctly?
Is you local oscillator set to the right frequency and power level?
Are you making sure you are keying in the correct special function?
Are you keying in the offset frequency correctly?
If you are auto-tuning try to manually tune it.
Are you getting any errors below -100
Try lowering your source in smaller increments
Don't have an answer for you. But we are getting a system set up pretty much identically. So I am not posting as an expert, but a co-user.
I believe our LO/downconverter generator is also an 8340B. We have a variety of 83623A/B's, 83630A/B, 8672x, etc.
I was just refreshing my memory about the frequency offset settings on the HP 8902A when using the downconverter. What difference frequency are you using? For example, if you are calibrating the UUT at a frequency of 2.00 GHz, what frequency are you setting the LO generator to? I believe I recall that the mixed down IF frequency should be between 10 and 700 MHz, so that once you key in the SPCL codes on the 8902A, that the frequency the 8902A Tunde RF sees is between 10 and 700 MHz.
Not sure if that is getting at the issues you have, but that's my two cents.
For best performance you should set you Lo to 120.053 MHz above the carrier...
That's what the procedure says but you could set your Lo to 100MHz and get the exact same result...
Thanks, the LO is set to 120.53 Mhz above the RF input and I am keying in 27.3 Spcl and telling the 8902A that is the offset, then keying it to manually tune to the RF input frequency.
I do have the 8902A configured as suggested in these replies. I have been talking to National Test equipment and the guys there have been very open about helping me over the phone. I've used them for repairs of our down converter and a 11792A sensor module. They are expensive but I think they're worth it.
I believe that since the problem only shows when the gain of the 8902A is 50 db or more the IF amplifier if defective.
I have found the AFMETCAL newsletter "8902A Microwave Measurement Receiver Tips" by Scott Weiss very helpful. It's in volume 28, issue #3 if anyone is interested.
I've had the same issues in the past. I have had very limited success at levels down to -100 dBm in the frequency offset mode. There are several things to consider already mentioned here like the offset frequency being centered in the 8902A internal lo octave bands i.e. 120.53 MHz being the first. The following is the Keysight description for the 11793A/8902S system which makes me wonder if measurements below -100 dBm were intended. There are product notes that state -100 dBm as the lower range as well. Also, depending on the 11793A options there are several different LO power levels recommended which can be found in the 11793A manual. From Keysight:
"The Keysight Technologies 8902S system extends the frequency range of the Keysight 8902A by adding the Keysight 11793A microwave converter and a local oscillator. With the Keysight 11792A sensor module, the system delivers the accuracy and resolution of a high-performance power meter to 26.5 GHz from +30 to -100 dBm. The extended system counts signals to 26.5 GHz with 10 Hz resolution and excellent long-term frequency stability."
"Extends the Keysight 8902A measurements to microwave
Accuracy and resolution of a high-performance power meter
RF Frequency Measurements to 26.5 GHz with 10 Hz resolution
Power Measurements from +30 to -100 dBm"
Hope this helps or someone can add to it to clarify.
Additional information can be found in the application note for operating the 8902A at microwave frequencies on the Keysight web page. The application note i'm looking at now is:
Copyright © 1985, 2000 Agilent Technologies
Printed in U.S.A. 9/00
5968-6296E
Years ago we operated a HP basic Signal Generator testing program, I believe the model # was 11708A, used a switch box so that a generator could be connected & left to the system to run. Anyway it had 2 configurations, one using the 8902A & 11722A and the other with the 11792A and 11793A with a LO. It could run a generator in whatever configuration was used, I was always reminding the operators that when a low freq generator was tested and the high freq configuration was used the RF power level testing cut off at -110dBm if I recall rather than the -127 as it would do with the 11722A config. Since it was all factory HP stuff I just presumed the higher freq setup just didn't have the dynamic range.
Nobody has mentioned the level of the LO generator.. That will have an effect on how low you can measure..
I normally set the LO to 8 dBm in my automation.
If the following link works there is some useful information in the event you haven't seen it before. Its a link to a document detailing low level power measurement techniques using the 8902A/11793A. It's old, but again there are references to sensitivity limits around -100 dBm. I really think the -127 dBm specification relates only to the 8902A up to 1300 MHz without the downconverter. I have seen many references to sensitivity @ -100 dBm to 18 GHz and -95 dBm above that to 26.5 GHz. I could be wrong but practical experience seems to agree. Anytime i write a Metcal procedure for a signal generator around the 8902A, I usually cut off the RF Level/Attenuation verification portion to -90 dB(m) when using the 11793A simply based on experience. (I'm operating an internal lab so I have the advantage of knowing the end user requirements though)
http://www.keysight.com/main/editorial.jspx?cc=US&lc=eng&ckey=1000002157:epsg:man&nid=-35196.536880545.00&id=1000002157:epsg:man
That document is very interesting. It does indicate the sensitivity of the system as I am using it is limited to -95 db. My system may in fact be operating in spec.
I am relieved we did not send the measuring receiver in for repair.
I had read another post that indicated the system sensitivity is limited above 1300Mhz. HP never seems to have been very forthcoming about this. It's a wonderful system but the documentation on it's specifications is difficult to understand.
I believe this resolves the problem. I thank all who posted regarding this issue.
And don't even get me started about Option 050. Headin out the door. Just saw your PM. Will answer in morning.
Your solution is spelled FSMR and NRP-Z37.
The old software I was referred to previously was the HP 11808A, it was really something in its day, it did have a "key" that had to go on the IEEE to operate.
Speak to m e about FSMR. I have a 26.5 GHz FSMR with the associated NRP sensor module. I like it and don't like it. I've tried going through the certificate of calibration and there are a LOT of useless measurements, and difficult to glean TRFL accuracy numbers. In a sincerely receptive way... talk to me about the TRFL numbers as relates to Opt 050 on the 8902A. We own some Opt 050 8902A's and even the cal kit for them (the attenuator that is a 355D with type N, cal void sealed cable and stick attenuator pad). All in a nice case.
We don't do a lot of volume in RF currently. But one of my back burner projects is to get accredited for RF Sig Gen cals, as there are some people who want to use us if we get accredited for that.
And I have the FSMR, and would like to use it and explore getting it accredited. But I have not found good TRFL numbers on the lengthy cert (from the OEM). We even sent, I believe, the module to Germany to R&S.
Hawaii,
I can't address your uncertainty questions right now, but I can tell you that I used both an Agilent PSA with N5531S Measuring Receiver option as well as an R&S FSMR daily for over six months, minimum, and the FSMR is much better and faster to use than the Agilent/Keysight PSA. Not that the Keysight doesn't meet its specifications, because it does. But the FSMR is significantly faster with a lower standard deviation in its measurements than its PSA/N5531S counterpart, in my experience. If it was my money, I wouldn't hesitate to purchase the FSMR as its a heckuva machine!
Are you aware that to verify the FSMR attenuation accuracy that the manufacturer sells a cal kit, the FSMR-Z2, calibrated by the PTB in Germany?
Good Luck on the Uncertainty Number Crunching!
I agree and have also worked with all of them in the past.. The PSA / N5531S has an issue with the CPU, it is just not powerful enough.
Last month I developed a new drives for the PXA for Metrology.NET, and I have to say Keysight got it right. The AM/FM/PM demodulation is super fast and the CPU has clock cycles to spare. I have not done a side by side comparison but I was able to test a PSG's demodulation in seconds.
The PXA doesn't have the N553x Configuration, but if uncertainties are an issue you are better off without the sensor heads. The heads... All of them add uncertainties to your measurement. The 117xx heads had a switch that introduces errors, and the N55xx heads have a power splitter and attenuator in the heads that introduces flatness errors.
My recommendation is to not use any of the heads. Take the ref reading directly with a power sensor then transfer that reading to the 8902A, PSA or PXA.. Then perform tuned level from that point down. (Note.. This is how Keysight does it!)
** It is important to note the PSA/N5531A can make accurate and faster measurements using the spectrum analyzer function. This is also available in the PXA.
I am recommending to all of my customers looking for an updated measurement system.. Look at the PXA or UXA with a Power Meter and Power Sensor, way faster and way more accurate.
Mike
CalLabSoloutions - Thank you for the information. I have run a couple of measurements of an 8340B attenuator at 20Ghz in TRF without the head and it works.
Since the only frequency the 8902A sees through the down converter is the offset (I am using 120.53 Mhz) I am now wondering why the 11792A head is there at all.
The head contains a switch to power sensor or thru path.
When you measure RF Power you are using the sensor, when you switch over to tuned RF you'll hear the switch get thrown and I you are using the 8902A to measure the IF (120.053MHz) relative to the RF level measured by the sensor or something like that.
I haven't used one of those heads since the military. I remember it being a pain to have to manually enter the cal factors. Much simpler to just use a power meter and sensor to measure RF power and connect directly to the 8902A for tuned RF.
I know what you're saying about entering the factors. We keep sensors married to a given 8902A (we have about three systems in service). So we only have to do it once a year for each system. I end up printing our the page from the operators guide each time, and after you've done a few, it's not too bad.