HP 8720C VNA - Displayed R dBm varies across power ranges but constant dBm out

[John_ITIC]

As the title says,

1) I set the 8720 to CW mode at 1 GHz for easier output power check.
2) I set the 8720 to display the power level measured on the R channel and enable the marker at 1 GHz.
3) I set the MENU / POWER level to -10 dB. The output is correctly -10dBm on my spectrum analyzer.
4) When changing the power ranges between [-10 to +10], [-15 to +5], [-20 to 0], [-25 to 5], [-30 to -10] the spectrum analyzer shows constant -10 dB.
5) However, the 8720 displays a varying power level -23, -18, -13, -8, -3 dBm respectively.

Images attached.

I'm confused by these results. I would expect the displayed R channel level to be constant -10 dBm.

Thanks,
/John.

[Wallace Gasiewicz]

John:
I do not have this fancy machine, so I am going to refer to my humble HP8714, which I have serviced too many times.
I am also going to refer to a trace thru the entire freq range  with a "thru" line attached to the two ports, rather than a single CW freq
Connect the "thru line" and look at Transmitted Power

The Transmit trace should be at ZERO dB
Go to CH 1 Measurement, Go to MORE, Go to Detection Options, Go to Narrow Band
You should see    A    B    R    A/R     B/R
The R trace  and the B trace should be at approx the dBm power level you choose
B/R is the Transmit trace, this is Transmitted Power referenced to Power Level, again should be Zero This is a dB level, not dBm

The A value is Reflected Power, A/R is Reflected Power referenced to Power Level. This is a dB level, not dBm
The B trace is power through the "thru" line and should be at about the Power Level you choose, this goes away if you disconnect the "thru line"
The R trace is a "Reference" of the "Transmitted power" This level is there whether or not the "thru line" is connected and should be at about the Power Level you choose

Check this and tell me what you see.

[John_ITIC]

Hi Wallace. Your test is essentially the same as mine  except you use a sweep while i used a CW to be able to confirm output level on my spectrum analyzer. I have checked before and I know the output power is always correct, both in CW and swept mode. It is flat all the way up to 20 GHz.

I have attached the R display when outputting 0dBm in the [-10 to +10], [-15 to +5], [-20 to 0] ranges. Just like before, in CW mode, the R display is -13, -8, -3 dBm too low.

Per the attached service manual excerpt, the A58 M/A/D/S unit is supposed to split the RF into two directions. One going towards the output port (confirmed correct) and the other going towards the A64 R sampler, which ultimately ends up as the S11 dBm reading on-screen in this test.

I'm suspecting something is wrong with the A58 M/A/D/S unit as it appears to consume too much current. The original problem with this 8720C was that the A58 M/A/D/S unit apparently started consuming too much power (1.3A) burning out a 2.2 Ohm / 3W resistor in the A59 control board. I replaced the resistor with a 5W version (which does get very hot too) but the unit came back to life.

https://www.eevblog.com/forum/repair/hp-8720c-vna-a58-modulator-amplifier-detector-splitter-how-to-test/

I now will have to back and look closer at the output from the A58 going towards the R sampler.

Thanks,
/John.

[Wallace Gasiewicz]

II looked at some of your unit's manual. Much different from my 8714.
Maybe the problem is in the attenuator, drawing too much signal?
Does the attenuator have mechanical relays?

[John_ITIC]

No, it's not the attenuator as the port output is correct across all power ranges and power settings. It is only the displayed (R) value in the analyzer that is incorrect. In the previously posted block diagram (from the service manual) you can see that the reference signal (R) output from A58 is different from the output going to the attenuator.

I could confirm that the R signal is varying with the attenuation range although the port A output is steady at 0dBm.
I also figured out that the R output does NOT change if the ALC is turned OFF. However, then the output DOES change.

As the A58-R output is varying as the menu attenuation is modified, it must mean that the A58-R output power is controlled via software at the same time as the output attenuator selection is changed. The A58-R level must be adjusted to keep the ratio between the R signal and the signal from the couplers?samplers constant. Either this A58 control signal is not working correctly or the A58 somehow is broken.

So, somehow, related to the ALC/A58 functionality. I'm trying to make sense of it all. The service manual is not much help.

The hear clicking when the attenuation is changing so mechanical attenuator.

Thanks
/John.

[Wallace Gasiewicz]

OK. I am looking at the manual. The R comes from the troublesome A 58 board and goes to the attenuator and then the relays and then ports.
It also goes to the A 64 R sampler
If the power out is changing correctly. I suppose the attenuators are just fine.

Troubleshooting the R path is on page 245 of the manual.
But perhaps you can measure it by connecting the R output of the board to your SA. But I do not know exactly what this value should be. I suppose it should varry insice the steps providedby the attenuator.
I guess I am not that familiar with your unit's functioning and am sorry I cannot be of more help

Wally

[John_ITIC]

Hi Wally,

But perhaps you can measure it by connecting the R output of the board to your SA.

Yes, I did this (see above) and the R signal changes in amplitude as it comes out of the A58 M/A/D/S unit (which is a sealed box which seems difficult to open). See attached picture. The A58 unit is the copper-colored box to the top/left. The rigid coax most to the left is the output to the attenuator, which ultimately leads to the output ports (per prior posted block diagram). The right rigid coax goes to a rear panel jumper (this is where i confirmed that the R signal varies when the power range is changed, with constant output level). After the rear loopback jumper, the R signal comes back and goes into the R sampler (in the center of the picture).

I think I next will have to figure out the exact functionality of the ALC circuit. It would be logical if the attenuation of the A58 and the attenuator always is the same such that the ratio A/R and B/R is contant regardless of the power level range and power setting. It seems the ALC ON/OFF is now either changing the attenuator OR the A58 R attenuation but not both at the same time.

Thanks
/John.

[Wallace Gasiewicz]

Manual Page 330 of  482
A69 Step Attenuator    The step attenuator provides coarse power control for the source signal. It is an electro-mechanical attenuator, controlled by the A7 CPU, that provides 0 to 55 dB of attenuation in 5 dB steps. It adjusts the power level to the device under test (DUT) without changing the level of the incident power in the reference path

So the attenuator provides 5 dB "ranges" within which the other stuff provides power level control (including ALC)
I do think these "ranges " are the same as the ranges on the screen.
Theoretically you should hear a click when you change power by 5 dB.

Did you say that the power to the DUT was unchanged by changing the ranges?

Edit Sorry,   I do NOT think these ranges are the same as those on the screen

[John_ITIC]

With ALC = ON : The output level does NOT change when the in-menu attenuation ranges change with the dBm value set the same. However, the R signal level changes within the instrument (as verified via rear panel Reference OUT/IN loopback coax jumper).

With ALC = OFF : The R level stays the same when the in-menu ranges change (dBm level the same) BUT now the output level changes.

I would think neither R or output should change level when just the range (but not actual dBm power level) changes.

I have not had time to think deeper about this. No further clues are available in the service manual.

Thanks
/John.

[John_ITIC]

Okay, I have found some more interesting things:

1) The attenuator has 5dB granularity, each step corresponds to the power range options, as you suggested. Thank you for this lead.
2) The ALC control is used for the fine adjustment within each step. It can be directly entered or tuned via the adjustment knob.
3) The changing ALC value affects both the source power as well as the Reference equally much.
4) Only the source power goes through the attenuator.
5) If the output power is configured the same (via MENU/POWER) and the power range (attenuator) changes to a lower range, the ALC level is adjusted up to bring the output power back to the constant value.
6) Because of 3-5, the A/R ratio will change as the power range is changed, although the source power is constant.

The below tables show the output level vs the Reference level  as measured on port A and rear Reference port.
The rear port Reference level was measured via a power splitter so -6dB lower reading than if direct connection OUT/IN.
The table values come from two external spectrum analyzers.

0 dBm in range	Pout (dBm)	Pref (dBm)
-10 to +10	-0.7	-25
-15 to +15	-0.7	-20
-20 to 0	-0.7	-15

-15 dBm in range	Pout (dBm)	Pref (dBm)
-25 to -5	-16	-25
-30 to -10	-16	-20
-35 to -15	-16	-15

-30 dBm in range	Pout (dBm)	Pref (dBm)
-40 to -20	-31	-25
-45 to -25	-31	-20
-50 to -30	-31	-15

-45 dBm in range	Pout (dBm)	Pref (dBm)
-55 to -35	-45	-25
-60 to -40	-45	-30
-65 to -45	-45	-15

I'm curious whether it is normal to have the ALC adjust the Reference signal level in such a way. It causes the A/R and B/R not to be constant across the different power ranges.

Thanks,
/John.

[grendle373]

Run tests 21, 22, and 17 ( system, service, tests) and note any failures. that should get you started in the right direction.

[John_ITIC]

Run tests 21, 22, and 17 ( system, service, tests) and note any failures. that should get you started in the right direction.

All these tests passed. The 8720C appears to work properly except for the observation that the Reference signal is not attenuated the same as the output signal across the various power ranges. The ALC (fine power control) does vary both R and output signal equally within a selected power range.

For instance, when measuring S11 with port 1 shorted, the display shows (at 10 GHz) the following S11 values as a function of the power range set. Note, the output power is constant -5dBm across these five power ranges.

Power range	S11 @ 10 GHz
-10 to +10	+10dB
-15 to +5	+5dB
-20 to 0	0dB
-25 to -5	-5dB

Edit: If anyone could compare my 8720C reference signal behavior with the same or similar VNA, such as 875X, 8719 etc it would be great.

Thanks,
/John.

[grendle373]

Yes, your unit is working correctly. 

R power level should not change with the attenuator but will vary some with the ALC, which is used in the MADS. R channel being the reference, is not run through the attenuator, switches or couplers, but is picked off directly from the MADS to the R sampler.


I wasnt sure how your set up was running originally, which was why I recommended 21,22,and 17. 21 & 22 check the ports, A&B sampler, switch, attenuator, and also the mads. We use those to isolate the ports, switch, attenuator, source, etc when troubleshooting. Test 17 isolates the ADC.

as an aside, the service manual can be found here: https://www.keysight.com/us/en/support/8720C/microwave-network-analyzer.html

Signal seperation can be seen on page 15-14.   
 

[John_ITIC]

Thanks grendle373.

I have spent weeks carefully reading the service and operating manual so i'm quite familiar with the instrument architecture by now. It is however still not clear to me how it is normal that the displayed S11 value can change depending on what the output power setting is. As things work now (which you say is normal) I get +10 dB to -5 dB error in the S11 reading depending on what power range is selected. Per the below table (with a -5dBm source power and short termination), the S11 value is only correct in the '-20 to 0' range (all R is reflected so 1/1 = 0dBm return loss), all others are incorrect due to the A/R ratio not being constant.

Power range   S11 @ 10 GHz
-10 to +10   +10dB
-15 to +5           +5dB
-20 to 0           0dB    <= Only this is correct.
-25 to -5          -5dB

Edit: I think i have figured this out now. As described, after having configured the power range and fine tuned the final power level, the A/R ratio will vary, depending on the range and power level set. An S11 OSL calibration is then needed to make the instrument display correct S11 data. I incorrectly assumed that the instrument would give ball-park correct S11 readings without having to do a simple OSL calibration.

The attached pictures now show the same S11 reading for two different power range values (both required separate calibration runs). If the attenuation is changed, the calibration becomes invalid and the instruments warns about this in red text. Fine-tuning the power level within a power range does not invalidate the calibration.

So, I think the instrument is now working correctly. I will now get on with my learning how to use it properly.

Thanks,
/John.

[Wallace Gasiewicz]

As I said before, I have an older VNA the 8714. This is somewhat different
Your explanation seems strange, does the manual state that calibration is necessary when you change ranges??
Calibrating every time you change the range seems like something that would make the instrument harder to use.
Or perhaps are the calibration of the different ranges are meant to be stored and then come up when ranges are changes?

[John_ITIC]

Hello Wally. The manual is vague regarding explaining that the calibration is invalid when the range is changed but the instrument clearly tells the user per the image in my previous port. The is only one current calibration and that is for the range in which the instrument was in at the time of calibration.

Once the range is changed, the warning appears and the corrections are turned off. If the range is then switched back to the original one (that the calibration data are valid for), it is possible to enable corrections again.

Unless someone can confirm that their 8720C behaves differently, I will have to assume this is just the way that the 8720C works.


Thanks,
/John.