HP Agilent Spectrum Analyzer 8595E CAL AMPTD error

[Miti]

Buy one of these:

https://www.banggood.com/35M-4_4GHz-PLL-RF-Signal-Source-Frequency-Synthesizer-ADF4351-Development-Board-p-1129041.html?rmmds=myorder

and one of these:

https://www.banggood.com/3Pcs-3_3V-8MHz-ATmega328P-AU-Pro-Mini-Microcontroller-With-Pins-Development-Board-For-Arduino-p-980290.html?rmmds=myorder&cur_warehouse=USA

Edit: You'll need one of these as well:

https://www.banggood.com/3_3V-5V-USB-to-TTL-Converter-CH340G-UART-Serial-Adapter-Module-STC-p-1232728.html?rmmds=search&cur_warehouse=CN

[charlyd]

ok i have two problems.  For one problem i got an other option 130 board  08591-60055 in and the tested that in the 8595E .. result Working now. 

This rises the next question is it possilbe to repair the board.
so i can use it in my 8594E. now i have two boards and i can compare for diagnosing.  i saw a lot of test points and this block diagram of the Narrow resolution bandwidths.

are there extender boards for these models SA?
connector type : AMP Brand 5-534204-5 on the card.  ( see picture )
connector type:  AMP Brand 5-102567-4 on the motherboard ( see picture ).  -> Connector Header Through Hole 60 position 0.100" (2.54mm)

If somebody can help me out with one or both of these connectors from a broken card or motherboard,  please pm me.


it is a start.

[charlyd]

In post #5  i was told by @analogRF you say : try to play the attenuator contacts several times and then run the cal again . i decided to refurbish the attenuator rubbers.

[analogRF]

In post #5  i was told by @analogRF you say : try to play the attenuator contacts several times and then run the cal again . i decided to refurbish the attenuator rubbers.

was it intermittent? if it was not, there is no need to mess with it really. Mine was intermittent and although it would work perfectly for a long time after playing the contacts several times but if i didn't use the SA for a couple of weeks, then the next time it would go crazy again so I had to repair it.

It's just the o-rings that dry up and break apart. Dont use any kind of oil. I bought  a bag of o-rings from china and they have worked for me on many occasions but how long they will last I will never know...

[Miti]

Thanks Mark!

Yes, that was the first thing I did when I got it, I saved the calibration data. But is the flatness data and the attenuator corrections all I need to save? It is not too much to restore if needed, but I see much more data there.
Also, I quote from the revision history: "Intro of 2X memory (4 2-megabit EEPROMS)". Do you know what EEPROMs are used? 27C020 is an EPROM not EEPROM.

The 8591E needs timebase, flatness, and attenuator to be backed up.  Once this data is restored, you can run the self-cal procedures described in the section "Instrument Calibration after Reloading the Correction Constants" to re-compute everything else.

The 27C020 is certainly an EPROM.  "EEPROM" in that document is a typo.  I've done the same thing many times myself.  I hardly ever deal with EPROMs anymore and my fingers are always typing "EEPROM" instead.

  https://octopart.com/search?q=AM27C020&currency=USD&specs=0

I ordered 5 EPROMs from UK, erased them, programmed 4 of them and my HP8591E has a new firmware, 980615. The only side effect was that it initialized the user memory and I've lost the EMC personality but that was quickly re-loaded. The calibration did not remapped, it stayed the same.

Edit: It passed calibration and confidence test after upgrade.

[Bicurico]

Hi,

I have a HP 8594E with the CAL AMPTD error.

I built the crystal shorts (but the resistors I used add up to 89.x Ohm instead of 90.0 Ohm - hopefully that is not critical).

Then I followed the first part of 2.5 Crystal Alignment.

I got a much better Marker Delta as described on page 52.

This did not fix the CAL AMPTD error.

So i carried on with the LC Alignment and LC Centering. This did not fix the CAL AMPTD Error either.

I then tried the LC Amplitude correction, but was not able to finish it due to wife calling...

However, the LC alignment and centering instructions seem wrong to me - it seems that the author did a copy&paste of the steps to move the shorts and they make no sense. I did follow this, positioning the shorts as I thought it should be, but I am not 100% certain I did it correctly.

I will try to repeat this ASAP.

Any suggestions, tips and tricks are welcome.

Thanks,
Vitor

[analogRF]

CAL AMPTD error is not always the result of filter misalignment.
also you dont need to mess with all filters...just notice on which RBW(s) you are getting bad results.

in any case I would mess with filter alignments as a last resort after everything else is ruled out...

[Bicurico]

I tried to swap A13, A12 and A11 between a known good unit and the problem remained. I assume that A13 and A11 need to be aligned for each specific SA, but I was kind of hoping that swapping A12 would solve the error.

As it did not, I tried to do the crystal alignment. Truth is that I never had done it before and as a hobbyist, reading the manual is not 100% clear, so it took me a lot more time than expected and I still am not 100% sure I did it right.

Regards,
Vitor

[Bicurico]

CAL AMPTD error is not always the result of filter misalignment.
also you dont need to mess with all filters...just notice on which RBW(s) you are getting bad results.

in any case I would mess with filter alignments as a last resort after everything else is ruled out...

CAL: RES BW AMPL FAIL occures when Res BW is set to 3MHz.

Any ideas?

Regards,
Vitor

[Bicurico]

After tuning LC it fails at 300kHz.

[analogRF]

I would try to troubleshoot the circuit board first and if all signals are ok then proceed to mess around with the tuning. at least that's how I approach my repairs.
all RBW >= 100KHz are handled by the two LC filters. The two crystals are shorted by a FET or MOSFET i think. If that Fet is not shorting the crsytal
when RBW>=100KHz, then you get errors. Different RBW>=100KHz  are selected by changing the bias on two varactors in the LC filters.
so you must make sure those 4 BW control signals are coming and they are within range and the bias transistors (for the varactors) are OK
if the biasing transistor is ok in one of the LC poles and dead in the other you can tune it kinda half ass but it will fail in some of the RBWs
there are 3 BW control signals that come into the bard and a 4th one is generated from one of them on the board. I dont have the schematic
on my work computer here to tell exactly where and what they are...so I would suggest you first focus on the BW control signals being in range
the varactor biasing circuit being ok and the XTAL filters being shorted by the FET or MOSFET when BW>=100KHz
if all is ok then perhaps tuning will be necessary
there are quite a few things that can go wrong on that board, it's not just the alignment. I think sometime ago someone had a problem where a capacitor on that board was drifted in value quite a bit...so I guess that could be one possibility too even though it's rare...

[charlyd]

unless, you have option 130. it is a easy check to pull the option board . it fixed mine.
after that i put in a other 130 board and all in calibration. but of course not all amptd errors are related to the 130 board. and for sure when it is not in..


so my last repair point is the amptd is a bit low ( see below) . But i found a new rubber set so i wanna swap them. i saw this could be an easy job and worth doing to refurbish the attenuator.
Comparing both pictures. Leftside -Internet Picture- and rightside -my picture-.

My amplitude looks much less from what i see from nearly all harmonics..

[Bicurico]

I am getting CAL: SPAN SENS FAIL now...

arghhh!

[analogRF]

you need to go back to the basics, I bet the original error was not even because of the RBW filters...

check the supplies for voltage and ripple and run the self diagnostic tests in the manual and see what you get for different tests

[charlyd]

hello how can help me out with the Part Number: 08595-90001 (Mar91. The 8595A is no longer sold by Keysight)

[analogRF]

you mean this?
https://literature.cdn.keysight.com/litweb/pdf/08595-90001.pdf?id=834233

why do you need this manual? yours is E model which is very different

[charlyd]

hello ah yes this is the one thanks analogRF.  ah ok.. is it really so different,
i came across the PDF tip somewhere on the keysight forum.  so now i understand it is not very usefull then.

So back to the 08590-90316.pdf, 8590DE-CLIP.pdf, 8590E-L-CLIP.pdf. that is already a lot but maybe there is something i missed. Is there more usefull documentation for the 8595E ?

i try to figure out if there is a way of simply testing my YIG. without have specific tools. i got the tip to order a cheap chinese SG which goes to arround 4GHz but with all the issues in china this can take up month to arrive.
i ordered a 5.6" LCD in china it will arrive in june somewhere they told me, the supplier live in the corona area. 

i mean is there something to compare. like the yig input/output ( with the Cal signal connected or so. )

for example,  what should i see on the input?
did anybody do some measurements in "the none working" situation and later in the "the working situation".
i found this.

[analogRF]

as I mentioned multiple times before, your YTO is fine. It is working otherwise you would not see a stable signal at ANY frequency at all
Also you would get unlock errors. Just tune the analyzer to different frequencies with span < 2MHz and if you dont get unlock error, your YTO is working which I am sure it is. Dont need to feed any signal.

your SYTF is probably not working and the only way to check is to have a high freq signal generator, there is no other way around it. Even if it is the switch that is bad not the YIG tuned filter, still you need a signal gen to check out the output of the device and by t he way  the switch is embedded within the YIG tuned filter, so if it is bad the whole YTF must be replaced

if you have another SA that goes to 6.5GHz, one thing you can do is to use your YTO (of the 8595E) as signal generator and feed it to the SYTF of your 8595E and let it sweep and then look at the output of the SYTF on the other SA.

Your YTO covers 3.92-6.82GHz when you sweep from 0 to 2.9GHz (up conversion) and it covers 3.12-6.82GHz when you sweep from 2.8 to 6.5GHz (down conversion). 8595E does not use LO harmonics for mixing in the high band.
 

EDIT: actually if you have access to another SA, you can just simply use the LO output of that SA as a signal gen and feed it to your 8595E and check the high band. Dont need to disassemble any cables.

[analogRF]

have you tried all the diagnostic tests in the service manual? I think one of the tests checks the YTF driver and at least you can see if the driver is ok

[MarkL]

I noticed that the CAL output is not looking clean.  Perhaps I've not seen enough CAL outputs on this series, but I've never seen one with a spur in the middle.  See photo.

Maybe there's something wrong with the CAL output?  Is that spur around 800MHz?  Can we get a closer look at it?  What is the exact frequency?

I'm assuming you don't have another SA.  What model SG do you have?  If it has a flat output you could at least confirm most of the low band is ok.  You could use your SG to slowly sweep the range up to 2.1G and look for anomalies.  Turning on trace B and setting it to MAX HOLD can be useful to capture the envelope at the same time.  Use the same settings you used for the cal screen shot.

If the envelope is flat, maybe it's the CAL output and for some reason the harmonics are dropping off at the higher end.

Can you find a frequency on the SG that causes a spur at ~800MHz?

If you reduce the RBW to lower the noise floor, can you see the CAL signal in the upper band?  It looks like there's a tiny peak there.

As others have said, a signal source that covers the high band is really needed.  Among the other suggestions, you could also get a passive frequency doubler (or other multiplier).  They're relatively inexpensive.  They need a fair amount of drive input from your SG, but they have a lot of good harmonics beyond double that could be useful for at least a go/no-go look at the high band.  Minicircuits.com has a wide range of options (https://www.minicircuits.com/WebStore/Multipliers.html).

[analogRF]

the CAL output is normal. It does look like that and does have a lot of harmonics

if you have access to another SA, as I explained, you can use th LO output of that SA as signal gen and crudely check  your 8595E. it cannot show you how accurate the high band is because LO output is usually not that flat over frequency but at least you can check if there are any holes in freq response

[MarkL]

the CAL output is normal. It does look like that and does have a lot of harmonics
...

Ok, thanks for the confirmation.  Disregard my post.

[analogRF]

oh, sorry, I just noticed you were talking about that extra spur just below 900MHz. That's not normal, it should not be there. Not sure if it's a byproduct of the SA. It could be the leakage coming from another device in the lab that is operating in that frequency (around 840-850MHz maybe?)
Using high quality coax cable and connectors should remove or reduce it if that's the case.

one thing that is wrong with that picture is that harmonics should be quite visibly larger even after 2.9GHz but they have disappeared
even below 2.9GHz on my units I see larger harmonic amplitudes. In that picture they fall off very quickly
it could be the frequ response calibration (flatness) that is wrong or lost or it could be the hardware....
having a signal gen and checking the output of each stage one by one would quickly show the problem

[MarkL]

Ok, so maybe the CAL output *does* need a second look.

Unless I missed a clue to the contrary, my point was that all the observations and attempted cal runs done by the OP have assumed the CAL output is ok.

I'm challenging that assumption because it's looking a little fishy to me with that spur in there.  It's also possible the CAL output is fine, and the spur is a clue to some other problem.  But my goal was to devise some way to show what level of confidence to have in the CAL output with the OPs limited equipment.

The second third harmonic also looks a little asymmetric, but it's really hard to tell at that span.  There could be additional garbage being obscured.

It's possible to use the SG to provide the 10Mhz reference to the SA and move the reference frequency around a little.  If that spur is an external influence, it should move around too.

Looking at the full span with corrections off might also be interesting.

Trying to think of what would be useful with the constraints of the equipment on-hand.  Otherwise, with the right equipment, I agree this would take about two seconds.

EDIT: Meant third harmonic.

[analogRF]

Unless I missed a clue to the contrary, my point was that all the observations and attempted cal runs done by the OP have assumed the CAL output is ok.

I'm challenging that assumption because it's looking a little fishy to me with that spur in there.  It's also possible the CAL output is fine, and the spur is a clue to some other problem.  But my goal was to devise some way to show what level of confidence to have in the CAL output with the OPs limited equipment.

The second third harmonic also looks a little asymmetric, but it's really hard to tell at that span.  There could be additional garbage being obscured.

calibration of BW filters is done at CF=300MHz and the cal signal is fine there. So that should not be a problem.
but the harmonics fall off too quickly and disappear in the high band which they shouldn't. Those two symptoms could be indication
of wrong flatness calibration, bad SYTF (hence no high band or very weak high band), or even mixer problems...

I still think that spur is from the outside and maybe it even switches on and off and we only got a still picture of when it was on
it is close to uplink and downlink of some GSM and LTE bands. I cannot think of any scenario that would generate that single spur frequency
in this instrument considering all the signal generation schemes that are used in 8595E... still I might be wrong and overlooking something...