EEVblog Electronics Community Forum

Electronics => Repair => Topic started by: rg58 on December 05, 2018, 06:49:47 am

Title: HP859xL Series Spectrum Analyzer - FM coil fault investigation
Post by: rg58 on December 05, 2018, 06:49:47 am
I'm in the process of repairing a HP859xL series spectrum analyzer with various issues. It is deaf and I already found the dual mixer (1st converter) to be faulty  (attenuation up to 60dB) as well as possibly the 2nd converter as well. But this is not my biggest concern right now.

When running the diagnosis to verify if the FM coil driver is working correctly below a SPAN of 10MHz I am getting a flat line instead of a ramp - test FAILED. Above 10MHz SPAN it passes. So I digged around and even went as far as taking out the huge analog board on the side to investigate the FM driver circuitry itself (as I don't have schematics). The FM coil which meassures 1.3 Ohm resistance in my case (=normal) is being driven by a push/pull transistor pair, in series with 2 parallel resistors (=190 Ohm). So I soldered wires across them and looked at the signals on the scope. The output looks fine to me. One can clearly see the ramp (below SPAN 10MHz) of 20ms which corresponds perfectly with the sweep time of 20ms as shown on the bottom of the screen (see pics).

Now I wonder, if the FM coil is being driven correctly and the coil resistance is normal, why do I get this error? It looks like the coil should be working. Is there a problem in the sensing, maybe?

ISSUE 2 that could be related: Under 10MHz of span, when turning the knob changing frequency, the display (peaks) become jittery and very hard to move in either direction. I guess some PLL is trying to lock it? I never got the analyzer to "lock" and all the self-cals still fail until I have fixed the mixer(s). The display continously shows (Phase) Look Off, sometimes SRQ 110 (which I believe means Service ReQuest).

ISSUE 3 that could be related: When I tune to 0Hz center frequency I see the large LO peak. However, it is off by a varying frequency, usually between -50MHz and -12MHz, same way as the frequencies from my signal sources are off. Why is that? Because the FM coil (for fine tuning) isn't working as it should be?

The service manual also pointed to the Counter Lock (A25) as a possible cause of error. So I went ahead and disassembled this unit just now. Nothing obvious to see here, but there is a trimming capacitor which could need adjustment?

I am stuck and don't know what to look for next. On this A25 board I can see TP3 and TP300 test points and an unsed J8, but I have no idea what to measure there. I have no CLIP for this unit 00591-60090 (or 08592-60108?).

What could be causing these issues? Where should I look next? Any ideas?

Title: Re: HP859xL Series Spectrum Analyzer - FM coil fault investigation
Post by: pbarton on December 05, 2018, 08:01:33 am
There are some VOLTAGE REFERENCES on the 'huge analog board' that produce +10VF and -10VF which is used by the YTO TUNE DACS AND RF ATTENUATOR DRIVER chipset on the same board.
You should verify that these voltages are present.

EDIT: The Analogue Interface drawings can be found at page 265 or 271 (of 340) at (
The COUNTERLOCK drawings can be found at page 319 (of 340).

EDIT: Also, check the three legged tantalum capacitors for shorts, on the analog board. (
Title: Re: HP859xL Series Spectrum Analyzer - FM coil fault investigation
Post by: rg58 on December 05, 2018, 08:25:04 pm
Thanks a lot!!! This document is VERY helpful indeed! It contains all the schematics I was looking for all the time. Why I couldn't find this myself? Probably looked up the wrong keywords in Google and these are scans so Google couldn't index their contents (what they do in a 'normal' PDF).  At least I got schematics now and can see how the FM driving circuit works (confirming my hand-drawn schematics of the output stage).

Will try to meassure the +/-10V somewhere in the unit today and also if there's any ripple on it. But the self-diagnosis on both voltages passed as I got a flat line on top and bottom of the screen, so somehow I think it is not related to this and the fault is somewhere else.
Title: Re: HP859xL Series Spectrum Analyzer - FM coil fault investigation
Post by: rg58 on December 06, 2018, 01:56:31 am
Could somebody please explain how this sampler in A25 works? Is it some kind of mixer? I have no clue how they get from 3-6.8GHz input + 277-298MHz down to 60-100MHz IF. You may refer to page 321 in above linked document (CLIP 5963-2951).

The problem seems to be somewhere here, but I have no equipment to measure over 3GHz, so I don't know if the input at J1 from the coupler is correct. The input from the sampling osc seems fine, although a bit low maybe?

Reasoning: I went backwards from the DISCRIM signal (that goes into the YTO Drivers section) and only found "noise" on there. In part (D) STABILIZER on the same page 321 we see the 7.5MHz from the PLL coming in which is OK. But the other input is supposed to be 8.25MHz +/-0.65MHz and I got nothing useful there at all. The amplitude of roughly 1Vpp is correct, but the signal is just another kind of 'noise'. This signal is also called 8MHz_IF_A coming from part (K) DIVIDE BY 10 AND 16 where I get the same weird noise on 8MHZ_IF_A and 8MHZ_IF_B (only inverted signal). Aren't we expecting a digital signal here when SPAN is 0Hz? They divide this 60-100MHz signal by 10 using digital parts (Flip-Flops), but they still call it 'IF'? I'm a little confused. Also, I get nothing out on FS1F/16 but maybe it's just missing some enable signal (COUNT_GATE?). And going further backwards, passing 2 low noise amplifiers, we end up at part (H) SAMPLER and part (G) SAMPLER MATCH.

So how does this sampler work? If it would have failed, and I get nothing in, the 2 amps just amplify noise that is being digitized at stage (K) Divide By 10 and that's the weird (bit) noise I see on the 8MHZ_IF_A input of the stabilizer?

P.S. The +/-10V look just fine, very precise -9.988V and +9.994V with noise below 10mV.
Title: Re: HP859xL Series Spectrum Analyzer - FM coil fault investigation
Post by: pbarton on December 06, 2018, 03:03:11 am
You state "You may refer to page 399 in above linked document (CLIP 5963-2951)."
I think you may be referring to .pdf page 320 in above linked document (CLIP 5963-2951), at least that's where your A25 attachment has come from?

I assume that you have a copy of "8590 Series Analyzers Assembly-Level Repair Service Guide" at (
A description of the Counter Lock and the A25A1 Sampler is contained within Chapter 9. See .pdf pages 361 and 362. But I guess that you already have this?
Title: Re: HP859xL Series Spectrum Analyzer - FM coil fault investigation
Post by: rg58 on December 06, 2018, 03:15:11 am
Sorry, in this document above it's on page 321 in fact, my fault. I looked up a few more PDFs today and found a similar document with 418 pages that I looked at. I edited my post above not to confuse people later. Yes, got that one already and will have a read in Chapter 9. Thanks!
Title: Re: HP859xL Series Spectrum Analyzer - FM coil fault investigation
Post by: rg58 on December 11, 2018, 10:13:41 pm
I continued my troubleshooting on this analyzer (my time is limited these days):

As the manual suggests, the mixing is done by using harmonics generated by the sampling oscillator. However, it is still not clear to me where exactly the selection/filtering of the required (Nth) harmonic takes place.

I found out that the 277-298 MHz signal from this sampling oscillator ([E] on A25 counter lock) has a low amplitude until a switching voltage (SAMPSW from [C] DIG INTERFACE) is applied. Then this signal goes up by approx. 30-50dB, reaching over 0dBm as far as I remember (should be +15dBm). However, I never saw that voltage being turned on, even when making several changes to the analyzer (frequency,span etc). It's most likely being switched by software/the CPU, but I haven't looked into that one yet. I switched it on manually by bridging a transistor C to E for testing. The frequency offset on the screen didn't go away from that unfortunately and was still about -50MHz for the LO peak.

Furthermore, on the DISCRIM output of the STABILIZER section on the A25 counterlock I see below 10MHz SPAN a waveform in the negative territory of around 5V peak as attached, but travelling BACKWARDS into this output! I concluded this because there is only noise on all inputs of this OpAmp. Must be the feedback of the opamp that follows. The timing is also strange as this signal has a periodicity of ~75ms whereas the analyzer was set to 20ms sweep time. Strange I thought...

So I still believe something is wrong with taking the LO frequency sample above from the sampler A25A1 that I cannot check without equipment reaching over 3GHz. After that and the 2 low noise amplifiers I only see random (bit) noise. I suspect either the sampler H (A25A1) or one of the low noise amps (Sampler IF1 & 2) or the Directional Coupler A3A10 itself has failed. Or maybe the YTO is outputting too low signal? But it produces enough output signal to let the analyzer work in the HIGH BAND it seems.

How could I narrow down this fault to the actual unit without being able to check over 3GHz?

Any ideas and suggestions what else I could try? I guess I need some input now from you guys to make this monster work again.

Thanks a lot in advance!

P.S. Interesting fact is that completely disconnecting and removing the A25 counter lock unit doesn't seem to make ANY difference. The analyzer seems to work just the same way, although a reference is missing!? The 10MHz oscillator is also disconnected. Maybe it sends 'guesswork' signals to the YTO in this case?