HP 8903A Oscillator Board problem

[Agosto]

I posted before about ongoing issues I am having with my newly acquired 8903A: After sorting out relay issues on the Input Board I tested the basic functions of the unit. Selecting 1000Hz, 3V AC I connected High Output with High Input and measured distortion. "60-72dB" at 1000Hz, getting progressively worse down to 20Hz with 35dB (readings not stable). Looking at the waveform at the High Output with a scope I noticed a distortion at the zero point of the sine wave fading away at around 500Hz. Looking around the Oscillator Board I re-soldered a portion of the board to no avail. Using an external signal source (1000Hz, 3V AC) I am getting 76dB on distortion measurement. So I am quite sure the issue is somewhere around the Oscillator Board (does 76dB sound ok in terms of performance for a $500 signal generator, GwInstek??).

Meanwhile, I got my hands on another Oscillator Board and used that instead: Upon initial start-up the distortion level improved to spec (around 87dB) but after around 2min and some faint (REED?) relay clicks distortion suddenly changed to the original board's level (60-72dB). So I am afraid I damaged the new board. Supply voltages are all within spec.

What puzzles me now is that the unit worked fine with the new Oscillator Board, although only for a short time. Am I wrong in thinking that the problem may originate somewhere else?
I am not experienced in servicing HP equipment so I hope someone here on the Forum can point me in the right direction. Thanks!!

[wn1fju]

After the two minute warm-up and things go astray, does the distortion measurement change when using an external generator?  In other words, can you isolate the problem to the generation side or the measurement side.

-76 dBc is a bit optimistic, in my opinion, for cheap signal generators.  Most of mine are in the low -60's, but occasionally you will find an oscillator that uses a Wien bridge circuit and those can easily top -80 dBc distortion.  If you have a function generator, you might see what the 8903 measures on a 1 kHz square wave and triangle wave.  You should get approximately -7 and -18 dBc respectively.

Also, when you say the power supply voltages are in spec, was this after the thing started to misbehave?  Have you looked at the power rails with a scope to make sure there is not excessive ripple?  Sometimes, measuring DC voltages with a voltmeter is not enough since it won't respond to AC ripple and just gives you an average amount.

I would be surprised if you damaged the other oscillator board.  HP's designs were usually pretty robust. 

There is nothing "special" about servicing HP equipment of the 1980's era.   Unless, of course, you understand that HP had superior engineers, produced superior equipment and published superior service manuals.

[Agosto]

Thank you for your feedback - much appreciated! To your questions:

1. Yes, using an external signal generator (the GwInstek I was referring to in my post) distortion measurement improves from 60-72dB/1kHz to -76dB. at 20Hz, improves from approx. -35dB to -76dB. So I concluded the root cause is the internal oscillator.
2. I checked the P/S voltages with a scope: No excessive ripple. The +15V rail is a bit low (14.7V) but as far I can read from the service manual that is still within spec.

For clarification: The new board showed excellent distortion numbers after my first use of the board. After the values went out of spec, measurements are consistent. There is no more spec ratings prior to a "warm-up phase". The bad readings are now consistent, no matter if cold or warmed up.

The change from in-spec values to the lower values happened instantly, so no gradual degradation.

However: (!)
Today, after 2 days off, I started the unit to measure square (Result: -7dB) and triangle (Result: -18dB) distortion rates. To my surprise, sine measurement at 1kHz/3V showed -83dB distortion again! So something is not stable...
I did a number of measurements  at different frequencies, with internal oscillator and external signal generators at 3V. Here are the results:

             Frequency     Int. Oscillator     Ext. Generator

                  1kHz              83.5                 75
                  500Hz            83                    75
                  250                79                    75
                  100                71.5                 75
                    20                58                    75

                  10kHz            83                     75
                  50kHz            88                     78
                  100kHz          89                     78

Although for unknown reasons the measurements are now better than what I originally reported, I believe that the results below 1kHz e.g. -58dB at 20Hz are still outside of spec. Correct?

Should I check supply voltages again?

[wn1fju]

For reference, I set my 8903A with the high output connected to the high input, no termination, 3V amplitude and dialed in your frequencies.  My unit defaults on power-up with the 80 kHz LP filter enabled, so that's how I left it.  Both front panel float switches were set to the ground position.

I got between -90 and -95 dBc at 20, 100, 250, 500, 1K, 10K and 50K frequencies.  With the 100 kHz frequency, I turned off the 80 kHz LPF filter and got -76.6 dBc.  With the LP filter off, the readings at the lower frequencies were all about 2 - 4 dB worse.

And while I'm thinking of it, you might want to reseat all the boards and give all the pots a jiggle (after you have marked their positions).  Perhaps a pot is dirty and needs a little jostling.  Also, give all the boards a good rap and see if anything on the display changes a lot.  The notch filter tune and balance adjustment is particularly sensitive in my opinion.  They use a single-turn pot and you had better have a very steady hand!  In fact, on my unit, I actually replaced that pot with a ten-turn variety so I could really dial in the lowest distortion.

[Agosto]

Thanks! Meanwhile, I was trying to do exactly what you described by wiggling the boards around and reseated all boards. Nothing - no value changed. But: The moment I switched the High Output float switch to float the 1kHz value came down again to 70dB. Any attempt to bring it back to 84dB by switching again, wiggling the switch or even shorting the switch on the back was not successful.
I will try the pots as you suggested. Other than that I am at a lost (at the moment). Thanks again for your support!

[Agosto]

Addendum: Jiggled all the pots - no change. What I can see is that the measured distortion is getting progressively worse the lower the input frequency is.

[wn1fju]

Well, good luck.  I don't have much to add at this point.  I suppose you could work your way through the procedures described in the service sheets of the manual.  Tedious at best. 

And here's one more "grasping at straws" suggestion.  The fact that there is a frequency dependence to what you are seeing might indicate a problem with a capacitor that finally failed somewhere in the signal path.

[Agosto]

Yes indeed, a failed capacitor is an option I am going to pursue. I think I can further narrow it down now that I replaced the oscillator board and the input board. Both replacements did not do anything for improvement.
On the other hand: The frequency dependency is only there when using the internal oscillator as source. So I am still convinced the issue is on the internal source side. But where...?
Thanks for your feedback.

[Agosto]

I finally resolved all issues with my new 8903A: For the last remaining item (high distortion readings) I had to replace 3 caps on the oscillator board. After going through the "Adjustments" chapter of the service manual, all readings are now within spec.
I previously tested the above caps with an LCE meter and they came out fine. Not sure how this works...

Thank you all for your support!

[DeepLink]

Out of curiosity - which 3 capacitors did you end up changing?

[Agosto]

C9, C20, C45