HP 8903A repair help please!

[bsgd]

Hi all,

So I've been trying to wrap my head around how to fix my beloved 8903A but no luck so far, and now I need help from the experts.

The problem is that the measured distortion is higher than what it should be, and keeps jumping all over from 0.02% up to 0.09% most of the time. I have borrowed a working 8903A so I managed to replace cards and now I know the problem lies in the A3 Notch Filter card. Replacing it with the one from the working 8903A fixes the issue.

Since the problem is a somewhat slight increase in distortion, following the service manual troubleshooting did not help much since all measurements seem to be within spec, so I started 'guessing' what might be wrong and trying to replace a few capacitors etc, but nothing fixed it.

I would like some guidance whether anyone came across such problem before or where to start looking. The A3 board is divided into what seems to be 2 'separate' sections, one for coarse tuning and another for fine tuning, and I still havent come to a conclusion on which part is most likely to be causing the issue. I have 'compared' measurements from the working A3 card with the non working A3 card but found nothing really suspect so far.

It also seems that the problem is not frequency dependant, as distortion still remains jumping around between 0.02% and 0.09% whether I dial in 100Hz, 1kHz or 100kHz.

Here is a pic from the temporary bench setup I have for comparing both.

Thanks in advance.

[cncjerry]

Depending on the voltage levels on that card, you can rig up a probe and feed it into a PC audio analyzer.  All you need is an fft and when you see the harmonics jump up, that is your stage.  I don't think you would see that level on your average scope FFF. Watch out for large DC levels.  I don't remember, I have one, but isn't that card mostly for the weighting? 

There shouldn't be too many things that could cause this.  Bias, offset adjustments would be fixed but resistors change value over time.  Interstage coupling would be a little obvious as there would be clipping, I would think, or approaching it.  Bad caps in the coupling would be a starting point.  If the voltages are off you would have a bias problem.  If you can't rig up a problem check the DC voltages accurately in the signal path with the same signal fed into the good one and the bad one.  That would help eliminate bias issues. At that level it could be a darlington pair that went off.

I love mine too.  I bought the matching signal generator, 8904?  I also have the spectrum analyzer, they call it a signal analyzer, and that is handy for finding problems like this as it has it's own source.

[macboy]

Have you looked at the monitor output with a scope? This is the distortion after the notch filter (and other filters is enabled). You should trigger the scope off the input signal to get stable triggering. Comparing the monitor output of the two units may be enlightening.

[bsgd]

Depending on the voltage levels on that card, you can rig up a probe and feed it into a PC audio analyzer.  All you need is an fft and when you see the harmonics jump up, that is your stage.  I don't think you would see that level on your average scope FFF. Watch out for large DC levels.  I don't remember, I have one, but isn't that card mostly for the weighting? 

There shouldn't be too many things that could cause this.  Bias, offset adjustments would be fixed but resistors change value over time.  Interstage coupling would be a little obvious as there would be clipping, I would think, or approaching it.  Bad caps in the coupling would be a starting point.  If the voltages are off you would have a bias problem.  If you can't rig up a problem check the DC voltages accurately in the signal path with the same signal fed into the good one and the bad one.  That would help eliminate bias issues. At that level it could be a darlington pair that went off.

I love mine too.  I bought the matching signal generator, 8904?  I also have the spectrum analyzer, they call it a signal analyzer, and that is handy for finding problems like this as it has it's own source.

I have tested all resistor in circuit comparing resistance with the known good A3 board and all is almost exactly equal, little variation of less than 5%. I also have recapped the board completely today and even replaced a few crital opamps. No change

I will have to get some extender boards to compare voltages.

Have you looked at the monitor output with a scope? This is the distortion after the notch filter (and other filters is enabled). You should trigger the scope off the input signal to get stable triggering. Comparing the monitor output of the two units may be enlightening.

I didint do that, good idea. Since residual distortion is only around 0.05% would that be visible with a scope?

[George Edmonds]

HI
With a typical modern digital sampling scope you will be lucky to see 10% distortion, certainly not 0.05%


G Edmonds

[macboy]

Have you looked at the monitor output with a scope? This is the distortion after the notch filter (and other filters is enabled). You should trigger the scope off the input signal to get stable triggering. Comparing the monitor output of the two units may be enlightening.

I didint do that, good idea. Since residual distortion is only around 0.05% would that be visible with a scope?

Yes, the monitor output is the output of the notch filter, which is also amplified depending on its level. This is effectively the noise+distortion that the instrument is measuring and comparing to the level of the fundamental to calculate the % distortion measurement. You will be able to see the nature of the distortion, such as whether it is mostly noise, or crossover distortion, or 2nd harmonic, etc.  You also need to monitor this output to perform at least one important adjustment (notch balance or something... It's been a long time).

[cncjerry]

Dumb question, does the distortion go up when you put your card in the other 8903? I just looked around on ebay and see a number of cards for an 8903.  I would never give up but you could have a fallback.

Jerry

[cncjerry]

if the distortion is jumping around, it could be a cracked trace.  Have you tried flexing the board a little while in the box?  Try doing resistance checks from point to point instead of components. Maybe run over all the solder joints.  I've had problems like this.  My brother has a printer he should have pitched a long time ago that suffers from flex problems.

[bsgd]

Hi all,

I tried all suggested and in the end, it was actually a cracked resistor. I tried flexing the board but no luck, so I decided to look with a magnifier for any visible problems on the traces and ended up noticing a suspect resistor that had a small crack. Not sure what happened and maybe it was this way from the beginning but only now developed an electrical problem.

Thank you very much for all the input, what a great forum!