I recently traded some gear for an HP8921A service monitor. It had a problem which I narrowed down to the high stability reference board. I ordered a replacement board and that fixed the problem. The 8921A now works great. However that still leaves me wanting to fix the original board. HP has never released schematics
for the 8920/8921 series of service monitors which is a shame. They have released board level repair information only.
Enough information was provided to power up the board on the bench. It has a TCXO running at 10 MHz with a 1 GHz VCO that is phase locked and also divided by two to output a phase locked 500 MHz. The main problem is that the 1 GHz output is noisy, the 500 MHz is also less then perfect but doesn't look as bad after going through its divide by two circuitry. I have gone through the amplifier stages and everything seems to look good. It outputs the proper levels. The power supply to the oscillator itself looks good as does the control voltage. I can see it changing to adjust frequency and it looks pretty clean with no noise on it. The main VCO transistor is an MRF581. I have output scans from both the good and bad modules - you can see the noise on the bad one and a slight variation in its output frequency. The various traces are min/max/average and instant.
I have also included a pic of the board. It is pretty easy to see the flow of RF through the left hand side of the board starting at the top.
I have checked all resistors, power supplies etc, all look fine. If anyone has any thoughts let me know. I already swapped the MRF581 as I happened to have one and it made no difference.