HP3336B Synthesizer/Level Generator repair

[tkamiya]

I have this new-to-me HP3336B Synthesizer/Level Generator that needs TLC.

Overall, it works but in limited fashion.  Power supply issue was there and it has been addressed.  Voltages are all correct (within the accepted limit).  No physical damage.  All connections has been cleaned and deoxed.  There is no obviously bad parts.  No signs of physical damage.  For its age, it's remarkably clean.  I have the oddball WECO connector.

Things I already fixed and how:

+5V was 5.3V. 
Remote sensing wasn't going through connectors.  Cleaned and reinserted.

No output.
Replaced coax between PCB and attenuator package.

No output at certain attenuation.
Removed and cleaned latching relay contact in Attenuator package.

Other stuff I need help with:

It has odd PLL situation.  If frequency setting is xxx.x00.000.0, PLL is stable and signal is stable.
If any of zeros are NOT zero, severe jitter as if PLL is not locking.

Odd Attenuator function.  Seems to be acting up.  If output is larger than 33dB, levels are reasonable.  If less than 33dB, signal actually gets stronger and repeats the decline again.

I don't have the exact frequency where this happens as it seem to me random.  But in lower frequencies, wave shape is not sine.  Voltage crosses zero and goes up.  Then there is this upside down triangle dip, then follows rest of the sine curve. 

I'm going through service manual but somehow, it is not as comprehensive as other HP manuals I've seen.  Help is appreciated.

[rf+tech]

tkamiya,

Odd Attenuator function.  Seems to be acting up.  If output is larger than 33dB, levels are reasonable.  If less than 33dB, signal actually gets stronger and repeats the decline again.

This behavior points to latching relay K3 not pulling in the 40 dB pad. K1 and K2 switch in 10 dB and 20 dB pads, respectively.

RF+ Tech

[tkamiya]

Thank you.  I hear relays clicking but at strange intervals.  To start, I will have to clean contact points again and start from there.

[rf+tech]

While you have the relays off, verify resistance of both coils.

The attenuator switching pattern should be consistent, with a 3 dB "offset." Since the maximum output level is +7 dBm, the 10 dB/step switching points will occur at -3, -13, -23, -33, -43, -53, -63 dBm; following weighted binary coded decimal logic.

RF+ Tech