My 8566AB from page 22 is now finally fixed.
Front panel got epoxied (sometimes added cotton fibers to expoxy for more strength) back into shape after removing original paint and metal coating. Added some plywood to make it more stable. Don't know why HP originally didn't do it that way.
Then I painted it with red steel primer paint and finally paint brushed it in black.
The PLL unlock faults were somewhat funny, because they appeared mostly at the start of a sweep. If the analyzer was set to long sweep times with narrow span, the PLL would lock and the errors mostly disappear. So I started probing around and found voltage spikes on most power rails at the beginning of each sweep. The A10A1-8 15-30MHz LF PLL would unlock or lock on a wrong frequency after each such spike. Since the YIG coils are tuned before each sweep and draw a lot of current doing so, I guessed that some of those silvery tantalum caps on PLL boards and/or YIG tuning boards were killed by ripple caused by the failed big blue 5V electrolytic cap, which I replaced previously.
Thereby I replaced most of those tantals on the YIG driver boards, A16 sweep generator board and some of those on the A10A1-8 15-30MHz LF PLL boards. As replacement I mostly used Wima PET foil caps (they last forever, but are only available up to 10-15uF and are quite large) and electrolytics (on places with less space or larger capacitance). After that, I adjusted the sweep generator pots on A16 board and the PLL unlock faults were gone.
Finally tested it with my HP 8684B cavity tuned oscillator. Frequency wise is not very stable (no PLL lock), so I couln't test the narrower spans, because the signal would drift away.
11GHz @ -45dBm indicated on 8684B:
Guess that the -52dBm on 8566AB is partly due to cable losses. Maybe the YIG preselector/filter in harmonic mixer assembly is also slightly misaligned. I also don't know how accurate the 8684B amplitude really is.
Second harmonic of 10GHz/+15dBm signal from 8684B (cavity oscillators produce quite clean signal, since high Q, so you only get very small harmonics):
Now I need to get a Gigatronics 1026 or something similar (HP 8340/8341 is usually much too expensive for me) along with power meter to verify the amplitude on all ranges.