I took my chances on a locally available 8591E. First power-on yielded an error message about missing calibration data, but I could get it working again thanks to MatL's advices over here:
https://www.eevblog.com/forum/repair/repairing-an-hp-8593e-spectrum-analyzer/msg990688/#msg990688Somehow expected, the instrument lost these settings immediately after shutdown, pointing to a depleted RAM Backup battery, supported by a sticker that noted last replacement to have happened in 2006, besides a 2nd sticker stating a preceding exchange in 1994. A quick check of the battery voltage gave a reading of 1.06V so a replacement Li-SOCl2 battery was ordered.
That battery arrived today, MFD date in 2024, 3.67V ocv. Soldered it in, watched the polarity, made a voltage reading on the board contact area, again 3.67V. All fine, I guessed. Put everything back together, ran calibration, set time and date etc. Had fun for an hour, switched it off for lunch to return an hour later. Switched on again: all settings gone (again). A brief moment immediately after power-on, a single message "EMPTY DLP MEMORY" can be seen before the instruments resumes operation with the usual complain about missing calibration data.
Further test show that the data is lost even when power off and on ist done within one second after the other.
So my question: is the RAM known to fail? I just replaced the SRAM in one of my E3631A so I know these can go bad indeed. But have there been similar cases before?
The service manual only offeres block diagrams, no schematics, so I need to remove everything above the memory section first and trace the battery voltage supply to the RAM chip to identify it and verify that there is nothing broken in between that hinders those 3.67V to reach the RAM. I could already see by using a small flashlight that my board is the newer revision with only two EPROMs instead of four.
Any help is welcome.
Christian