keithley 197 reading stopped updating. (& ghost gpib signals?) Suggestions?

[cnt]

I recently got a very cheap 197 and i "fixed" the switches with some deoxit and replaced the cut off power cord and all was well.  I left it on measuring 6v supply last night and after being powered up for 24-36 hours i noticed the display stopped updating.  changing the mode to ohms yielded one update then it stopped, same changing it back to volts.  A power cycle from the front panel switch seems to have fixed it, but I'm wondering what happened and how to troubleshoot if the problem comes back. 

I'm guessing it lost a clock somewhere, the manual mentions half a dozen clocks involved with the ADC.


UPDATE:  Now its regularly turning on "RMT" and going into cal mode with the "C" on.  And occasionally flashing "out", or "out" gets stuck on.  "out" is whats displayed when you try to save calibration without the switch in the back enabling cal mode.  Seems like I've got a ghost in my GPIB board.

[cnt]

Well looks like I found the problem, or at least _A_ problem.

the GPIB board's got its own transformer tap with a bridge rectifier, 2200uF cap, and a 7805.  I was probing voltages and all looked good, but I thought I'd take a look with the scope on the 5V rail.  I got this weird trace doubling about 100mv tall and near the cutoff frequency of my scope (100mhz), which I later traced to my probing and I think some EMF from my chinese florescent desk lamp light bulb.  But, while probing I accidentally coughed and bumped something and got this on the 5v rail:
 

I can't make it happen at will but it seems some combination of jiggling the filter cap and or breathing too heavily will trigger it. 

Probing it in circuit (its relay isolated when un powered) shows more than half the microfarads fell out at some point and I suspect there may be a crusty connection somewhere internal to the cap too.  Figures the ONE Richey cap is bad and the rest are nichicon.  I guess 35 years is a good run.

16v 2200uF cap these days are almost all much smaller apparently.  Searching by lead spacing and diameter and height and maximum expected life yielded a 25V 105C (up from 85) so hopefully this new cap will last another 35 years.

[trobbins]

Yeh, I traced a noisy measurement value to the filter cap on the GPIB accessory board - easy to identify by disconnecting that board.  Still got some last digit or two noise, but for the moment that's acceptable.