Well, I replaced the acquisition SRAM and it fixed most of the problems I was having at the beginning of this thread. Now the scope accurately shows my input signals over the entire sampling interval. The random stair-stepping issue is also gone, no idea what was causing it. As far as I can tell, the scope is functional.
Unfortunately, it still doesn't pass the diagnostics self test. The new error that appears (inconsistently) is:
diagAcq_peakDetectAcq Peak Detect mode is a special acquisition mode described in the technical reference as
"In Peak Detect mode the sampler holds the minimum and maximum values
over a sample interval. The data points are transferred point by point to be
digitized and stored in the acquisition RAM as they are acquired."
I have never used anything like it before, so I can't really say if it is working properly but it looks like it is. It would be nice to get rid of this error but I'm not sure it's worth the trouble. I do plan on replacing the Dallas RAM and RTC eventually, but loosing the existing calibration during the extraction process would turn this functioning scope into a brick; I don't have the equipment to recalibrate it.
P.S. The CPU errors were caused by a partially disconnected communications option board. It was unable to find the floppy drive which causes a save/recall memory error.