my Data General Nova 4/X, disk woes continue

From: "Tom Jennings" <tomj at wps.com> I've been working on my car project (http://wps.com/AMC/1970-AMC-Hornet/index.html) because car season is coming up soon enough, but I peeled time out for the Nova here and there. I originally repaired a popped +5V logic bypass cap on the Read/Write Board, (some monolithic ceramic job), and changed all the funky old caps for new ceramic disks. That worked for 20 - 40 hours of operation without a single error. Disk had formatted with zero media defects. Then the disk started throwing read errors that moved around (bad blocks later good; more and more bad blocks...). Diagnostics never indicated any error other than read errors; seek, format, diags all run 100% perfect except read errors. I suspect the Read/Write Board. So I ordered modern monolithic replacements for the remaining old mono chips from Digikey, and shotgunned all the old electrolytics etc for good measure, in two places electrolytics replaced with tantalums (didn't have .82uF's). Install board with all new chips and alcohol wash, format -- hard 100% read failure! D'OH! Long story short, I found internal +6V was not; turns out I installed a tant backwards. doh. New tant, +6V fine. Formats, reads OK, then errors again! Sheesh! System has been on for two hours (I'm getting paranoid, so temp stabilize the whole thing); halted diags, loaded disk heads, installed drive cover, pushed into the rack, left for an hour for the drive temp to stabilize (this is a religious rite, might as well kill a chicken) then format and test again. After warm-up, it's worse. Signals all through the path look OK, hot and cold. I can see no difference, but this is 10 MHz NRZ data, in a two-state amp with AGC. I think I'm screwed. The R/W board is a piece of work too. I suspect it's an early revision, as the board is a hack job, covered in cut traces, components and jumpers on the bottom, and there's a !#$%!! trimpot crammed in somewhere as well as kludgey RC (lopass) networks in some digital logic. Plus, the PC board was milled out for access to a middle layer. Not Good. Plus, it doesn't match the schematic (though it's close enough to poke around in with a scope). Photos of the nasty board here, http://wps.com/NOVA4/pitchas.html About halfway down. Anyone got a spare board?!