malpass@LL-VLSI.ARPA (Don Malpass) (12/14/87)
My problem has been fixed. There were two causes, both self-inflicted. But since one part of the diagnosis may be useful in the future to someone, I'll post the saga. 1. The problem with the 8" drives was NOT a problem, and simply misled me. I have a toggle switch on the back of my 8" drive cabinet to switch between single- and double-sided disks (since some of the former are still kicking around from my old CP/M days and I live under the delusion that "someday, sometime, when I least expect it" I'll want to read one of them). I must have snagged the switch with my color monitor cable when I recently lent the monitor to someone for a demo. So that one was easy. 2. The 5" problem turned out to be the precomp adjustment, as one of you out there suggested. I must have turned it with my thumb while changing controller cards last week. That is VERY easy to do, and I think I'll glue a block of plastic or something to the edge of the card to guard the pots from careless fingers - which I seem to have lately. The important clue is that a freshly formatted 5" disk would check fine (with Norton's DISKTEST) up to around cluster 203 and then be no good. The cluster number was obviously different if the disk was formatted with the 8-sector or the single-sided option. But the book says that precomp changes at track 23, and sure enough, if I counted "thumps" as the head indexed, all went well for 23 tracks, and then went bad, coincident with the diagnostic printing out "203". So it was obviously precomp. The hardest part of the fix was finding where I had put the little test board Heath had me use to calibrate the board when I first built it. But all is well now. Further, in skimming through the manuals, I found that the 8" MFE drive I use would really like to be calibrated at a higher value of precomp than 120 uSec, so I should get improved reliability on disks I format from here on, although I've never detected any problems in the past. Thanks for all those cards and letters. don [malpass@LL-vlsi.arpa]