core stacks

From: Eric Smith <eric_at_brouhaha.com>
Date: Mon Apr 19 19:37:10 1999

Jay wrote:
> ISTR that if you had a core set for an 8 that was "mismatched", the hardware
> handbook had the procedure for using a scope to "tune" a set into the
> appropriate range.

Chuck wrote:
> Not exactly, this does allow you to "fine tune" a set that already is close
> enough, but it doesn't necessarily get you close enough if you weren't
> there already. There is a graph in of the hysterisis(sp?) region in the 8/a
> book and tuning consists of making sure that the write current and inhibit
> current cancel, and that write current from both the x and y drivers will
> just flip the bit but not risk flipping other bits.

So you're saying that whenever Field Circus had to replace one of the boards
of a set, they replaced all of them, because there wasn't enough adjustment
range to calibrate them? That doesn't match my recollection; I think it was
commonplace for the H220 to get replaced by itself if it got damaged, and I
think I recall seeing them replace at least one failed G104.

The biggest problem is that the diagnostics can't stress the memory system
as much as data-break peripherals can. Tuning the adjustments for reliable
operation is more difficult than the maintenance manual might lead one to
believe.
Received on Mon Apr 19 1999 - 19:37:10 BST