28 Feb
2014
28 Feb
'14
9:19 a.m.
False alarm. I cleaned the connectors and plugged
things back together. I've got 100% core working
again. I'll have some spare driver chips anyway.
Earlier I had a problem where 2 of the banks
had a bit 11 stuck high. I'd thought is was a wire
wrap problem but removing the write and read boards
and the data bits were all connected when checking with
an ohm meter across connectors. Again, cleaning the
connectors seemed to fix it. I hope the problem doesn't
come back.
The first bank worked fine, just the other two were
not working on the same bit?
Maybe just wiggling things by pulling boards in and out
got the wire wrap connect to work again?
Maybe both had the same bit needing the connector
cleaned ( somehow, a 1 in 20 probability ).
I'll have to keep a watch on it.
Dwight
----------------------------------------
From: dkelvey at hotmail.com
To: cctalk at classiccmp.org
Subject: Core problems
Date: Fri, 28 Feb 2014 06:43:36 -0800
I've been bringing my Nicolet 1080 back on line,
gradually. I'd written a simple RAM test. No address
uniqueness, just a simple read read complement write read
complement write sequence.
I found a dead stretch of memory on one of my 3 core
banks.
I figured it might be an address driver or a broken wire
in the core.
I checked my documents and wouldn't you know it, the wire
list didn't include that part of the cores. Only up to
the driver board inputs ( typical wire wrap back plane ).
I still need to determine if it is a core or driver problem.
I can swap things around to determine that though.
I was looking at the schematics and found that the drivers
were SN75324N. I'd checked on ebay and sure enough, they
had them. With shipping they're little over $4 each,
for 4, but available ( note that the plastic and ceramic packages
have different pinouts ).
I suspect the core but that is a project for today.
Just getting the machine back to taking instructions and
running code is fun.
Dwight