[vcf-midatlantic] Core Memory Purchase

Jonathan Gevaryahu jgevaryahu at gmail.com
Thu Dec 3 01:18:03 EST 2015


Here is a response about reviving old core memory boards, from Kevin 
'kevtris' Horton, who has revived 3 of them in the past.


-------- Forwarded Message --------
Subject: 	Re: Fwd: [vcf-midatlantic] Core Memory Purchase
Date: 	Thu, 3 Dec 2015 01:15:52 -0500
From: 	K Horton <kevtris at comcast.net>
To: 	Jonathan Gevaryahu <jgevaryahu at gmail.com>



On 12/2/2015 22:17 , Jonathan Gevaryahu wrote:
> kevtris, you have some experience resurrecting core memory. Do you 
> want to respond here?
>
>
> -------- Forwarded Message --------
> Subject: 	[vcf-midatlantic] Core Memory Purchase
> Date: 	Wed, 2 Dec 2015 16:56:35 -0800
> From: 	John Sully via vcf-midatlantic 
> <vcf-midatlantic at lists.vintagecomputerfederation.org>
> Reply-To: 	vcf-midatlantic 
> <vcf-midatlantic at lists.vintagecomputerfederation.org>
> To: 	vcf-midatlantic at lists.vintagecomputerfederation.org
> CC: 	John Sully <john at csquare.ca>
>
>
>
> I'm planning to resurrect some old core memory as a hobby project.  I was
> wondering if anyone with experience on getting this memory working could
> provide some insights on what I should look for.
>
> EBay seems to have a large number of expensive ones that don't appear to be
> in working order - along with some much cheaper (relatively speaking)
> russian made stacks.
>

yes, this is a big problem.  if the core memory is shown "open" on ebay, 
don't buy it.  there's a 99% chance it's going to be broken.   The cores 
and wiring are extremely fragile, just blowing on it can destroy it 
(well not THAT bad, but it's still pretty insanely fragile).

I have three fully working core memory boards:  two 16K*18 and one 
32K*18 boards.   I had to fix all three of them, since they had bad 
parts here and there, but it wasn't TOO hard.

All of them work similar; there's a hardware state machine that runs a 
little sequence for reads that do this:  read the word, then write the 
same data back (since reading core is destructive).

I successfully dumped the data that was sitting on all three of my core 
boards; the 16K boards were repaired and put in stock as spares 
apparently, and just had a 0101010101 pattern,  while the 32K board 
contained actual NASA code!  I dumped it but lost 256 words of data, 
unfortunately.   That board had some problems with reading/writing 
because there was a bad transistor in the X driver, and there was a bad 
TTL chip.

All of them had bad tantalum caps that I had to replace;  the 32K board 
kept randomly drawing enough power to trip the current limit on my power 
supply every few hours, so I finally found it and replaced the cap which 
fixed that.

The 32K board is extremely "advanced', compared to the 16K boards which 
are more traditional.  The 32K board only has a single X and single Y 
driver instead of a bunch of each.  All of them have a forest of 
transistor arrays, diodes, and pulse transformers to do selection 
because it saves a ton of hardware this way.

I have a dir on my website with all the stuff I did on it and if there's 
interest I can make a video showing off my rig and how I tested the 
memory.  I had schematics I drew for the boards but I think I might've 
lost them.

there's some PDFs in there that give esoteric details on the care and 
feeding of core memory.

http://blog.kevtris.org/blogfiles/corememory/







More information about the vcf-midatlantic mailing list