IBM 1620

[Tothwolf]

On Sat, 29 Aug 2015, Paul Koning wrote:
> On Aug 29, 2015, at 12:36 AM, Lyle Bickley <lbickley at bickleywest.com> wrote:
>
>> I was not on the Team that did the memory analysis and the ultimate 
>> "modern" replacement memory. However, when I joined the Team, I asked 
>> similar questions and was told that the core memory was literally 
>> "falling apart" and was not repairable.
>
> I sure have a hard time fathoming any of this.
>
> Quite apart from the obvious one of building a pin-compatible modern 
> memory, another answer comes to mind.  If a core plane has come apart, 
> the cores could be recovered and restrung.  That requires patience and 
> dexterity, but it should be doable.  It also requires a device to hold 
> the cores correctly for threading; something like that could easily be 
> made in a a 3d printer.

Those particular cores are quite small and I'm not sure a 3d printer would 
be able to print a jig with the tolerances required. IBM had special 
machines to position and thread them. 
http://ibm-1401.info/IBMCoreArraysIEEEMagnetics1969.pdf

I can certainly understand why repair wasn't initially attempted, however 
that doesn't mean it is impossible. Given the rarity of the system, 
welding new stubs of wire to the original enamel wire or completely 
rebuilding the core planes with all new wire might still well be a 
worthwhile project at some point in the future. It might even be possible 
to keep most of the cores in position and rethread just one portion (X, Y, 
sense/inhibit) at a time.

Another plus is that because of the way the wires are threaded, it is 
unlikely that any of the cores have been lost, even if a large number of 
wires have broken at the terminals.