Disasters and Recovery

From: Eric Smith <eric_at_brouhaha.com>
Date: Sun Jan 17 20:23:42 1999

Tony wrote:
> I doubt if any semiconductor memory, even mask-programmed ROMs, will last
> 100 years.

Masked ROMs, particularly older (lower-density) ones, might have a reasonable
chance. Most, but not all, of the electromigration occurs when the part is
under bias.

Note that the old nichrome fusible-link PROMs suffer from regrowth. I'm
not sure whether the slightly newer titanium-tungsten PROMs have this problem
as bad. A friend has reconstructed PROM contents by optical inspection. The
regrowth is visually distinctive from original unblown fuses. I speculate
that data might also be recoverable by programming the part; regrown fuses
shouldn't take as much current to blow as unblown fuses.

> There are reliable electronic stores. Braided wire ROM is one of them.
> But none of them are high-density, or are anything like the typical
> memories of 1999.

I wonder how well the HP 9100 calculator will hold up? It uses wire braid
ROM for its microinstruction decoder, and PCB memory for its control store.

How long will a CRT hold a usable vacuum?
Received on Sun Jan 17 1999 - 20:23:42 GMT