Old MOS Mask-Programmed ROM forgetfulness?

[Eric Smith]

On Mon, Feb 15, 2016 at 11:25 AM, Rick Bensene <rickb at bensene.com> wrote:
> I know some of the early MOS ROMs had issues with metallization creep
> that would cause data loss/corruption.

That would certainly do it, but it's the first time I've heard of it
with regard to masked ROMs.

> I have three different EPROM programmers, but sadly, none of them have
> the capability to read these parts.   I was I had a Data I/O programmer,
> but alas, haven't come across one with all the Unipak modules I'd need
> at a price I can afford.

It's not obvious that even a Data I/O would read them. With at least
"recent" Data I/O models, they often won't read masked-ROM
microcontrollers because they detect that the supply current is out of
range for EPROM parts.

If I were going to read those, I'd wire them up to the pins of a
microcontroller. I've been doing that lately for other non-standard
ROMs, such as the Western Digital microcode ROMs used in the LSI-11,
Alpha Micro AM100, and Pascal Microengine.

Obviously you should try to read them by electrical means first.
However, if you eventually become completely convinced that you
haven't (or can't) get the correct data out of them in that way, it
will be time to start thinking about decap. It seems fairly likely
that it will be possible to optically distinguish the intended states
of the bits, as opposed to the actual electrical connectivity.  This
has been done successfully with NiCr bipolar PROMs suffering from fuse
regrowth, though the problem for masked ROMs isn't completely the
same.