> I was proposing that the 8 (or more if you sacrifice the ID locs) test
> memory words be used to estimate endurance of a particular 16C84/16F84
> before it is first used.  I hoped you could shed some light on whether
> this idea is a non-starter.  It may be that the test memory has
> already been used by Microchip quality control for example.  If the
> idea has any merit it could be a "value-added" feature of an '84
> programmer.

Actually, an idea I'd like to incorporate into my programmer, if I can get any
"official" word that it'll work, is an automatic chip-detect, at least for the
'84 (and probably '822) parts since they require a different algorithm from
the other parts.  While the '84 programming algorithm would probably work okay
on other OTP parts, it put extreme stress on the 16C622JW I (accidentally) ran
it on; after erasing that chip, the locations I had previously programmed as
zeros could not be rewritten but would permanently read as ones.  If I could
determine whether a chip was an '84 versus "something else" before I program-
med it that would help greatly in avoiding this kind of disaster.