I suspect that the Silicon Welfare League would target me.  I am having
medium to long term problems programming 16F870's and 16F877's with my
homemade parallel low voltage programmer and software.

Initially all starts off well and I can reliably blow quite significant
programs on fresh silicon (typically 256 - 512 words in length located both
at the low and high end of the memory map).  As I develop these programs
(including a butchered version of the Microchip bootloader) I find that my
chips start ailing.

Initially this shows up as unexpected run time behaviour (e.g. sending
spurious characters to the serial port).
Then programming becomes hit and miss (I can program 8 words or so - the
exact number varies - of program memory then it will fail.  Repeating will
result in a further 8 or so words beng programmed before failing)
Finally the IC dies and will not be programmed at all (returning 3fff).

Typically I have been having these problems after 10 - 20 successful cycles.
I recently got my bootloader working and I was therefore hoping that all
these problems were behind me but after somewhere around 50 cycles my
bootloading IC exhibited the same problem.  (I tried reprogramming the
bootloader but to no avail)

So I guess the question must be

Any ideas on what I am doing wrong?

I have discovered the RB3 thing, and I do stick to the rules.  However,
after 4 dead F877's and 2 dead F870's my patience is running out.    Oh, a
couple of supplementary Q's

If the RB3 rule is violated does this cause irretrievable damamge to the
PIC?
Or does a reset sort things out?
Is there a set of steps guaranteed to get a working 16F87* back to the state
in which it was shipped by Microchip?

Thanks for any help

Ian

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.