>     Can you say, "You get what you pay for?"
>
>     This list and the Microchip BBS are full of complaints about
>     PICSTART reliability.  As far as I know, no one has EVER had a
>     problem with PRO MATE reliability.
>
>     Also, the PRO MATE (unlike the other two) is a "production"
>     programmer; it verifies chips at both Vmin and Vmax.

How much of an issue is this?  One of the projects I'm designing includes
a connector so that, if needed, firmware updates may be done in-system (it's
a 16C84).  Is there any reliability issue with just programming it at 5 (the
system will only run at 5 volts)?  My conjecture is that, for any reasonable
number of program/erase cycles the probability of the chip's EEPROM failing
is much less than the likelihood of any cobbled voltage-selector failing.

Also, if a firmware update requires only changing a few words, is it better
to erase and reprogram the whole array or to simply reprogram those words
that have changed?