> Just to add my 2 pennies to this discussion, I had a problem with a
> so-called universal programmer of commercial origin. A particular line of
> EPROM's I was trying to program had high failure rates when it came to the
> verify cycle. I kept the chips that had failures and one day when I had
some
> spare time (or it might have been that we were running out of chips and I
> desperately needed some) I got stuck in to investigate the problem. It
> turned out that the programming pulse waveform was nothing like the timing
> WRT the other waveforms that the datasheet called out. I ended up building
a
> little circuit on matrix board using some one-shots that timed according
to
> the data sheet, and an address counter so I could copy from one EPROM to
> another. The result was I could now verify the "faulty" devices as good.
>
> It appeared that the original testing worked OK on the devices they had
> tested with, but a process change by the EPROM manufacturer meant that the
> waveforms had to be followed exactly for correct programming of newer
> batches. It is this sort of problem which plays havoc with any attempt at
a
> hobbyist universal programmer, as it is always "someone else's design"
that
> needs fixing. Witness the problems some people are having (as reported on
> this list) with the 16F84A chips compared to 16F84 chips.
>
> --
What you describe above is the best scenario Alan. A considerably worse
scenario in a Commercial environment is a programmer which does not follow
the chip manufacturer's programming algorithm but does result in
a verified chip. However, it can result in a cell capacitor charge which is
marginal. Correctly charged the manufacturer may guarantee a cell charge
retention of say 10 years. With a marginal charge, Murphy's Law more of less
guarantees the failure of shipped equipment within the warranty period due
to one or more EPROM cells flipping state.

Now within 10 months of being shipped, equipment starts being returned as
faulty and the fault proves to be due to flipping cells in the EPROM. It is
found that the programmer does not conform to the chip manufacturers
algorithm.
These are the first units shipped out of a total of over 10000.

What does one do?

There is the Douglas Adams approach : Don't Panic, put on the Peril
Sensitive Sunglasses, and hope that all the other units fail after 1 year
and 1 day. (i.e. Out of Warranty Period)

or one could sell the business quickly at a knock down price due to ill
health or any other excuse you can think of

or one could wait for the inevitable bankruptcy

Chris Carr
Owner of a pair of Peril Sensitive Sunglasses.
The best investment I ever made.

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu