I have now done my next step: equipped a new board with a brand-new
AT90CAN128: on this part I can do correct single-byte eeprom write
operations both with the JTAG MKII and the JTAGICE3 without any problem.
And I still can reproduce the fact that another target CPU does not
present any problem when written with the JTAG MKII, but serious
problems when I write with the JTAGICE3 (and I do not have the time to
start an experiment to measure after how many writes this new part will
start failing when writing it with the JTAGICE3).
It looks like the problem of writing eeprom with a JTAGICE3 is specific
to parts where an eeprom position has already seen some 100 write
operations (and that is still perfectly writeable with a JTAG MKII) -
but to make this kind of statement, I should know more about EEPROMs.
Note that there is a problem of statistical relevance: when I say
"always" or "never" I mean within the number of tries I can do within
the limit of time I have, and my reluctance to do too may writes on a
given position. But, "no problem with JTAG MKII" means that in the last
> 5 years and handling some 50 parts I had to throw away a single part
that I used for much development work out some 50 which never presented
a problem, meanwhile there is an epedemy when I use the JTAGICE3.
So, it looks like a good hypothesis is to assume that the JTAGICE3 is OK
for new parts, but presents serious problems as the number of eeprom
writes to a given position increases.
Yes, I think that this is worth submitting a ticket to Atmel.
Juergen
_______________________________________________
AVR-chat mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/avr-chat
