Conny Andersson wrote:

> I use Windows when developing my programs, maybe slower than DOS but
> with many advantages (you can easily switch between word processing,
> coding, simulating ...)

I agree with you. My assembler/debugger (called Picasm) is a DOS
application, but it runs nicely under Windows (e.g., releases
time-slices, etc.)

>> I have also found a bug with the TO (TimeOut) bit in the Status
>> register. Sometimes the bit is not set correctly after a MCLR reset.
>> If anybody is interested I can post some more info.

> Yes, post it on the piclist.

OK, I'll make a separate posting later since is not related to the
subject of this message (EEPROM) - in fact it's not even related to
PIC16C84 devices.

>> ...
>> To test, I wrote a program that initiated a write and then incremented
>> the EEADR register every 0.5ms. Result: 4 (!) EEPROM locations were
>> erased and the data was written into a 5. location...
>> Looks like Microchip forgot an address latch.

> Interesting, I guess the people at Microchip have the answer...

That was a good guess, because I got a mail from Tom Salt at Microchip.
Here is what he wrote about the problem:

> You caught me here. During the design I knew it would be hard on the part
> to change the data during a write, so I disabled writing to the EEDATA
> register during a program cycle. It wasn't until later that I realized
> that a write to the EEADR register would be just as harmful (and easy to
> do). Changing either of these registers may reduce the endurance that may
> be expected from the part. We are currently developing the 16C84A which
> will have this problem solved. I currently have silicon which does not
> allow either of these registers to be changed during a program cycle.

I'm glad to here you have solved the problem. All we need to know now is
when the new 16C84A will be available ?

BTW is this problem documented in an errata sheet ? I assume it isn't,
tecause I asked exatec (Danish distributor) for some detailed info on
the EEPROM - and got zip :(
IMHO: it would be of benefit to both Microchip and all us PIC users, if
ALL information on the PIC were available - so we won't have to "dig it up"
by ourselves. Intel didn't exactly gain popularity on there "We'll tell
you what you need to know" policy on the FDIV bug in Pentium - If you know
what I mean...

Michael Brinks