Question about eeprom in 16F886 ?

sorasit46 · Joined: 07 May 2007 Posts: 69

Hi!

PCM programmer · Joined: 06 Sep 2003 Posts: 21708

sorasit46 · Joined: 07 May 2007 Posts: 69

Thanks alot.
Yes I write to the same eeprom address continuous.

Ttelmah · Guest

You need to 'rethink' how you are working then.

Multiple approaches:
1) Only actually write the data when power is turned off. This is surprisingly 'easy'. All you do, is add circuitry that detects the incoming supply falling, and ensure that the capacitance on the PIC power rail is sufficient to run the chip for the time needed to write all the data when the incoming supply reaches this point. Then have an interrupt when the power fail is detected, and write the data in this.
2) Only write the data at quite long intervals. For example, on a 'position' controller, which accepts keyboard movement commands, I will detect when a change is initiated, and only write the data, when it has changed, and then stayed the same for 30 seconds. When people are adjusting the system, issuing dozens of movements, no 'write' occurs. Only when they are happy, and leave the value alone, is this recorded.
3) Use another storage medium, like FRAM, that does not have this limitation.

While '100K' cycles, sounds a lot, if you write once a second, it only takes just under 28 hours to destroy the chip. Write faster, and the time is less....
The EEPROM, is not designed for 'dynamic' data, but for storing configuration data that changes infrequently.

Best Wishes

sorasit46 · Joined: 07 May 2007 Posts: 69

Ohh! Thanks. Now I see.
Really I'm destroy the chip in 1 day, because I write as dynamic data.
Now I must detect falling supply before power sufficient to run the chip.
This is new problem.
regards

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

I recently toasted the eeprom in an f887 after as few as 50 or so erase cycles to reprogram the part. I had noted that erase ( on my programmer anyway)
erases the user EEPROM too. no problem.

On the part I wrecked - the symptom was that an int16 checksum stored after last byte in EEPROM kept failing with a rock solid known good code library.

As soon as I uploaded the same hex file into a new F887 - problem solved EEPROM worked like supposed to

The odd thing was if I erased the bad '887 some more - it would work for a while - but then in a day or so - be right back to the no go mode
where I update the eeprom content
and on next power UP - it get the failed checksum message on the data put there. spooky

TX PCM for your input