Mike Harrison wrote :

[snipped data sheet quotes, they are well known.... :-) ]

> I can see 2 possible interpretations : 
>
> a) if you write data infrequently, you need to do an 
> occasional eeprom refresh

*Only* if there at the same time are one or more EEPROM
cell that is written frequently.

> - but how often ? 

After D124 writes to the *frequently* written cell(s) (if any).

> b) If you write data frequently, you need to to a refresh at 
> least once per D124 (1,000,000) writes

Again, *only* if there at the same time are one or more
EEPROM cells that is/are written *in*-frequently. Any write
will be a "refresh" of that specific cell, so if you only have
"frequent cells", there is no problem.

Or in other words, there is only need for a "refresh" when
there is a *mix* of frequent and infrequent written EEPROM
cells. And the definition of freq/infreq is param D124.

Or yet i other words, if the you have reached a total of
EEPROM writes = D124, and there are still cells *not*
written dring this timeframe, a refresh should be done (of
those cells).

Each write to *any* eeprom cell, slightly discharge all other
cells. After a specific number of writes (D124), there is a risk
that un-written cells are loosing the charge, and has to be
"refreshed".

This also means that if your application knows which cells
are "infrequent", it (the app) only has to refresh those cells.
All other cells has been refreshed while written, of course.

> - does this mean that the endurance is infinite as long as 
> you do this?

No, all we have talked about so far, is the number of writes
before a refresh is needed. Note that this is the total number
of writes to the whole EEPROM, not to a singel cell.

The edurance is the "life-time" (measured in number of
writes) of any *single* cell, which is something quite
different, and has nothing to do with the refresh issue
as such.

And, the data sheet is correct, but maybe not *that* clear... :-)

Jan-Erik.



-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist