1
\$\begingroup\$

I'm trying to determine the life of some EEPROM. I will be erasing a byte (setting to 0x00, clearing a parity bit) as frequently as I will be writing to it. So if setting it to 0x00 counts as a wear cycle, I will effectively be having the number of cycles I can use that address.

Does writing a 0 count as a wear cycle or not?

EDIT: it is the act of flipping a bit back and forth that I am trying to estimate wear on, not writing a 0, since EEPROM is 0xFF when erased. So, if I start with erased memory (0xff) set an address to 0x00, then back to 0xff, then back to 0x00, does this count as two wear cycles or three?

\$\endgroup\$
2
\$\begingroup\$

See also 100k EEPROM writes "per bit" or as a whole?

The polarity is the other way up for most chips: "erase" sets a byte or group of bytes to all 1s. So you need to be a little careful. Because writing the byte to 0x00 incurs an erase and a write, and then setting it to e.g. 0x01 incurs another erase (to 0xFF) then a write.

\$\endgroup\$
  • \$\begingroup\$ ... which suggests that wear could be reduced (to half?) by checking the status before writing. If it's correct then leave it. \$\endgroup\$ – Transistor Oct 6 '16 at 14:23
  • \$\begingroup\$ @Transistor: unlikely to gain you anything for those where you erase in blocks \$\endgroup\$ – PlasmaHH Oct 6 '16 at 14:25
  • \$\begingroup\$ Thanks for the quick replies. To clarify: sorry, I haven't looked into the logic enough to know if this chip is default 1 or 0. My point is, does resetting the memory to its default value cause a wear cycle, or is it just switching it from the default to the non-default? In other words, if I set a bit, clear the bit, then set that bit (not getting hung up on whether setting/clearing is 0 or 1), does that count as 2 wear cycles or just 1? I can always invert my logic, it's how to count wear cycles that I'm hung up on. \$\endgroup\$ – Bob Oct 6 '16 at 14:41
  • 3
    \$\begingroup\$ I believe that wear is specified in terms of erase cycles. So 0xFF -> 0x00 is a write (not a wear cycle), then 0x00 -> 0xFF does count as a wear cycle. Note that you can clear bits one by one without incurring wear cycles, e.g. 0xFF -> 0x7F -> 0x3F ... -> 0x01 -> 0x00 does not count as a wear cycle. \$\endgroup\$ – pjc50 Oct 6 '16 at 15:10
  • 1
    \$\begingroup\$ @pjc50 I think you should include that technique in your answer \$\endgroup\$ – pericynthion Oct 6 '16 at 16:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.