0
\$\begingroup\$

When data is written to the EEPROM that is identical to what is already stored, does it have any different effect on endurance than when there bits are toggles?

Is there an effect, like in flash memory, where the bit is hard erased and written again regardless of the state?

I know I can protect the endurance by first reading, comparing and possibly overwriting the bytes, but I'm interested in this from a hardware point of view.

\$\endgroup\$
7
  • 1
    \$\begingroup\$ Typically data is written in pages so even if you modify only one part, you still have to erase and re-write the whole page. \$\endgroup\$
    – floppydisk
    Commented Jan 18, 2023 at 12:57
  • 2
    \$\begingroup\$ That depends on the implementation. Sometimes you have separate erase and program cycles, so you can program without erasing. Sometimes you get automatic erase+program cycle. There may be different implementations what happens if you program an already programmed cell, but of course you can read the cell and program only the cells that need programming so they are not programmed multiple times but just once after erasing. \$\endgroup\$
    – Justme
    Commented Jan 18, 2023 at 13:30
  • \$\begingroup\$ Some EEPROMs have pages and blocks, where a block is the smallest amount that can be written, and writing a block is transparent to the user. The details are often not included in the datasheet. If you are pushing the limits, you may need to talk to the vendor. \$\endgroup\$
    – Mattman944
    Commented Jan 18, 2023 at 13:44
  • 1
    \$\begingroup\$ Not what you asked about, but a collogue of mine once did an FPGA for some company to interface between an EEPROM and the rest of the system because they wrote to it constantly, wearing it out. For some obscure reason, they couldn't touch the code so the solution was an FPGA which looked for identical data already stored and mapped accordingly. If new data, block write as little as possible and internal table of which blocks had been written how many times to do wear leveling. Similar to calling NASA to land a rocket on a nail because you're not allowed to use a hammer. But solved the problem \$\endgroup\$
    – winny
    Commented Jan 18, 2023 at 16:50
  • 1
    \$\begingroup\$ @winny, "For some obscure reason, they couldn't touch the code" I've seen that a lot in systems I've worked on needing intense software testing/approvals, like in defence and military. Software can't change because it breaks the approvals. HDL isn't subject to the same level of approvals testing so hardware/HDL can be added/changed. Amused me that 80 years ago, engineers dreamt of a future where a thing called software could be changed so hardware didn't have to be and changes would be quicker and easier, yet those systems say "Don't change the soft-named bit, only the hard-named bit" :-) \$\endgroup\$
    – TonyM
    Commented Jan 18, 2023 at 20:57

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.