[U-Boot] OT-ish: NOR flash write speed degradation?
Hi all,
As I feel this list is a good resource for embedded design minds, please forgive this elementary question.
As NOR flash ages, does write speed degrade? Or do writes take place at roughly the same rate over time until the part reaches the write limit (100k-1000k writes)? Any pointers in the right direction would be great.
Thanks for your time.
On Mon, Aug 17, 2009 at 12:58 PM, Jake Peavydjstunks@gmail.com wrote:
Hi all,
As I feel this list is a good resource for embedded design minds, please forgive this elementary question.
As NOR flash ages, does write speed degrade? Or do writes take place at roughly the same rate over time until the part reaches the write limit (100k-1000k writes)? Any pointers in the right direction would be great.
Thanks for your time.
I believe it's the erase time that grows - it takes longer to tunnel the charge off of the floating gate as it ages (which is a separate mechanism from the write).
With some of the original flash parts (Intel 28F008 & friends, IIRC), you had to time the erase operations yourself, and you could get some extended lifetime from the parts by just allowing more time for the erase to complete. These days that's all handled by embedded state machines in the parts. Also the old parts didn't have an internal charge pump, you had to feed them +12V for programming and erase ops.
Dear Jake Peavy,
In message c5221b990908171058s5e3b5f76j1aa6c6e0abc269b9@mail.gmail.com you wrote:
As I feel this list is a good resource for embedded design minds, please forgive this elementary question.
You are welcome.
As NOR flash ages, does write speed degrade? Or do writes take place at
Yes, it does. Both erase and write times grow, sometimes significantly compared to "virgin" values when the flash was new.
roughly the same rate over time until the part reaches the write limit (100k-1000k writes)? Any pointers in the right direction would be great.
We've seen doubling of erase and write times much earlier, say after some 10...100 erase/write cycles.
Best regards,
Wolfgang Denk
participants (3)
-
Andrew Dyer -
Jake Peavy -
Wolfgang Denk