Eraseblocks torture: OneNAND results

[Ricard Wanderlof]

On Fri, 15 Dec 2006, Enrico Migliore wrote:

> I've recently read  the Micron's "TN-29-16 Boot from NAND..."  technical
> note and got puzzled by the following:
>
> "Avoid excessive reads to the area of the NAND Flash where code is
> stored. When
> repeated accesses are required, the code should be copied to the DRAM.
> This minimizes
> the probability of read-disturb errors in the NAND Flash device."
>
> Does that mean that excessive readings might compromise the integrity of
> NAND cells?

That's the way I would interpret it too. (BTW, the paragraph wasn't there 
in the first edition of the document, rev.A from March this year...).

Although I'm also somewhat puzzled at the conclusion. I always thought 
read errors could occur at any time, the probability of which could be 
increased by reading or even more writing to the device. But not that it 
was localized to the area which you are reading. Although it makes sense I 
guess; accessing a bit cell requires some activity in the vicinity of the 
cell in order to read the bit, which I guess would increase the 
probability of failure.

/Ricard
--
Ricard Wolf Wanderlöf                           ricardw(at)axis.com
Axis Communications AB, Lund, Sweden            www.axis.com
Phone +46 46 272 2016                           Fax +46 46 13 61 30