On a nrf52832 what are the chances of spontaneous flash corruption?

[Don Garrett]

I'm working on a risk analysis, and need to understand the likelihood of a spontaneous change of values stored in flash, along with what usage patterns most affect that.

The primary usage pattern I'm looking at involves writing a page out, then reading it back about once a minute for 10 years. Assuming that the initial write is verified and successful, what are the chances that we will see any form of flash corruption (such as a bit flip) spontaneously happen?

[Susheel Nuguru]

The retention of contents of the data is qualified for 10 years, so if you write once and read it back every one minute for 10 years, then it is still under the qualified range and should work as intended. 

[Don Garrett]

Yes, but what the the odds that it won't? 1/1,000,000? 1/10,000,000?

[Susheel Nuguru]

Hi,

I am sorry, I was wrong. It is not just the writing It is also the read cycle that has retention limits. It is given here

We do not know the odds of failure, as this qualification data comes from the production line and there is no more info provided than this.

[Don Garrett]

Thanks!