Ticket#2022070610000149 - Colibri T20 NAND flash durabilty

Hi,

we observe flash memory becoming corrupt too fast in one of our applications.
The application is running on a Colibri T20 module with the Windows CE 7.0 image. Most of the affected modules are using image version V1.4.

We have been in contact regarding flash issues already in V1.3, where we moved to 1.4 then.
(In the meantime most of our modules are delivered with V2.2, but not all controls in the field were updated so far.)
There is more or less no difference of the amount of written data to our other applications, where we do not face the problem.
The only difference could be the CPU and module temperature which is definitely higher.
Here are the temperatures measured over a long period of time (via the Toradex Task Manager)
Module: 64°C to max. 74°C, most of the time 70°C and higher
CPU: 71°C to max. 82°C, most of the time 77°C and higher

The T20 datasheet defines the max. operating temp. with 70°C, the max CPU temp. with 90°C.
Can I assume that the operating temp. from the datasheet corresponds to the module temperature of Task Manager?
Are there any experiences if the flash is worn much faster when the temperatures are too high?

Here you can find two logs from the serial interface:
5049784.log (303.8 KB)
3001243.log (9.1 KB)
The file name is also the serial number of the module.

Both modules do not start anymore. Each log contains outputs of “Ecc.Corrected(1) Blk…”
As far as I know, these outputs are an indicator that the flash is getting worse. Is that right?
Interestingly, both modules give different messages. Of course I checked more than these two modules, but the outputs are identical to those.
Is there any way to tell from the logs if the flash is corrupt or if the module is unable to start for some other reason?
Is there possibly any other indicator in the log that shows how good/bad the flash is?

Best regards