unintended flash erase

RE: unintended flash erase

gorazd — Mon, 27 Aug 2018 13:23:05 GMT

We discarded the module that was causing the problem. Other modules did not exhibit the problem.

RE: unintended flash erase

kamisen — Mon, 27 Aug 2018 09:31:53 GMT

Had your question been resolved?

RE: unintended flash erase

kamisen — Mon, 27 Aug 2018 09:30:20 GMT

Had your question been resolved?

RE: unintended flash erase

MartinBL — Fri, 20 Oct 2017 05:50:58 GMT

Ok, let me know how it goes then.

RE: unintended flash erase

gorazd — Thu, 19 Oct 2017 05:30:25 GMT

Tnx I will attach the resistor. We also found out not all of the modules exhibit the problem or the problem is meybe harder to reproduce. But we had one module which we could reproduce the problem many times in a row on.

RE: unintended flash erase

MartinBL — Tue, 17 Oct 2017 13:48:33 GMT

Hi,

I'm unable to open your files. They seem to lack file extensions. You can edit your question and attach them directly in your post.

The nRF51 has a design flaw related to the SWD and reset pins and is known to unexpectedly reset itself and maybe enter unknown states. This may be caused by noise on the SWD lines generated by for example magnetic fields. The workaround is to attach a 1k pull-down restistor on the SDWCLK pin as shown in the latest reference schematic. Have you done this? Anyway, I'm unsure how this would result in an erased flash, but the cases I have seen related to magnetic fields have also been related to noise on the SWD lines.

RE: unintended flash erase

gorazd — Mon, 09 Oct 2017 11:52:00 GMT

Additional information: we have a level shifter TCA9406 to communicate with motor over uart which is on 5V. Also, we suspect the flash erase happens after the motor (we dont even start the motor spinning and advertising works just fine up until then) is plugged out which results in capacitors slowly discharging and eventually putting our module into reset. After that the module is dead (flash empty) and the module needs to be reprogrammed to work again.

RE: unintended flash erase

endnode — Mon, 09 Oct 2017 08:56:26 GMT

Really looks nicely erased... to do such mass erase it would probably need to write into UICR and then reset the chip. Not sure how probable this is in case you have the chip in strong magnetic field. But it would be more probable that some corruptions happen already before so it would be interesting to dump the chip while it is still working and compare against the original HEX image.

RE: unintended flash erase

gorazd — Mon, 09 Oct 2017 08:43:13 GMT

I leave it by default (unprotected)

Outside the housing everything works fine (This is inside the flash)

Once in housing it stops working (This is in flash)

EDIT: broken links fixed

RE: unintended flash erase

endnode — Mon, 09 Oct 2017 08:27:53 GMT

This is quite strange. How exactly you check the flash state? You leave it unprotected (RBPALL register not set) and then once it stops working you dump it and whole flash is erased meaning set to 0xFF..FF? Isn't there other kind of corruption before?