NFC tag stops working suddenly in very rare cases

RE: NFC tag stops working suddenly in very rare cases

Yannik — Wed, 19 Jan 2022 15:28:07 GMT

Hi,

Maybe i got the same or a similiar issue on SDK17.1 with nrf52840.

First i got the lost event and aftert that the detected event.

<debug> nfc_platform: Field detected
<debug> nfc_platform: Field lost
<debug> nfc_platform: Field detected
<debug> nfc_platform: Field lost
<debug> nfc_platform: Field detected
<debug> nfc_platform: Field lost
<debug> nfc_platform: Field lost
<debug> nfc_platform: Field detected

When this occurs nfc stops working. No assertion or anything.

Best regards

Yannik

RE: NFC tag stops working suddenly in very rare cases

Otso Jousimaa — Fri, 16 Oct 2020 12:41:46 GMT

I haven't tried on 17.0, I'll check it out

RE: NFC tag stops working suddenly in very rare cases

Jared — Fri, 16 Oct 2020 12:41:02 GMT

Hi,

This fix is included in SDK 17.0. Have you been able to reproduce it with that version?

RE: NFC tag stops working suddenly in very rare cases

Otso Jousimaa — Thu, 15 Oct 2020 13:02:58 GMT

Any update on this one? I diffed SDK17.2 files to SDK15.3 files and found only changes in comments and binaries

RE: NFC tag stops working suddenly in very rare cases

Jared — Thu, 29 Aug 2019 14:37:27 GMT

Hi.

Sorry for the late reply.

Thank you for a good explanation of both the bug and the steps for reproducing it. I've reproduced the bug and have reported it. I assume that it will be fixed in a future release.

Best regards

Jared

RE: NFC tag stops working suddenly in very rare cases

timo151 — Fri, 23 Aug 2019 08:16:54 GMT

Hi,

Yes, I am seeing the bug every time the FIELDDETECTED and FIELDLOST event is generated simultaneously. When these event are not generated simultaneously, there is no issue.

The problem to reproduce is that it's difficult to generate FIELDDETECTED and a FIELDLOST event simultaneously. You can increase the probability of error case, if you spend more time in interrupts (e.g. add a delay in an interrupt handler --> I added a 10ms delay at the end of nrfx_nfct_irq_handler to reproduce it). Then you have to read the NFC tag multiple times with a mobile phone.

You can use an example of the nRF5 SDK v15.3.0, e.g. examples\nfc\record_text. Add the macro DEBUG_NRF to the Makefile in order to see the assertion. It also helps, when you enable the log in nrfx_nfct.

RTT Viewer output, when the error occurs:

0> <debug> NFCT: Field detected
0> <debug> NFCT: Field lost
0> <debug> NFCT: Field detected
0> <debug> NFCT: Field lost
0> <error> app: ASSERTION FAILED at ../nRF5_SDK_15.3.0_59ac345/integration/nrfx/legacy/nrf_drv_clock.c:328

Regards,
Timo

RE: NFC tag stops working suddenly in very rare cases

Jared — Tue, 20 Aug 2019 11:55:49 GMT

Hi,

Thank your for a great explanation of the problem

[quote user="timo151"]It happens very, very rarely. Therefore it is difficult to reproduce.[/quote]

What happens when the issue doesn't reproduce? Is the problem that it's difficult to generate FIELDDETECTED and a FIELDLOST event simultaneously, or is that the driver sometimes doesn't ignore the FIELDDETECTED event which doesn't cause the bug to happen? In other words, are you seeing the bug every time the FIELDDETECTED and FIELDLOST event is generated simultaneously?

I would have to reproduce the issue before i could report it as a bug. Could you share a project that would be able to reproduce the issue sometimes and can run on the DK , and could you give a rough estimate % wise of how frequently you're seeing this bug?

regards

Jared

RE: NFC tag stops working suddenly in very rare cases

timo151 — Fri, 16 Aug 2019 10:50:19 GMT

Hi Jared,

Yes, this is working fine.

The problem is really in the NFCT IRQ handler. Only one field event is handled (field detected or field lost) at once, even if both events has occurred. This can cause the lost event to be handled twice in a row. The first time, when the filed is gone. And the second time (later in time), when the field was there for a really short time and both events were handled in the same IRQ handler call. In this case nrf_drv_clock_hfclk_release() was called one time too much, which leads to an assertion.

It is really a bug in the NFCT driver. It can be fixed as described.
It happens very, very rarely. Therefore it is difficult to reproduce.

I hope, my explanation is clear enough...

Regards,
Timo

RE: NFC tag stops working suddenly in very rare cases

Jared — Fri, 09 Aug 2019 13:54:05 GMT

Hi,

I'm suspecting that the issue might be related to this errata for the nRF52832. Is TASK_SENSE triggered right before the FIELDDETECTED event occurs?

regards

Jared