Direct test mode sample crashes when using LE coded PHY

RE: Direct test mode sample crashes when using LE coded PHY

Simonr — Thu, 26 Nov 2020 08:36:04 GMT

[quote user="basvkesteren"]I think integrating the nRF5 code into that would be tricky due to the different API's and such...[/quote]

Indeed.

[quote user="basvkesteren"]If I disable the assert-check in nrfx_timer_enable(), the code runs. Dunno what I'm breaking by doing that, though[/quote]

Erratum 172 refers to BLE long range co-channel performance in the nRF52840 where you can experience packet loss when a blocker signal is present at the same or a nearby RF frequency. The consequence here is that the BLE test with co-channel interference (RF-PHY/RCV/BV-29-C) will fail without the correct FW workaround.

It seems like this workaround might not be implemented correctly in NCS as of yet. I will report this internally and come back to you with any updates on this. Thank you for updating us with your discoveries. I can't guarantee that this will be prioritized and fixed immediately though, as we already have a functioning DTM example in the nRF5 SDK, but it should find its way into an NCS release eventually.

Best regards,

Simon

RE: Direct test mode sample crashes when using LE coded PHY

basvkesteren — Wed, 25 Nov 2020 15:30:52 GMT

Hi Simon,

Late reply, sorry about that; have been focusing on some other problems.

I took a look at the nRF5 DTM source; I have the DTM code integrated in our own (nrfconnect based) application; I think integrating the nRF5 code into that would be tricky due to the different API's and such...

I did do some more debugging on the nrfconnect code: after receiving a packet, something goes wrong with the anomaly 172 workaround code: in dtm.c, dtm_wait_internal(), the following piece of code:

if (dtm_inst.state == STATE_RECEIVER_TEST &&
      nrf_radio_event_check(NRF_RADIO, NRF_RADIO_EVENT_READY)) {
    nrf_radio_event_clear(NRF_RADIO, NRF_RADIO_EVENT_READY);

    if (dtm_inst.anomaly_172_wa_enabled) {
      nrfx_timer_clear(&dtm_inst.anomaly_timer);
      nrfx_timer_enable(&dtm_inst.anomaly_timer);
    }
  }

The nrfx_timer_enable() call results in an assert:

ASSERTION FAIL [m_cb[p_instance->instance_id].state == NRFX_DRV_STATE_INITIALIZED] @ WEST_TOPDIR/modules/hal/nordic/nrfx/drivers/src/nrfx_timer.c:150

If I disable the assert-check in nrfx_timer_enable(), the code runs. Dunno what I'm breaking by doing that, though 🙂

I tried to compare this piece with the nRF5 code; it looks similar. Did notice that the nRF5 code lacks the assert-checks the nrfconnect code has.

RE: Direct test mode sample crashes when using LE coded PHY

Simonr — Mon, 26 Oct 2020 12:58:13 GMT

Hi Bastiaan

Quite a bit has changed between the nRF52 and nRF53, so applications made specifically for one device is not necessarily made for the other.

The DTM example is rather simple, which is why the 53-version works for nRF52 devices as well, but the Coded PHY layer requires some configuration changes in both cores of the nRF53, which is likely why your application is crashing.

I suggest you use the nRF5 SDK's DTM example which is specifically made for the nRF52 series and supports all these PHYs by default.

Best regards,

Simon