Very large drift of HFCLK vs. RTC (LFXO) on nRF52832

[Timofei]

Hello!

I am trying to understand why the HFCLK of the DecaWave DWM1001C module based on nRF52832 has a significant (7000-10000 ppm) frequency offset w.r.t. its nominal value.

I configure RTC1 (with LFXO) to produce ticks at 1024 Hz and every 1024th tick I log the counters of both RTC1 and Timer1. Timer1 is set to 1MHz frequency. The first page of the attached PDF shows the frequency offset of Timer 1 from the nominal 1MHz (y = delta/1e6 - 1), with respect to the RTC.

The second page shows the same but for the DecaWave internal clock vs. the RTC, and there the offset is moderate 70 ppm, which tells me that the frequency of LFXO is quite close to the nominal 32.768 kHz.

What surprises me is that the frequency offset of HFCLK seems huge, and also that it is very unstable, changing in time within 3000 ppm.

I would appreciate any suggestions of what to check. I am wondering if it is caused by wrong components used in the module (there are some indications of that here: https://decaforum.decawave.com/t/dwm1001-issue-with-32-768khz-xtal-for-nrf52832-rtc/6591).

Thank you!

Timofei

drift.pdf

[Jay Tyzzer]

Hi Timofei,

The nRF52832 has 4 Oscillators .  a 32Khz RC and Crystal and a 32Mhz RC and Crystal.

Are you sure you are turning on the 32Mhz crystal Osc and not running off the internal 32Mhz RC? The RC is on by default. 

Please review section 19.1 in the product spec. v1.4

Br,

Jay

  

[Timofei]

Hi Jay,

You are right! Thanks a lot for the very quick and precise reply!

Best regards,

Timofei