32.768Khz crystal frequency offset detection

[james]

Hi,

   We had a product developed with nRF52810. We found some product would frequently disconnect with samsung cellphone.We tested 2.4G frequence of the product in fix frequence, the 2.4G frequence was fine. So we guessed it may be caused by 32.768Khz crystal ,  we changed 32.768Khz crystal , and then it didn't disconnect with samsung cellphone. Is there a way to measure  32.768Khz crystal?

[haakonsh]

You can f.ex. measure the crystal's signal directly by using an oscilloscope to get start-up time, frequency offset, jitter, etc. 

One pitfall here is that a typical probe adds 9-12pF load to the signal. This can be overcome by either using an inductive probe, active probe (1-2pF load), or temporarily change the value of the crystal's loading capacitor so that the total capacitive load matches the original load cap value. 


If you only need to measure offset you can can set up a GPIOTE channel to toggle a pin on RTC compare/overflow event, and then measure the pin with an analog or digital scope. Here you'll want to run the RTC without prescaling and toggle the pin at large time periods, in order to average out the timing "error" of the PPI peripheral. PPI is bound to the 16MHz clock and need one clock cycle from a received EVENT to trigger a given TASK (min 61.25ns, max 122.5ns).

An RTC overflow event will happen every 1s. PPM can then be calculated: ( 1s - (1s / (1s + error))) * 10^6.
The maximum measurement error introduced by the PPI system is -0.1225PPM at 1s, which should be more good enough for clock accuracy verification. 

[james]

Hi haakonsh

  Thanks so much.It helps me a lot.