How much time drift is expected on NRF52832 with an external 32Khz 20ppm over a year

RE: How much time drift is expected on NRF52832 with an external 32Khz 20ppm over a year

hmolesworth — Wed, 17 Jan 2024 16:36:49 GMT

Only 2nd place, but happy with that :-)

RE: How much time drift is expected on NRF52832 with an external 32Khz 20ppm over a year

toastedowl — Wed, 17 Jan 2024 15:06:40 GMT

Thank you very much for this detailed information!

I hope you won the contest!

RE: How much time drift is expected on NRF52832 with an external 32Khz 20ppm over a year

hmolesworth — Tue, 16 Jan 2024 17:16:09 GMT

About 20 years ago I wrote a contest entry for RTC 32kHz crystal temperature compensation for the MSP430 but it is equally applicable to the nRF52832. Feel free to use anything useful, with credit for anything I wrote.

I'll attach the c, h and xls files together with the pdf and xls if this system allows all that.

devzone.nordicsemi.com/.../2475.RtcTempCompensation.c

devzone.nordicsemi.com/.../3542.RtcTempCompensation.h

devzone.nordicsemi.com/.../3482.MSP430-RTC-Temperature-Compensation.pdf

devzone.nordicsemi.com/.../RtcCrystalFrequencyCompensation.xls

RE: How much time drift is expected on NRF52832 with an external 32Khz 20ppm over a year

toastedowl — Tue, 16 Jan 2024 15:29:06 GMT

Here is the specification for the crystal.

https://support.epson.biz/td/api/doc_check.php?dl=app_FC-135&lang=en

Assume temperature is 0 - 40 Celsius. If easier, maybe the calculations at the boundaries and middle would shed enough light - e.g. what would the answer to the original question be if (a) it was a constant 0 degrees the whole year (b) it was a constant 20 degrees the whole year (c) it was a constant 40 degrees the whole year.

RE: How much time drift is expected on NRF52832 with an external 32Khz 20ppm over a year

Susheel Nuguru — Tue, 16 Jan 2024 12:16:45 GMT

I think this question is not completely related to nRF related solutions but a general question in how much drift to expect over time given the accuracy of external clocks.

[quote user=""]What is the worst case scenario for the difference between 365*24*60*60 - (latest local timestamp - original local timestamp)[/quote]

nRF app timer totally relies on the internal RTC ticking and the accuracy of the internal RTC is directly related to the accuracy of the clock being used. So you need to check the specifications of the external oscillator or crystal manufacturer to see what is the worst and best case accuracy in different temperatures and humidity. You should be able to calculate the average drift of this clock over any period of time based on this worst and best accuracy given in ppm.

[quote user=""]Is the delta mostly linear so I could then adjust each 10 minute timestamp to get a close match of the real time the reading was taken?[/quote]

This again depends on the external clock being used and there is no generic answer that can be given. But most likely the delta is not linear in my experience and the delta curve will be different with time/temperature and humidity. Only the manufacturer of the external clocks can give better range on how the delta behaves in different conditions.