Asymmetric Crystal Load Capacitors

[Kyle]

We are trying to do final optimization of the load capacitors connected to the 32MHz crystal on the nRF52840.  A value of 13pF for both capacitors results in a frequency that is about +13ppm over spec.  Using 15pF for both capacitors, results in about -13ppm error.  Using two 14pF capacitors results in almost no error, but 14pF capacitors are difficult to source. 

Is it acceptable to use one 13pF capacitor and one 15pF capacitor?   What are the risks?

Final tuning of the 32kHz crystal is next on our agenda.   Does the same answer apply to it as well?

Thank you!

[Joakim Jakobsen]

Hi!

Is it acceptable to use one 13pF capacitor and one 15pF capacitor?   What are the risks?

There is no problem using two capacitors with different value, if that is what you are asking. Using both a 13pF and a 15pF would be acceptable.

Final tuning of the 32kHz crystal is next on our agenda.   Does the same answer apply to it as well?

Yes, the same answer applies to that as well.

Best regards,
Joakim.

[Kyle]

Hi Joakim,

Thank you for the quick response.  Why does the Nordic white paper (015) that addresses crystal oscillator design state the following:

"The layout of any crystal circuitry should be symmetrical (that is, the same capacitive load present on both crystal pins) to ensure stable oscillation."?

I realize that 13pF and 15pF are similar values, but I'd like to better understand the concerns expressed by the author of the white paper.

Best,
Kyle

[Joakim Jakobsen]

Hi Kyle,

Yes, the whitepaper states that the crystal circuitry should be symmetrical (same capactive load on both pins).
This is to ensure that the oscillator works properly, and that the oscillation frequenzy is as specified.

However, using 13pF and 15pF would be inside the range of acceptable values. Using slightly different values is fine, but I would not recommend bigger difference like, i.e 10pF and 15pF.

Note that the "Crystal Oscillator White Paper" is written for older chips (not nRF 51/52), so there might be some differences to account for.
i.e the input capacitance is ~4pF for nRF52832 and ~3pF for the nRF52840, not 1pF as stated in the whitepaper.

Here is some more information that might be interesting for you:
Tutorial - General PCB Design Guidelines for nRF52-series
Capacitor values nRF52832
nRF52840 PS

Best regards,
Joakim