Clock-source selection for ANT

[Richard Phillips]

Hello all,

Am I right in thinking that to use the ANT SoftDevice, I basically must use an external 32 KHz crystal, in order to meet the accuracy requirements?

Looking at the list of possible clock-sources for sd_softdevice_enable(), it seems that only the crystal options have ppm <= 50, and I read that this accuracy is needed for ANT?

I'm puzzled as to why the synthesized clock-source is only listed as 250ppm accuracy - shouldn't it reflect the accuracy of the high-frequency crystal (10ppm in our case)?

Regards, Richard.

[Petter Myhre]

The LF clock source must have a frequency tolerance better than 50 ppm for ANT applications.

So you can use a 32kHz crystal oscillator, with an external 32kHz crystal with less than 50 ppm frequency tolerance. Or you can use the 32kHz synthesized oscillator, with an external 16MHz crystal with less than 42 ppm frequency tolerance.

The synthesized oscillator will draw much more current compared to the crystal oscillator, because it requires the 16MHz crystal to be running.

The run current of the 32kHz crystal oscillator is IX32k=0.4uA.

The run current of the synthesized oscillator is ISYNT32k=15uA, in addition to the run current of the 16MHz crystal oscillator, which is IX16M = 470uA.

This means that you need to replace IX32k with ISYNT32k+IX16M in(or with ISYNT32k in stages where the IX16M is missing) table 20 and 21 in the S210 v2.0 specification.

I'm not sure why NRF_CLOCK_LFCLKSRC_SYNTH_250_PPM is the only one in the list. I'm quite sure it is a bug. In my opinion it should be NRF_CLOCK_LFCLKSRC_SYNTH, since it depends on the choice of 16MHz crystal.

It should be safe to use, but please keep the current penalty in mind.

[Richard Phillips]

Great, thanks - in my application, NRF_CLOCK_LFCLKSRC_SYNTH (to give it it's new name!) should be fine, then.

[Petter Myhre]

Np. Please accept my answer if you find it sufficient.

[Richard Phillips]

How do I do that?!

[Petter Myhre]

Click the grey circle with a check mark in it :)

[Richard Phillips]

Shall do, but can I suggest it's changed from "mark this answer as correct" (which suggests I'm confirming your answer as correct, which doesn't make sense, seeing as I asked you the question!), to "accept this answer" :)