Errata 146 of nRF52832

RE: Errata 146 of nRF52832

Jeff4BLE — Mon, 06 Nov 2017 17:16:17 GMT

I believe S132 v4.0.3 is the first softdevice release to allow setting RCOSC to 500 ppm. Please review S132 v4.0.3 Release Notes for more information.

RE: Errata 146 of nRF52832

Remo — Tue, 09 May 2017 15:05:15 GMT

Thank you very much!

RE: Errata 146 of nRF52832

FormerMember — Tue, 09 May 2017 12:27:12 GMT

Yes, that's correct! When using S132 v.4.0.0 and specify the RCOSC accuracy to 500 ppm, the RCOSC should not cause any linkloss.

RE: Errata 146 of nRF52832

Remo — Tue, 09 May 2017 06:31:45 GMT

Thanks for your answer.

I've got one last question just to be sure that I understood it correctly. When I use S132 V2.0.1 there may be some link terminations because the SD expects a LF RC oscillator accuracy of 250ppm instead of 500ppm, but when I would use S132 V4.0.3 and specify the RCOSC accuracy to 500ppm there will be NO problem. Is this correct?

RE: Errata 146 of nRF52832

FormerMember — Mon, 08 May 2017 11:11:35 GMT

1) Unfortunately, I cannot comment on this one.

2) It's not possible to mask this problem fully other than to update to a softdevice that support 500ppm tolerance for the RC clock.

3) Unfortunately, I cannot comment on this one.

4) The latest version of the softdevice should have an option for setting the RC-oscillator frequency accuracy to +/- 500 ppm

RE: Errata 146 of nRF52832

Remo — Wed, 26 Apr 2017 07:40:49 GMT

Thank you for your response. I cannot live with that some devices would always lead to link terminations. This would be a worst-case scenario for us! Therefore, I have some further questions to clear things up a bit:

How many percent are "some very few devices"?
What calibration interval is necessary to mask this problem?
This means that the accuracy is never 500ppm directly after a calibration?
Why does Nordic state that for the BLE stack an accuracy of 250ppm is necessary while the BT core specification requires 500ppm?

RE: Errata 146 of nRF52832

FormerMember — Wed, 26 Apr 2017 06:47:23 GMT

At some, very few nRF52832 devices, the frequency tolerance of the RC oscillator may/will exceed +/- 250 ppm. For the affected devices, the result is larger a packet loss rate, which can lead to a disconnect.

For S132 v.2.0.1 with the RC oscillator, it is only possible to set the LFCLK accuracy to +/- 250 ppm. If you experience that most of the devices work fine, but a very few of the devices exhibit a disconnect problem, those very few devices are most likely affected by Errata 146.

If you can live with having a few devices where the RC oscillator exceeds +/- 250 ppm, there shouldn't be any problem continue using the RC oscillator with a +/- 250 ppm setting.

For BLE, the sleep clock accuracy should be within +/- 500 ppm (BT Core specification v.5.0, Vol 6, Part B, chapter 4.2.2).