Rare counter error

RE: Rare counter error

CheMax — Thu, 16 Aug 2018 08:25:14 GMT

Hi,

It is very likely that this is the case. Did you see any discrepancies in the scoped signal that correlates with errata #155 ?

I Did not see. The analyzer does not show (checked at a sampling rate of 100 MHz), the oscilloscope is difficult to catch this moment.
Perhaps this is a consequence of the effects of static voltage.

RE: Rare counter error

Håkon Alseth — Thu, 16 Aug 2018 08:14:21 GMT

Hi,

The specification does not describe this region of memory. I understand correctly that this is an additional configuration register?

This forces the peripheral clock to be always on in GPIOTE IN module, similar to how the GPIOTE IN behavior was on the nRF51-series devices.

This can be the effect of impulse (> = 750 kHz) interference?

It is very likely that this is the case. Did you see any discrepancies in the scoped signal that correlates with errata #155 ?

Best regards,

Håkon

RE: Rare counter error

CheMax — Wed, 15 Aug 2018 15:27:36 GMT

So, I tested for an hour and a half of uninterrupted recording. This error did not appear.

changes except those described in the Erata *** I did not add.

Can you tell us what exactly this record:

*(volatile uint32_t *)(NRF_GPIOTE_BASE + 0x600 + (4 * GPIOTE_CH_USED)) = 1;

does?

The specification does not describe this region of memory. I understand correctly that this is an additional configuration register?

RE: Rare counter error

CheMax — Wed, 15 Aug 2018 12:10:12 GMT

Hi,

Yes, Channel 8 (AFE Data Ready) is used as a base for signal generation. Signal parameters:
1. the duration at zero is approximately 12.5 microseconds
2. a period of 1 ms

The error occurs entirely by accident. It can occur 1 time in 10 minutes, it can 3 times in 2 minutes.

On the analyzer's record, there is no double front (sample rate was 16 MHz).

This can be the effect of impulse (> = 750 kHz) interference?

I'll try the solution you proposed and write about the results.

RE: Rare counter error

Håkon Alseth — Wed, 15 Aug 2018 11:29:19 GMT

Hi,

Is channel 10 counting based on the channel 8 signal (CSN of AFE_SPI)?

How random is this behavior? 1 in 10?

If one of the input signals is quite fast, you might be running into this errata:

IN event may occur more than once on input edge

Could you try adding this workaround to see if this has any effect on the behavior?

Kind regards,

Håkon