0
\$\begingroup\$

Posting here, since Nordic DevZone is broken for me (Made a long post that didn't upload, just kept loading).

I am currently approaching this from HW point of view (I personally not developing firmware).

Using this scheme: nRF52 circuit powering scheme for nRF52

Before nRF goes into low power mode, the nRF_COIN_EN is pulled high to attach the coin cell. Every other GPIO is pretty much disconnected (not to consume anything else). When 3.3V disappears, the Vcoin is the only thing keeping nRF alive. nRF comes back up when 3.3V is applied (detecting this via nRF's LCOMP). RTC keeps track of time, when Vcoin is above 2.5V (nRF sees ~2.4V). Under that RTC data is gone.

Shouldn't it be that the nRF is able to run even down to 1.8V...? but I can't seem to get it so.

The current draw is ~80uA peaks (I think when RTC is counting), 11uA average. The voltage levels are indistinguishable when RTC data is preserved or when it's not. I have noticed that powering on 3.3V when Vcoin is below a certain level it always loses RTC data and when above a level, RTC data remains intact.

CH1 - yellow - VnRF CH2 - blue - Vcoin

nRF52 power levels

Vcoin (blue line) was about 2.51V The drop during current peaks is about 80 - 100mV. This remains the same whetver the coin cell is full or even lower than 2.3V (when RTC data for sure is gone).

nRF52 power levels, math measurement

MATH measurement. Measured the difference btw Vcoin (blue) and VnRF (yellow). I don't see anything dramatic here to suggest that RTC has lost it's data.

Any ideas what else should I check ? Maybe there are some firmware things we need to check to allow nRF to run down to 1.8V ?

\$\endgroup\$
0
\$\begingroup\$

The current draw is ~80uA peaks (I think when RTC is counting),

That is probably too slow measuring equipment. The peak sould be MUCH higher, as the RTC is very simple and needs the CPU (in)frequently. The CPU core requires current in the mA range - which your coin cell cannot hope to provide.

Luckily, the current draw is rather short. A capacitor in the 100µF range across VnRF might do the trick.

Almost forgot: The NRF chip triggers POR when VCC changes more than 0.3V in a short time frame. This might bite you here, too.

\$\endgroup\$
  • \$\begingroup\$ Hei, @turbo Thanks for the info! For clarification, doesn't the capacitor banks on the nRF supply pins (which I'm essentially powering from coin cell) already provide enough buffering ? Will do some more tests based on this. Will report back. \$\endgroup\$ – crypton Jul 31 '18 at 8:21
  • \$\begingroup\$ Confirming that when nRF gets a power voltage increase (and probly same for decrease as well) of about 300mV, faster than about 100ms, the nRF resets. When I implemented a slow Vcc ramp times (~100mV/200ms), the resets and data loss was gone. \$\endgroup\$ – crypton Apr 13 at 20:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.