Recently I've been trying to measure the deep sleep current of the CC2650 running Contiki. Unfortunatley I am measuring ~40uA average power usage when the board is in deep-sleep mode (standby).
To conduct my test I have a simple test app (https://github.com/bkozak-scanimetrics/CC2650_lpm_test I see this problem in both the 3.0 release of Contiki and the master on github.
A detailed discussion of what I've been doing be found at http://sourceforge.net/p/contiki/mailman/message/34433588/
To summarize my findings thus far:
- I see the 40uA consumption both on the CC2650EM (multiple boards) and on a sensortag
- With a DC analyser I can see that the current is changing between ~50uA and ~0uA (averages at ~40uA) at regular intervals.
- The only way that I can reduce the power consumption is by changing the chip's recharge configuration (HWREG( AON_WUC_BASE + AON_WUC_O_RECHARGECFG ) = 0x00A4FDFD is one configuration which gives very low current usage)
- The RECHARGECFG change is all or nothing, with some configurations I see >1uA and with all others ~40uA. It only seems to make any difference at all if the recharge period is sufficiently large (is this just the chip malfunctioning because charge is lost?)
I've read through the documentation on power modes several times and really can't see anything wrong in the way that Contiki enters deep sleep. I'm afraid there might not be much left for me to try on my end.
Any help would be appreciated.
Thanks!
