As discussed here [20736]Fine timer fires during sleep ? using the Fine Timer significantly increases the current consumption of the 20736/737 during sleep, as the timer can't truly be disabled or killed. As also discussed here Turning off the timers in BCM20736S and here Re: FineTimer the bt_clock_based timer is a good alternative to the fine timer. Using the bt_clock_timer reduces my sleep current consumption to close to spec.
However, now I've started implementing a master/slave scenario as described here Testing Master and Slave Roles
I am finding that the master does not hold connections very long (or sometimes at all) when I'm using the bt_clock_timer. If I revert to using the Fine Timer for my application-specific routines, then the master/slave works flawlessly but then I'm back to consuming significant current during sleep.
Have the software development team put any thought into allowing the Fine Timer to be truly disabled during sleep (or even to make the timer period modifiable at run-time), so that current consumption is reduced?
I'm going to dissect my application routines and examine the timing, to see what I can do to make the bt_clock_timer more stable in the central role, but there may not be much I can do there. Are there any other thoughts on what I can do to balance my need for stability Vs sleep current?
