Jee Labs Talk

This forum has moved to forum.jeelabs.net - this read-only archive is for reference only!

1. andras
   

   I noticed that the batteries of my rooms nodes with DS18B20 get drained in a week or so. I was surprised because in the december blogs on power optimization the average measured current drawn was found to be less than 0.1 mA.

   I measured the current drawn of the most recent rooms sketch with my multimeter and found a seemingly constant 15mA draw. This was with #define SHT1X 0 because I have the DS18B20 instead of the SHT11. Then I changed this back to 1 (even though I don't have the SHT11) because the 0.1mA was measured in a setup with the SHT11. Now I saw the measured current alternating between 0.1 and 15mA.

   So I was wondering wether the low-power consumption is only achieved with the SHT1X configuration or that I have overlooked something here. Does anyone else also have a similar experience with the DS18B20 rooms configuration ?

   Posted 17/02/2010 09:55:59

2. jcw
   

   Hm... 15 mA sounds like the RFM12B module is on. No idea whether that's a hint of any kind.

   Posted 17/02/2010 12:11:37

3. andras
   

   Yes, it must indeed be the RFM12B module. I noticed that after a new measurement message was sent and succesfully acked ( rf12easyPoll() returns 1), the radio is switched off OK, but after about a second it is switched back on. It remains on untill the next message is sent; again being switched off for only about a second.

   So then the radio must be on even though !radioOn is true. In periodicSleep() I changed the criterium to switch the radio off from if (radioOn && rf12easyPoll()) to if (rf12easyPoll()) to make sure that it is turned off when the polling mechanism has nothing left to do. After that, the current measurement went back to below the accuracy of my multimeter, so that worked.

   Could it be that it is then rf12easyPoll() itself that switches the radio back on ? In that case a better solution would then be to change the polling mechanism to prevent the radio from being switched on when !ezPending is true.

   Posted 02/03/2010 14:30:29

4. matajumotors
   

   I also found out that rf12easyPoll() wakes RF module http://talk.jeelabs.net/topic/162#post-793 . So wrote my own logic using rf1easyInit(0). and that logic calls rf12easyPoll() only if data has changed. But rf12 drive suposed to do it by it self..

   Posted 02/03/2010 18:20:22

5. jcw
   

   Well, it's clear that the room node battery saving logic is not quite finished - most of the battery-powered nodes I have around here have died by now. In a way, that's good: it means I've GOT to dive in again to get it 100% right. The bad news is that I'll need some serious undisturbed time to do so - it's tricky stuff, and these days I'm a bit too hectic / disorganized to reach that level of zen...

   Posted 02/03/2010 22:32:49

6. andras
   

   No problem JC. The simple change described above will still let the JeeNode be in powersave for 99% of the time. That ought to do for the time being so you can continue on focussing your zen on the jeemon stuff ;)

   Posted 02/03/2010 22:55:30