ESP8266 Community Forum

- Mon Jan 12, 2015 4:48 pm #7132 The module from Olimex uses different Winbond chip - W25Q16B. It says 25uA idle (no communication) and 1uA in power down mode.
What is interesting is the "release power down" is done by "read chip id" instruction. So if ESP boot code issues this read chip id it will effectively wake up the SPI flash.
This needs to be proven with logic analyzer on the SPI bus and the fun can begin...

But still this idle consumption is not a problem - even in most extreme scenarios of rarely waking sensor nodes, the average active of 70-100mA for 2-3 seconds every 30minutes will be higher that the 40-50uA of the chip (or chips).

I suspect that some of those sleep uAs might be caused by floating inputs? Has anybody tried to solve this by setting fixed levels with external weak pull up/down resistors?

- Mon Jan 12, 2015 6:38 pm #7144

But still this idle consumption is not a problem - even in most extreme scenarios of rarely waking sensor nodes, the average active of 70-100mA for 2-3 seconds every 30minutes will be higher that the 40-50uA of the chip (or chips).

This is an excellent check before launching into a mission to reduce power off draw - look at the contribution the wake-time makes.

6 seconds per hour at 100mA for wakefulness (to join wifi and send data) = 6/3600*.1 = 0.00016 Ah
3594 seconds per hour at 50uA for sleep = 3594/3600*0.00005 = 0.000049 Ah
This is equivalent to an average drain of approx 200uA.

So the sleep drain seems to be a smaller factor than the awake power draw, but definitely still significant.

Please be kind to me if I've made a late night basic maths error, btw

Last edited by gwizz on Tue Jan 13, 2015 6:23 pm, edited 1 time in total.

- Tue Jan 13, 2015 1:26 pm #7212 Yes, your calculation are correct. I am sure that the idle comsumption can be optimized further - right now I am quite busy but I plan to check if some floating pins (inputs) could add to the consumption.
Another test would be to pass the power supply of the SPI flash through a jumper or switch, and then to bring the module in deep sleep, turn the switch off (or unplug the jumper). At this point I can separate measure the SPI flash consumption and the ESP8266 consumption.
Without proper documentation we have to be prepared that those chinese modules might have some leakage somewhere.
Another good test will be to measure how the chip consumption in active mode is affected when CPU is clocked to 160MHz.

- Tue Jan 13, 2015 2:37 pm #7218 Keep in mind also that your battery will have a self discharge current... dependent on battery technology, temperature and remaining capacity. I believe 50 uA is normal for a LiPo, much worse for LiMH and NiCd.