Battery measurement

[theopensourcerer]

I noticed your comment about measuring the battery voltage...

I am simply using the internal measurement of the ESP (this means the ADC doesn't work but you get a decent idea of the power anyway.

This screenshot above was a test of an environment monitor that slept for 10 seconds, then it ran for about 5 seconds to send data back to my server. I wanted to see how long an 18650 (~2500mAh) would last. (About 36 hours).

But the interesting data is the relatively slow decay of voltage as seen by the ESP. I am using an MCP1700-3302E LDO regulator: http://uk.rs-online.com/web/p/low-dropout-voltage-regulators/6989044/

[Zentris]

I think, this LDO is very tightly dimensioned:
I have measured peaks up to 350mA while the WLAN transmitter is active...
Futuremore i think, the wakup/sleep ratio (5s/10s+5s = 0.3 ) is to big:

A medium current of 200mA assumed:
200mA * 0.3 = 66mA => 2500mAh / 66mA = (around) 37h working time... :-)

Suggestion: Try to increase the sleep time (if possible) to 300sec (!).

[theopensourcerer]

I have :-). That was my first test just to check reliability.

The same circuit board is now running on a 10 minute sleep cycle - which is how it will be when it goes into my Polytunnel. This run is on a different battery and I have a suspicion that some of mine are less good than others.

But the point was that voltage measurement doesn't need to use an external divider and the ADC port.

https://github.com/theopensourcerer/esp8266_envmonitor/blob/master/code/src/main.cpp#L46

The graph clearly shows how the voltage degrades. The good thing was that the ESP was still able to transmit when the supply VCC was only 2.63v.

[theopensourcerer]

My Environment test is still running since the 6th February and the VCC measurement of the ESP has not dropped below 3.20v yet, although I suspect it is about to. It has been "bouncing" between 3.21v and 3.20v almost since the very start of the test, but now it is spending more time at 3.20 and less at 3.21v.

[theopensourcerer]

Just to follow up on this, my test on a 10 minute sleep cycle ran for approximately 28.7 days. Which is OK for my needs.

[lh84]

@Zentris when you place a big cap of 1000uF between vcc and gnd of the ESP you can minimize the peaks very well. See the video of Andreas Spieß. He explains it very well. https://youtu.be/6SdyImetbp8