ESPimatic battery issues

---

Just for comparison: I’m now doing the same with a arduino nano V3 (5Volt 16 MHz version) with a ds18b20 and an rf-transmitter.
Leds removed, power regulators still on, power down for 10 minutes, running on 3 AA batteries.

I was glad to be able to swap my RF433’s for wifi ESP’s, but not if I can run the RF device for months compared to a week on ESP.

@Harry-van-der-Wolf said:

I was glad to be able to swap my RF433’s for wifi ESP’s, but not if I can run the RF device for months compared to a week on ESP.

And that’s the point. I also tried to run the ESP on batteries and run into problems. Sleep, deepsleep, problems with wakeup from deepsleep, reboots and so on.

What do you wanted to replace with Esp8266? Sensors or actuators?

Did some measuring on my nrf + pro mini + PIR. I’m always getting confused with mA > uA

Sending: 17 mA
Sleeping PIR low 0,13mA (130uA)
Sleeping PIR active: 0,2mA (200uA)

Lets say batteries have around 2400mA capacity. Lets use 2200 to be realistic. and lets say the pir uses 0,25mA avarage = 2200 / 0,25 = 8000h / 24h = 366 days.

Measured another pir and that one used 0,08mA (80 uA) and active 0,16mA (160uA). That one should get 458 days (calculating with 0,2mA average.

Conclusion, even with the same hardware it can much differ. Sending could even be lower but i havent removed led13. But I don’t want to remove it for debugging.

I also removed the regulator on the pro mini which uses around 0,1mA. And i’m not even using it. And don’t use internal pull-ups! they use around 60uA.

So for the ESP users. Remove the LED’s, regulators and don’t use internal pull-up (if needed use a 1M external). Power it directly on the vcc with 2 AA batteries. I’m interested to see what the consumption will be. I think nodemcu’s are not good for batteries.

Does some1 have a barebone ESP? I’m interested to see what the consumption is with direct battery and a pir or something.

Off-topic but stil relevant (I think).

@Harry-van-der-Wolf said:

Just for comparison: I’m now doing the same with a arduino nano V3 (5Volt 16 MHz version) with a ds18b20 and an rf-transmitter.
Leds removed, power regulators still on, power down for 10 minutes, running on 3 AA batteries.

And I stopped this one again. In 3 days (or so) my pimatic hung twice. Just because I connected those 433 devices again. Unfortunately I had also removed my “reset espimatic”, but also the “never failing because hardware triggered” watchdog as described here didn’t help.
I now ordered this set.

My devboard is still working after more than 7 days with 10 minute deepsleep on 3 batterys. To be honest : i’m sending the ADC value so no sensors attached.

It stopped last night after 14 days.
The batterys weren’t fresh and due to a config mistake it was up for 9 hours without sleeping.

Not sure if it would be twice as long without these sleep issues and fresh batterys.

14 days and still sending
https://forum.pimatic.org/topic/1500/espimatic-battery-issues/42

EDIT: It stopped after 33 days.

A few days ago I saw a youtube video with some nice tips and tricks to get the most of your battery life out of an ESP. The author has been able to run his ESP-07 for 17 days on a LIR2450 button cell and extrapolated it to 425 days on 2 AA cells. Quite impressive IMHO.
for those interested:
https://www.youtube.com/watch?v=IYuYTfO6iOs

Maybe it is the weather that more people get interested again but this afternoon (outside under the parasol) I wrote a simple sketch to try to maximize the battery life.
I put it here on my github.

I started it this afternoon on a fresh set of 3xAA duracell Simply batteries connected to 3.3V. I did not remove the power regulator. Simply because I still don’t know which chip it is on the nodemcu 0.9.
I did remove all the leds.

So lets see how far it gets.

If I recalculated my last test, according to the video:
2200mAh / 500mAh = 4.4
60 min / 15 min = 4
33 days * 4.4 * 4 = 580.8 days!

And I sent the real data from DHT11 (temp and hum).
The problem is that once an hour is not enough. Maybe so in the greenhouse.

@wutu how can that calculation be correct? What is that 500mah and 2200 mah?

To my eyes that calculation looks totally wrong

@wutu how can that calculation be correct? What is that 500mah and 2200 mah?

Conversion capacity of the battery. Exactly how he Recalculating between AA and coin cell.

To my eyes that calculation looks totally wrong

I’m not much at math, it’s quite possible. I’m not sure recalculation interval. Calculation is obviously very rough.

I would suggest something like:
http://oregonembedded.com/batterycalc.htm

@wutu @sweebee
His final calculation was based on one measurement per day, not per hour. That was a bit tricky and not completely clear. In that case your 580.8 days becomes 580.8/24= 24 days. So your 33 days was even slightly better.

But in powerdown state it also uses power. This is more important that the power it uses while measuring and transmitting.

I already know these nodemcu’s will never take as long as a mysensors device, but one of my mini pro/ds18b20/l2401 combi’s stopped after 2½ weeks while the batteries were still good. All wires are OK and I also changed the transmitter already, but it simply doesn’t function anymore.
I already mentioned earlier in one of the mysensors threads that my first pro mini was a damaged product. Maybe I have a bad batch.
Anyway: I have 4 nodemcu’s and they simply function and are also much simpler to handle. If I calculate my hours (to the hour price my company calculates) I should have completely switched to ESPs, no matter the higher energy use.
There is of course a high hobby factor involved in these kind of projects but by now I’m a bit fed up with mysensors.

I’m currently running on 3xAA batteries. I’m also thinking of running 2 3xAA battery packs in parallel to maintain the same Voltage level (otherwise the power regulator will cut it off anyway).
And next time I take Ikea’s : http://www.batteryshowdown.com/results-lo.html

I think the largest improvements from the video are:

1. Take out the LED(s)
2. In case of Lithium based cells replace the voltage regulator by a (very) low drop regulator with a very low quiescent current, i.e. the HT7333, to get the most out of your cells.
3. In case of 2 AA batteries completely lose the voltage regulator as it is not needed --> no quiescent current from the VReg.

with deep sleep the ESP should then use about 15uA.

Just encountered this. There seems to be some extra magic: “This magic sauce command, lowers the power consumption of the chip by 30% when it wakes up to perform sensor measurement.”

I assume that without the power regulator it will be much better.
But if you have a development board: How will you flash it after you have removed the power regulator? I assume the 5V USB connection uses the power regulator. And as they are soldered dev boards: how are you going to flash it again?

Or you should use one of those esps like in this post from @leader21

I found this “devkit schema” by the way: https://raw.githubusercontent.com/nodemcu/nodemcu-devkit/master/Documents/NODEMCU_DEVKIT_SCH.png

Dev boards are generally not meant to be power efficient. Use more ‘bare’ ESP variants to achieve that, if you would want a breadboard friendly package there are indeed several adapters and/or hacks available.
Flashing them is also possible with USB<–>Serial converters, a lot of them are 3V3 capable as well.