Measuring power consumption

[Atlask1]

Hi,
I've got an esp32 module soldered on my pcb, with a minimal circuit and with 2 transistor for auto reset .
I'm under Arduino .
The sketch put esp32 in deep sleep and there's a wake with 2 ext interrupt + timer wake every minutes.
I'm trying to measure the power consumption and also if I'm happy for results, there's some things that I don't understand.
First test :
Device powered from 3.3V of external USB/TTL CP2102, with serial connection and RTS/DTR, digital multimeter between 3.3V - CP2102 and VCC of device. During wake, without WiFi connection I measure 20-24 mA and during sleep I have ~3 uA.
Second test:
Device powered from 3.6V of Saft 14500 battery, not rechargeable, 2500mA, with serial connection, GND and RTS/DTR, digital multimeter between 3.6V - battery and VCC of device. During wake, without WiFi connection I measure the same 20-24 mA and during sleep I have 4-5 uA.
Is always a great result but I don't understand why I absorbe less with a USB/TTL connected to USB of PC .
Probably is my misunderstanding and I know that isn't so important 1 or 2 uA with this kind of results , but I want to understand why.
There's someone can help me understand?
thanks in advance

[boarchuz]

What is the actual voltage on the battery, under load?

[Atlask1]

Actual voltage is 3528 mV. But is a particular battery, you don’t see a realistic voltage decrease, you read a voltage around 3.6V until battery id near the end, after that the voltage drop down ( you have some oscillation of reading).
I’m also trying to discover a system to control the state of this battery when is powering the esp32. Now it’s empiric and I’m trying to consider time and mA or uA during sleep / wake / wifi connection/ http post .

[HermannSW]

That is an interesting battery, exactly delivering the absolute maximal rating input voltage to the ESP32:
https://www.espressif.com/sites/default ... df#page=40

I had to use step-down converter for 3.7V lipos (they show typically even 4.2V when fully loaded).

[Atlask1]

That is an interesting battery, exactly delivering the absolute maximal rating input voltage to the ESP32

Yes it is !
If you look it’s datasheet you will see the behaviour and as I have understand, it give 3.6V until the end of his life.
so I can't rely on the remaining mV, but I have to necessarily consider the mA , time and finally mAh.
My sketch do always the same things: stay in sleep for 1 hour waiting for an ext interrupt ( or for the timer), if interrupt or timer wake the device, do wifi connection and send a httpPOST to a server, then return in deep-sleep.
I’m trying to measure time and mA of wake, connection and post phase for estimate of consumption and approximate battery life.
measuring the current with the tester I found the behavior in question.

[Atlask1]

unfortunately these batteries do not allow me to measure the voltage and understand the state of charge, I should measure the current.
I was thinking of verifying the time spent in each operational phase (wake, connection, transmission) and adding the consumptions or an average of the consumptions measured with the multimeter, so as to have approximately the total mA consumed and compare them with the declared 2500mA of the battery.
Does this make sense? can it be corrected? any other idea?

what confuses me the most is the correct calculation of the final value, I don't understand if it is better to bring everything in A or mA and the time from micro sec to minutes or hours, being the consumption measured in Ah.