Battery Power for EzSBC-ESP32-01

[MicroTinkerer]

I was delighted to test your EzSBC-ESP32-01 board and achieve a deep sleep current of only 13 uA -- and it's made in the USA! My application spends most of the time in deep sleep until a reed switch wakes it up to send an ESP-NOW message to another ESP32, then goes back to sleep. With your board, it could run as long as seven years on one set of AAA batteries, ignoring the effects of outside temperatures and self-discharge. I am now working out how to power the device and could use some advice.

The device works fine with three AAA Ultimate Lithium batteries in series (4.0 - 5.2V), but I read that only two such batteries in series (2.7 - 3.5V) are necessary for the ESP32, and the extra power from the third battery is just wasted as heat by the LDO regulator. I tried two batteries with the ESP32-01 board and it did not work consistently, so I decided to do some testing with my bench power supply.

For Vin = 4.0 - 4.5V the device worked as expected.
For Vin = 3.5 - 3.9V the processor wakes up, does something, and goes back to sleep, but the ESP-NOW message is not received.
For Vin = 2.8 - 3.4V the processor wakes up, but never goes back to sleep, and no ESP-NOW message is received.

Does this sound right, or am I doing something wrong? Yes, I can just use three batteries for this application, but are my power options really that limited? If this is right, then I can't use a lithium-ion or lithium-polymer battery to power the board, for example. For comparison, my Lolin D32 boards run fine with a lithium-polymer battery (3.7V), but not consistently with just two AAA cells (3.0V).

I'm not an engineer, just a hobbyist, but I am eager to learn all that I can about processors and IoT. Any advice you have to offer on powering the ESP32-01 board with batteries would be greatly appreciated.

[Daniel]

Your results are different from what I got when I tested low voltage operation. It may come down to something easy like long wires between the PSU and the module. These modules draw fast, large current spikes and unless the wires are pretty short you may get enough dip to trigger the brown out behavior of the Wroom-32.

A more likely issue is that the input voltage does something unexpected. There are two possibilities. If you are powering from an external power supply the current spike of the ESP32 when it wakes up can upset the regulation loop of the power supply leading to a too low input voltage. With cells it is possible that the cell voltage drops abruptly when loaded after a period of low load. The module has quite a bit of capacitance on the input, 47uF, but it may not be enough to hold up the voltage.

Below 3.6V you are in the region where the onboard LDO enters into the dropout region with loads above 100mA and I found that the behavior of the module depended on the program that was executing on the ESP32.

[MicroTinkerer]

Thank you, Daniel, for your reply. It is helpful to know that the LDO enters the dropout region below 3.6 V, since my app does briefly spike above 100 mA. I'll stop trying to over-optimize the battery configuration and stick with three cells. That has tested well and should work for the useful life of the batteries.