ESP32 Devkit with LiPo battery triggers brownout

[spystath]

Greetings,
I'm trying to explore powering the ESP32 Devkit (WROOM-32) with a battery through the 3V3 connector, hoping to bypass the power-hungry AMS1117. To do so I use a 2Ah LiPo battery at 3.7-4.2V through a voltage regulator. I've tried both a Torex XC6210B332MR (700 mA - ~100 mV dropout) and a Microchip MCP1725 (500 mA - ~200 mV dropout). The XC6210B332MR is coupled with 1μF input/output capacitors and the MCP1725 with 1μF/4.7μF as per the respective datasheets. Voltage is indeed 3.3V on the Vout. However I'm facing a brownout issue just before the radios are turned on and the Power ON LED seems to be dimly lit when compared to the 5V USB input. Reading through the and the documentation of the Devkit it seems that the board can be powered by the 3V3 pin and to my arguably inexperienced eyes the schematic seems to hint that the 3V3 should bypass the existing voltage regulator (and the LED I suppose?).

Is there something wrong with my setup or is this a limitation of the Devkit board? My program is fairly standard; connect to wifi, fill a struct with data; send a UDP packet out of it; sleep; repeat. I'm mostly a software guy so apologies if I'm missing something obvious wrt circuitry. Measuring with an inexpensive multimeter current draw seems to be in the 70ish mA range before the brownout is triggered. Program works as expected when connected through 5V USB.

Thanks in advance!!

[ESP_Sprite]

The power LED lighting dimly is normal when you're feeding the unit from the 3.3V input: the LED is connected to the 5V and probably lights up a little bit because of reverse leakage through e.g. the AMS1117 or something.

The brown-out is not, however... What does your physical setup look like? If you e.g. have too long or too thin wires running from your LDO to the module, it can have too high a resistance or impedance to get power to the ESP32 quickly enough when it does its RF calibration.

[spystath]

Hi! Thanks for replying. This is how things are wired up in the breadboard. I use single core 22 AWG wires, roughly 5 cm long, directly plugged into the breadboard vertical power rails for all power delivery (battery to breadboard to voltage regulator; voltage regulator to ESP32).

[ESP_Sprite]

If anything, the fact that you use a breadboard may be an issue here; at least in my opinion, these things can have a pretty high contact resistance. Quick and dirty fix may or may not be to plunk a big cap over the power rails for prototyping.

[spystath]

Indeed a 10uF capacitor on the power rails fixed the issue. I'm seeing a slight dip in voltage ~100 mV when the radio turns on. Is that normal? The firmware behaves as expected.

[ESP_Sprite]

I'd think that's normal. During RF cal, the ESP32 can use up to 500mA; it's pretty easy to have assorted parasitic resistances and impedances have the rail drop by 100mA at that.