Current draw of a D1 Mini in deep sleep

To measure the deep sleep current draw of the D1 Mini, I hooked up 4 NiMH AA batteries in series (~5.2V) to its 5V and GND pins.

The sketch uploaded to the D1 Mini was a nominal:

void setup() {
  ESP.deepSleep(60)*60*1000000UL, WAKE_RF_DEFAULT);
}

void loop() {
}

The current draw was a pretty steady 0.8mA, or 800uA. That's a obviously a far cry from the sub-100uA reportedly achievable with the barebones ESP-01 due to all the extra components on the D1 Mini.

Reported deep sleep current draw for the D1 Mini is all over the place, from 0.21mA (5V), to 0.3mA (3.3V), to 6mA (USB)!

Notes:

1. Connecting a 18650 battery (~4V) to the 5V pin did not work. In theory, the MC6211 LDO used by the D1 Mini means anything higher than 3.56V should work. But when connected, the onboard LED started to flash in a slow but erratic fashion, I suspect it is randomly resetting (because each time the D1 Mini powers up, the onboard LED flashes briefly).

2. Connecting a 18650 battery to the 3.3V pin did work, though in theory it shouldn't be done because the 3.3V is not connected to any voltage regulator. So the 4V goes directly to the ESP8266, which has a theoretical upper limit of 3.6V. But it did work for me (others have reportedly connected up to 5V to the ESP8266 with no problems, but don't push your luck!), and the deep sleep current draw was ~1.5mA. I suspect if I could hook up a regulated 3.3V source to the pin, the deep sleep current draw should fall closer to the 0.8mA range.

3. Connecting 2 x alkaline AA (3.2V) to the 3.3V pin did not work. The onboard LED started to flash erratically again. In theory, this should work, since the ESP8266 has a minimum operating voltage of 2.5V, but there have been many reports of the chip being finicky with the input voltage, so who knows.

ESPCLOCK1 / ESPCLOCK2 / ESPCLOCK3 / ESPCLOCK4