I've been testing the deep sleep mode on the ESP-03. I loaded the frankenstein firmware from here https://github.com/nekromant/esp8266-frankenstein onto the module to test the deep sleep mode.
Note, you need to tie gpio15 to GND in order to boot any firmware on this module as shown in this table
|L||L||H||UART||Download code from UART|
|L||H||H||Flash||Boot from SPI Flash|
|H||x||x||SDIO||Boot from SD-card|
On the ESP-01 modules, GPIO15 must be tied to GND on the board. But on this module because the pin is broken out, you need to make sure that it's grounded at boot.
So, with the firmware running, I connect GPIO16 and RESET together and send the module into a deep sleep...
=== Press enter to activate this console ===
(c) Andrew 'Necromant' Andrianov 2014
This is free software (where possible), published under the terms of GPLv2
frankenstein > deepsleep 10000000
Frankenstein ESP8266 Firmware
Powered by Antares 0.2-rc1, Insane Mushroom
While the module is awake, current draw is round 70mA. In deep sleep this drops to 600uA which rather high (stated deep sleep current is <10uA) I try it again, and this time the current drops to 45uA, which is a lot better.
What I'd like to do, is create a battery powered sensor running of a old Nokia 890mAh 3.7v LiPo
Some quick calculations:
- Sleep current: 50uA
- Wake current: 100mA
- Wake interval: 5 minutes
- Wake time: 30 seconds
So each hour, the device would be awake 12 times an hour for 30 seconds at a time, (360 seconds or 0.1 hours) and asleep for the remainder of the time (0.9 hours). So average current consumption in any given hour is (100*0.1) + (0.05*0.9) = 10.045mA
So the 890mAh battery would only last 89 hours, or less than 4 days. That's not really good enough for a battery sensor. (Compare this to some other battery sensors that I've build using NRF24L01+ modules which get about 10 weeks life out of the same battery)
I'll do some experiments to validate my estimates (especially the awake time) to see if this can be improved.