Advanced Users can post their questions and comments here for the not so Newbie crowd.

Moderator: eriksl

User avatar
By eriksl
#91268 So finally I have started my first "mobile" project where the ESP has to be battery powered. It's a "WiFi remote control" to control lighting in my house (also controller by ESP8266).

The idea is like this:
- connect two rotary encoders to a hardware "hex" debouncer (MC14490)
- 6 (!) outputs of the debouncer go to a cmos 8-port AND gate (CD4068B) (as the idle encoders are not pulling the pullups, so input and output of the debouncer are idle high). Two spare inputs are of course tied to +.
- output of the above AND gate (normal output) goes to an 8-port OR gate (CD4078B). The output of GPIO 14 also goes to one of the OR gate inputs. Other inputs are tied to ground.
- inverted output of the OR gate goes to a one-shot 555 circuit that makes sure output is kept high for some time after a trigger (from either the rotary encoders or the GPIO).
- output of the 555 goes to CH_PD of the ESP8266.

It actually works, so that's not the question ;).

Please note all four used IC's are low-power CMOS variants, power usage is in the nano- and pico-ampere range.

Still, after the ESP8266 is powered down (CH_PD is set to low), I see for one second a power usage of 1 mA, which is not too good, but still more or less reasonable. But after a second it goes to 14 mA, which is totally unacceptable for battery operation. It stays at that usage, so it's not a capacitor charging.

For the p-o-c I am using an ESP-201, which has a power led on board. I already killed that ;). Also I set all unused GPIO to be an input, but it doesn't matter. Anyway, as I am using pullups of 10k or more, that cannot be the explanation for the high power use.

After the CH_PD has been down and it set to high again, the ESP8288 is clearly booting, so it really has been off.

The power used is measured between DC/DC converter and the general power rail, so it can't be a DC/DC converter issue.

If I disconnect CH_PD altogether, power use is 0.6 mA which I think is acceptable and also proof that it's not the other four IC's that are the problem. You'd almost suspect that CH_PD isn't drawn "low" enough and some current keeps flowing inside the chip just not enough to keep it running? Strange, I'd suspect a schmitt-trigger input there, really.

The 555 is a NXP ICM7555, CMOS version of the 555, output is push-pull so no pull-up required, so it's connected directly to the CH_PD. I think it's hard to believe that the 7555 doesn't deliver a proper drive to ground, it can drive up to something 100 mA... Also I also can hardly believe the 7555 will need to supply a high amount of current to the CH_PD pin. On my board there is nothing else connected to it, but who knows on the ESP201 board...

Might be interesting to measure the current flowing there.
User avatar
By eriksl
#91269 Current between 7555 and CH_PD is unmeasurable, in both states, so we can scratch that.

There appears to be very little documentation about this mode of operation. Most describe operation of the internal selective sleep modes (API), not the operation of the CH_PD power gating input.
User avatar
By btidey
#91283 Very strange. I'm not familiar with that particular module but it looks pretty basic with no obvious excess current draw.

I have used ESP-12 for all my battery projects including ones with rotary encoders and find that current draw of the module is of the order of 15uA with CH_PD low. That input is very high input impedance. With a low quiescent regulator I get about 30uA total draw which gives very good basic battery life.

Is the flash memory chip in standby with CS high?

When you say disconnect CH_PD do you mean pulled down to 0V?