Let's Control It

I once made a battery powered sensor using a wemos d1 mini clone.
With no modification it was drawing around 300uA of power while in deep sleep which is ok but not good enough.
With the proper modifications which included lifting the ground pin of the serial chip and a changing the ldo regulator i got the consumption down do around 50uA which is good enough for me.

Two days ago i found myself in need of a water leakage sensor. I could have simply bought one but since i found this instruction https://esp8266--server-de.translate.go ... r_pto=wapp i thought this seems easy enough since i already have all the parts i need laying around.
A quick and dirty setup and some measurements later i realized that even with the aforementioned modifications i couldn´t get lower than 1,8mA
I spend the whole yesterday to figure out what was wrong. And i found out that some wemos d1 minis without mods got down to 300uA in deep sleep and some go only down to 1,8mA.
One night sleeping over it brought me to the right path:
The serial chip!
Both type of boards have slightly different layout but the serial chip is labeled the same: CH340C
The chips are the same but the connections differ? That means one of the chip is not an CH340C since one draws significantly more power than the other (300uA vs 1,8mA!)
A continuity test revealed the ground legs of the sketchy chip, which i desoldered. New measurement: 60uA.
I hope that revelation helps some people that stumble upon the same issue.


In the picture you can see on the left the standard clone and on the right the clone with the strange serial chip.
The red circles indicate the ground pins wich need to be desoldered/lifted. And here the quick and dirty water sensor which sends a notification with the battery status to ntfy and activates the buzzer for audible feedback as soon as the contacts are submerged in water:

Yep, external components are easy to overlook.

I'm also looking into the differences between pull-up and -down pins as this has been mentioned a few times deep in some datasheets from Espressif (mainly regarding ESP32 families).
You can sometimes set some GPIO pins to a specific state when going to deepsleep to make sure the power draw is minimal.
And on the other hand, you can also lock pins to make sure their state doesn't change, even during reboot/crash/deepsleep.
This can be especially useful for operating some switches like relais. But I was also thinking about some circuit where the ESP may control some FET to either keep it powered during deep sleep, or let the ESP decide to turn it off completely and let this fet also be switched by some external trigger to turn it all on.
This way you can get really low power consumption. (still a work-in-progress though)

TD-er wrote: ↑30 May 2023, 14:04 Yep, external components are easy to overlook.

The problem with these "fake"/different serial chips seem to be their inability to enter a "sleep mode" when not in use

The chips are the same but the connections differ? That means one of the chip is not an CH340C since one draws significantly more power than the other (300uA vs 1,8mA!)

Good find!

- Thomas