battery drain in deep sleep

[rapid kiln]

Hello guys. I have custom split keyboard with dongle, I installed zmk-battery-center to monitor battery state.
I encountered a problem: i switch physical power button (dissconnect battery) it works as expected, no battery drain, but when the keyboard goes into deep sleep (after ~30 minutes of inactivity), I notice significant drain. For example, if I leave it overnight, I lose about 20% battery. Interestingly when I use it in the office for 8 hours it goes down only mayby 1-2%

Any ideas what might be the issue?
I'm forgetfull sometimes so I stays on over the night.

Idk if its important, but dongle stays in the dock and it stays on without anything to connect.
there is my github repo:
https://github.com/JohnnyProg/keyboard_test_config
https://github.com/JohnnyProg/keyboard_test_config/blob/master/config/boards/shields/corne/Kconfig.defconfig (not a corne keyboard, just used already existing config as startpoint)

[wooden cipher]

if you built with ["superminis"](#hardware message), you should [measure their pull-up resistors](#1157825408130101349 message).

No idea what hardware this actually is since all you've said is that it's not a Corne, but at a glance it doesn't look like it has any LEDs, displays, or encoders

[rapid kiln]

it doesnt have any leds, displays or encoders. It is ali express version of nice!nano v2

[wooden cipher]

then yeah, commonly recognized as a supermini
you need to measure its pull-up resistor

[rapid kiln]

ok, thank you. I wont be able to do it, I don't have small enough probes in my multimeter to do it. Could you please explain why this resistor would cause leak only during sleep (over night) and not (much smaller) during use?

during use (8hr) battery goes down 1-2%, but overnight while sleep (16hr) almost 20%

[wooden cipher]

deep sleep turns VCC off: https://zmk.dev/docs/features/low-power-states#deep-sleep
(for why VCC being off would result in higher power draw, see previous message and read links)
no, I don't know why VCC would be on given you don't have peripherals which require it—but I don't have a full history of your hardware/firmware
perhaps it's a side effect of repurposing corne code

...that said, I also can't rule out [random jank](#hardware message) since the hardware is inconsistent and unreliable
(note the huge variation in idle power consumption; stroma didn't see such a big change with deep sleep—but if idle can vary that much, deep sleep can too)

if you can't deal with the resistor (though [you can just crush/scrape it with a small screwdriver, no need to be too fussy about it](#1157825408130101349 message)) I'd probably turn off deep sleep

just because that seems to be working better for you runtime-wise