AFAIK, those things estimate charge based on voltage. If a battery heats up, it’ll have higher voltage. Not necessarily for a good reason…
Is that why they’re so inaccurate that they always die around 20-30% and never charge to 100%? I figure that phone battery meters are accurate cause they can track usage habits, but how would you do something like that with a power bank?
Yep. State of charge is almost entirely voltage based.
As a battery loses charge, the voltage sags.
What’s happening in the OP is that the batteries are getting a voltage bump, likely from the conversion to/from 5v on the output and the conversion back to battery charging voltage on the input (or the thermal/internal resistance is changing)… One of those things.
Either way, the conversions are not 100% efficient, so basically all this does is turn your battery bank into a heater, slowly sapping the power away from it as heat until dead.
With phones, it can also be battery degradation, that the voltage drops off at a higher “state of charge” level than when the battery is new.
Voltage sags can also be induced by load. If you go from a high drain state on your phone to a low drain state (say, going from playing a 3D mobile app to idling at the lock screen) the state of charge % can actually increase.
Cold temperatures can also increase the internal resistance and cause batteries that are not fully discharged to stop operating as well, only to work again after being warmed up.
Current battery tech is wild, and the state of charge indicator of voltage can be extremely inaccurate.
