As your battery looks (not that I’m some expert) like slightly degraded there, please ensure that some power supply/bank is still connected (you’ll be there where you want to be when state outputs fully-charged, as my last two screenshots above show) and:
BQ25890 works by itself during reflash process, under Jumpdrive … either when connected to some computer USB port and where BPP-L503 battery cell actually discharges itself (at slower rate, yet discharging while needs to add above 0.47A necessary power that i.MX8 MQ and its periphery are requesting in order to successfully complete the job/task):
or even when actually charging this Li-ion battery as connected to adequate power supply/bank/dock, isn’t it?
And yes, I already gave + to @dos post here for some reason:
This very kind advice applies here as well (providing correct answer), I’m quite sure:
If you use the provided PD charger or an equivalent which is able to supply 3A, I wouldn’t bother with turning the phone off during charging since it’s unlikely that you be hitting that limit, at least as long as you won’t let it pause the charging due to high temperature.
Well, unless you do something like docking it to an external screen to play an online 3D game that uses a camera using cellular connection in a place with poor reception, then the charging will likely slow down due to power draw:)
I thought that this thread applies to the recorded value/output or rather artificial number (and actual main issue/behavior of inserted battery here / of some wrong charging setups on some phones / when those run only on its battery) when energy-full at 26.3287 Wh (like my screenshot shows) or even at:
Difference between energy-full and energy-full-design. Basically what the gauge thinks is the actual battery capacity in relation to stated design capacity. energy-full gets updated after every full charge cycle.
As I said above, that value is determined by gauge’s heuristics and is updated at the end of charging cycle, so if your gauge is still calibrating it will likely fluctuate a bit. That said, this is simply a not-100%-accurate representation of the actual physical property, and that physical property tends to only go down over time with battery age. Seems like your gauge simply took a pessimistic stance at first ;]
When the L5 is fully charged from the original power supply and the red LED goes off, I power it down by SSH with sudo systemctl poweroff and without disconnecting the power supply from the L5. As you can see in the photo below, it keeps charging with 5.231V and 0.087A which is nearly 0.5W. A reboot after half hour shows now:
energy-full: 12.9961 Wh
energy-full-design: 13.8572 Wh
energy-rate: 0.705233 W
voltage: 4.20969 V
percentage: 100%
temperature: 31.2 degrees C
capacity: 93.7143%
Great job done there (please do not forget to mark your last post here as solution)!
Also (IMO), there is no charging of battery going on after red LED light off (although around 0.9A shown, in range from 5.00V up to 5.30V relates to which power supply used there), yet repeating (up to your description) two or three times sudo systemctl poweroff and power on over power button (power supply still connected) will get your inserted battery to 100% (I’m quite sure, just follow when red LED light present, leave Librem 5 charging, either with OS showing 100% yet red LED light still … or when phone powered of). Thanks for your post as now I’m quite sure (as well) that actually repeated boot into this great Linux Kernel adjusts what you were into in this thread:
PureOS Kernel is not only supposed to read actual BPP-L503 battery cell voltage/capacity, this Linux Kernel indeed “configures the gauge with correct values” or reads exact stage (capacity number for example) from within Librem 5 belonging battery, from the very precise bq25890-charger connected to the inserted Librem 5 battery (even when brand new fully charged to exactly 4.20V or properly serviced one just inserted) during booting phase into PureOS GUI. IMHO, here related Linux Kernel does what supposed although other HW related questions being quite complex to be understood right out-of-the-box as self-explanatory (for any end-user):