Not sure what’s going on. After doing a search on this problem, the only thing I found was something in the manual indicating I was supposed to initially take the charge all the way to zero and then fully charge it.
That sounds like something we had to worry about back in the days of NiCAD batteries, but not Lithium batteries. I rarely go through the manual when I get a new device?
What could be the problem?
This:
The combination of Meshtastic device (BLE MCU + LoRa Radio), power bank and wireless charger is sufficiently complex to warrant documentation which wasn’t written because someone in the Heltec office was bored.
The most likely explanation for cycling the battery is to calibrate the fuel gauge chip. The batteries will be shipped with ~70% charge, most likely as delivered by the battery company to Heltec manufacturing, so the battery % is meaningless until the device sees how much charge a full battery takes.
Thanks for the reply. The battery came almost empty. So this is more about bad calibration of the fuel gauge (because I didn’t RTFM)? Should I then assume it is actually taking a full charge, but just not reading as such?
How can you tell? Until it’s calibrated, the unit is making a vaguely informed guess based on voltage which is based on lookup tables, many hours of R&D plus some secret sauce mostly made up of Horse Manure by PR depts.
No, this is more about NO calibration resulting in the fuel gauge chip not knowing the parameters of the battery it is monitoring. The fuel gauge doesn’t need calibrating.
You can, it’s highly likely, the drop off curve on LiPo’s makes the bottom half of the voltage measured very hard to put a % against. So as the battery would have some charge (storage is normally at 3.6V), you push some charge in, the battery gets full but the fuel gauge chip has only seen 87% and reports that. The gigantic market for battery monitoring has resulted in many low cost voltage sensing mechanisms that, well, need a bit of help getting the numbers right. Whereas more sophisticated coulomb counting gauges may report a %, but if the firmware of the device knows how big the battery is, between them they can report properly without having to empty & then fill up.
I ran extensive tests with a branded premium voltage sensing gauges along side a (different) branded premium coulomb counting gauge and an INA219 voltage/current sensing module with some solar panels this last summer. The voltage sensing unit curve was a big S - it was OK at the top & bottom of the capacity but only vaguely tracked the other sensors. They are better than nothing but they aren’t super accurate.
Given the unit is designed to charge your phone, just let it charge your phone until the MP is flat, your phone will be charged which will give you time to charge up the MP. Job done and you can move on to off-grid encrypted messaging with some reliable indication of state of charge.
Thank you so much for a very thorough explanation. I mean it. Whether or not I resolve this discrepancy, I will start by draining the battery while Qi charging my phone. I’ll then charge the Meshpocket for a long time with a fast charger.
Ultimately, I didn’t buy it to use as a battery bank so regardless of the battery level reading, I have a lovely Lora device whose charge lasts a long time.
I’m running BaseUI on it and I’m enjoying the experience.
Thanks again.
Drained it until I had to wake it from sleep multiple times just to get as far down as possible. Then plugged it in to charge. The led just now stopped flashing green and is solid green, indicating a full charge.
Reads 87%. Oh well.
Bummer.
Due to the screen not updating from the start and the stupidly small foil antenna, mine is so heavily customised that I can’t do a reset to test the out-the-box experience.