Bangle.js 2 - high idle drain on stock setup (2v29), is this normal or a faulty cell?

[spkocz]

New Bangle.js 2, about a week old. I'm seeing what looks like excessive idle drain and I've run out of software explanations, so I'd appreciate a sanity check before I go back to the retailer.

Setup

• Firmware 2v29 (DFU 2v29) — latest release
• Factory reset: "Install default apps" + "Reset Settings", then the 11 pending app updates applied
• Default apps only, nothing custom installed
• Bluetooth OFF (via Settings menu) for all idle measurements
• Wake on button only — twist/wake-on-movement not available/enabled
• HRM interval: off
• Watch lying flat on a desk, not worn, not moved

Measurements

I'm using raw analogRead(D3) rather than percentages, since the percentage readout jumps around with instantaneous load. Approximate voltage = raw / 0.314 * 4.2.

Time	analogRead(D3)	E.getBattery()	~V	Notes
T0	0.31421	100%	4.20	full charge, left on charger well past first 100%
+8h	0.30591	91%	4.09	idle, BLE off
+20h	0.29858	83%	3.99	11 app updates over BLE, after that: idle, BLE off
+30h	0.29321	69%	3.92	idle overnight, BLE off

So roughly 0.021 raw consumed in ~3 days, from a usable span of about 0.074 (0.246–0.32). The last clean idle window is 0.29858 → 0.29321 = 0.0054 raw in ~11 hours, with radio off, on the flat part of the discharge curve.

Power Manager log (same window, 10h49m57s)

Deferred function calls
Timeouts/intervals ran for 43.86s (0.12%)
  39.05s  89.04%  function () {drawTimeout=undefined;draw();}
   4.67s  10.65%  function () {return WIDGETS["bat"].draw()}
   0.13s   0.30%  function () { [native code] }

Hardware power
   8.87s   0.03%  LCD

CPU wakeups look healthy to me: 0.12% of the time in timeouts, 0.03% LCD on, no runaway function (nothing like the 99%-of-calls case in discussion #5977). The clock redraw at ~39s over 11h is consistent with a once-per-minute redraw.

Question

With the JS side apparently idle and the radio off, is this drain rate within normal range for a 2v29 stock setup, or does it point at the cell / a leak? For reference, HughB reported 37 days idle from a single charge on a comparable configuration (~2.5%/day) in discussion #6500, which is roughly an order of magnitude better than what I'm seeing.

Is there anything else worth measuring before I treat this as a hardware fault? Note that Power Manager's hardware log only lists LCD — I assume softdevice/BLE draw wouldn't show up there at all, so if there's a way to rule out radio-side or leakage current from the JS side, I'd like to try it.

Also: charging pins are unmodified (not taped/lacquered) — happy to test that if it's a plausible leakage path.

Thanks.

[gfwilliams]

Hi - sorry to hear this. So have you run it down until it turns off at all? How long did it take?

Also I should add that the power manager itself really hurts battery life because it's waking up every so often in order to check JS execution state - so you shouldn't be concerned about high battery usage if you have that installed.

And just to check - do you leave it to charge over night? Because the battery percentage meter uses voltage, and the voltage rises above normal when the watch is charging, the battery percent meter will show 100% when on charge long before the battery is actually fully charged. So if you take the watch off charge as soon as it says 100% you won't be getting anywhere near a full charge in it.

The raw readings are a bit difficult, because the discharge curve of the lipo is very nonlinear:

If you're looking at voltage drop (eg the analogRead value) you're going to see it fall off a cliff right after it comes off the charger, and then it should drop much more gradually as you're seeing at the 20-30h mark.

charging pins are unmodified (not taped/lacquered)

You mean you received the watch without tape on the contacts? Where did you purchase it from?

HughB's 37 days is likely with the watch not worn much (it can drop into a lower power mode for the accelerometer when it doesn't detect movement) but even so you'd hope to get at least 20 days off it being worn with default apps (not with power manager installed though).

[spkocz]

Thanks for the quick reply - and good to know about Power Manager itself. That undermines my measurements, since it was installed and logging during what I thought was my cleanest idle window. I'll remove it before running the next test.

Answers to your questions:

Run down to shutdown: No, I haven't. I've only measured partial windows so far, so I have no full-cycle figure. That's the test I'll run next.

Charging: Charged for about 3 hours in one go. analogRead(D3) reached 0.31421 (~4.2V) and stopped rising, so I think it really was full rather than an early false 100% - but I take the point that the percentage meter can't be trusted for this.

Charging pins: My mistake in the original post - I checked again and the two middle pins are covered. Only the outer charging contacts are exposed, as they should be. So no issue there.

Retailer: BerryBase (Germany).

Plan for the next test:

• Remove Power Manager
• Default apps only, plus the percentage battery widget so I can read the level without connecting
• Full charge (confirmed by analogRead(D3) plateauing rather than by the 100% indication)
• Then run it down to shutdown, left idle on a desk

I'll report the actual runtime rather than extrapolating from partial windows. Thanks for the pointers.

[enaon]

you can use D25 to determine when charging is done.

I also use D3 to get battery state, the graph is quite linear, not like the image above implies.

when left idle on the desk consumption is very low, I get over 40 days, as Gordon said the accelerometer goes to sleep mode when no movement is detected, so this is not a fair test, and it will take too long. :)
The BangleJs2 idles at ~150μΑ if not moved, the battery is ~180mAh, ~50 days when not moved at all.

just do an D3 read test for a day or so, it should stay above 4.15 volts for 3-4 days when idle. Do this on the on the "clean boot" mode (press and hold the button till you get an options menu).

do a

let hexString = ("0x" + (0x50000700 + (D3 * 4)).toString(16));
poke32(hexString, 2);

to release D3 after reading the voltage, else it draws 70μΑ.

I'm using raw analogRead(D3) rather than percentages, since the percentage readout jumps around with instantaneous load. Approximate voltage = raw / 0.314 * 4.2.

Also, it could be that your charger shuts down when too little of a load is detected, 0.314 is not 4.2 volts, ~0.319 is 4.2 as far as I know. I use this: 13.245 * analogRead(D3). It is best to use D25 to determine if charge is really done, it will go to false when done.

but most importand, this is strange. E.getBattery() may go one or two numbers up or down when the screen is on/off, this is ~15mA, but that is it. I have a torch on mine, 2 led's for a total of 60mA, toggling the leds does not make E.getBattery() fluctuate more than 2-3 percent. If you really get values all over the place, something is not right.

[gfwilliams]

Thanks for the info! Obviously if there does turn out to be some issue we'll get it sorted. With Berrybase it's probably better that you go through us for a replacement anyway.

Testing as you suggest sounds great. But rather than checking on the charge state or voltage - maybe just leave the watch charging over night? As @enaon says it'd be good to make sure you have one that doesn't shut down early though just in case. Checking for constant voltage isn't enough:

The charge process is in two main phases - constant current and constant voltage. So if you disconnect the charger as soon as the voltage is flat, you're not really fully charging your battery. Although 3 hours is probably ok, especially with a new watch it'd be best to make sure you charge it overnight to make sure the battery really does get fully charged.

Also just so you know, even without the Power Manager app the default Health app logs the battery level every 10 minutes, so that can still give you a pretty good idea of discharge. Just go to https://banglejs.com/apps/?id=health and click the floppy disk icon when connected to the Bangle, then Download CSV and you'll get a column for battery percentage.

[enaon]

The charge process is in two main phases - constant current and constant voltage. So if you disconnect the charger as soon as the voltage is flat, you're not really fully charging your battery.

I think going over 4.15 is not important. The last part of the charging process is crucial, but only when more than one cells are connected in series. This last stage is the balance stage, but only makes sense when balancing is needed, capacity wise it is irrelevant, and it does not really help the battery longevity, is anything it is bad for the battery when it is one cell only.

with an idle draw of ~150μΑ, the bangle should stay on 4.15 for a day, if it does not, the battery has a problem I think.

[spkocz]

Thanks, and good to know about the Health app logging battery every 10 minutes, I'll use that if I can.

On the charging: I didn't disconnect as soon as the voltage went flat. Following @enaon's tip I watched digitalRead(new Pin(25)) and left it on the charger until that flipped from 0 to 1, then kept it charging for roughly 3 more hours after that. So it should have had the full CV phase, not just the CC part. (Incidentally D25 isn't defined as a global pin object on my build - analogRead(D25) returns NaN - so new Pin(25) is needed. Might be worth noting for anyone else trying this.)

It's a fairly basic charger (without FastCharge nor PD), but the last reading before I unplugged was 4.178V with D25=1.

After that: clean boot (button held to the options menu), Bluetooth off, and it's been sitting untouched on a desk since yesterday 22:00.

Current state, purely from the battery percentage widget so I don't have to connect:

• 22.07; 22:00 - 100%
• 23:45 (1h45m) - 99%
• 23.07; 09:00 (11h) - 91%
• 12:00 (14h) - 90%

I'm not reading much into those yet given the discharge curve is steepest right after charging, the interesting part is what the next couple of days look like.

Plan: leave it completely alone for a full 24h, then take one reading, then another 24h, and so on. I'd rather not restart the test from scratch, so I'll try the Health CSV export first - though I'm not sure whether Health is actually logging after a clean boot, since that skips app boot code. If it isn't, I'll fall back to connecting over BLE and reading D3 directly, making sure to release the pin afterwards:

var v=analogRead(D3); poke32(0x50000700+3*4,2);

(That 70μA leak from not releasing D3 was almost certainly polluting my earlier measurements - along with having Power Manager installed - so thanks to @enaon for pointing both out.)

I'll report back with the numbers as they come in.

[enaon]

It's a fairly basic charger (without FastCharge nor PD), but the last reading before I unplugged was 4.178V with D25=1.

D25.read() should return false when charging is stoped on the ic level.

This is not importand capacity wise, but so to be clear: D25.read() will go false, when the watch is plugged in, the charger is active, and the watch it self stops the charging process. If charging seems to have ended, but D25.read() is true, the charger ended the process, not the watch.

the reason analogRead(D25) returns nan, is that D25 is not an analogue pin. Do a D25.read(), or a digitalRead(D25) instead

After that: clean boot (button held to the options menu), Bluetooth off, and it's been sitting untouched on a desk since yesterday 22:00.

Current state, purely from the battery percentage widget so I don't have to connect:

• 22.07; 22:00 - 100%
• 23:45 (1h45m) - 99%
• 23.07; 09:00 (11h) - 91%
• 12:00 (14h) - 90%

nice, my take is that your battery is not ok, is should stay on 100%, over 4.15 volts, for a day, this is the case on my watches.

I'm not reading much into those yet given the discharge curve is steepest right after charging,

I think you should, this is the point of the test. Over 4.15 the capacity is minimal, very small, so it is very easy to see battery health. With a load of ~150μΑ, a healthy battery will stay over 4.15 for a a day at least. If not, the load is greater than normal, or the battery is not healthy.

just so that you may have a base point, I did charge a bangle yesterday right after my last post, left it till it went up to 4.20 and D25 reported false, it now repots 4.19 volts, BT is enabled with 1500 advertising interval, this adds ~25μΑ, so not really important.

this question will follow I think, so am asking it :)
have you maybe placed the bangle under water, could it be that it got wet?

[enaon]

and one more thing :)

After that: clean boot (button held to the options menu), Bluetooth off, and it's been sitting untouched on a desk since yesterday 22:00.

Current state, purely from the battery percentage widget so I don't have to connect:

this makes no sense to me, how could you get the battery widget if you are really on clean boot mode? Clean boot means that nothing bangle related is running, i suggested it so that you may rule out apps related issues. On clean boot you will not get any widgets running.

the idea is: Fully charge, or at least over 4.15volts, enable clean boot, let it be for 12 hours or a day, then connect frmo the ide and do a D3 read. If it is not over 4.15, the battery is not healthy, or the bangle has a load greater than normal, maybe it got wet.

leave the BT enabled so that you can connect when you need to from the ide, when not connected the draw is very small, on defaullt advertising interval it is still less than 60μΑ, it will not mess up your testing.

[enaon]

and a side note, if one cares about a simple testing of the battery, or in general does not mind spending 12 euros, this tester is excellent.

https://www.aliexpress.com/item/1005002287714032.html

it will stop chargers from auto shutting down, as it presents enouph load to the charger to keep it alive, it will monitor capacity going to the bangle during charge, it has BT, and I do have an app for connecting it to the bangle if one cares.

[spkocz]

You're right, the fact that I could read the battery widget at all means I was not in clean boot. I most likely tapped Reboot, which sits right next to Clean Boot in that menu. To be honest I didn't know what Clean Boot actually looks like when it's active, so I had no reason to suspect anything was wrong. This is my first Bangle.

Also, correcting something I posted earlier: D25 is defined - the NaN was only because analogRead() won't work on a digital pin. D25.read() works fine, so my new Pin(25) was an unnecessary workaround for a non-problem. Sorry for the noise.

On the water question: no, never. It's never been in the shower, a pool, or rain. It's been on a desk or on my wrist indoors since I unboxed it a week ago. The two middle charging pins came covered from the factory and I haven't touched them, and the outer contacts look clean. So I don't think there's a moisture leakage path, though obviously I can't rule out something internal.

One thing I'd like your read on, because my watch seems to behave the opposite way round to your description. Right now it's sitting on the charger and I have a USB meter (FNIRSI FNB58) inline. It reads 0.024A flowing into the watch - so it's definitely still charging, that's far above the ~150μA idle draw. At that exact moment D25.read() returns false.

So on my unit, false appears to mean "charging in progress" and true "charging finished" — which is the reverse of what you described. That also makes my earlier reading consistent: I got D25 = 1 (true) at 4.178V at the end of the previous charge, which under this polarity simply means the charge had completed, rather than my charger having given up.

Am I misreading something, or does the polarity vary between units/firmware? I'd rather get this straight before I use it as a stop condition, since "charge until D25 goes false" would mean "charge until it starts charging" on mine.

For now I'll use the USB meter as the criterion instead, waiting until the current drops to essentially nothing, since actual current flow seems less ambiguous than either the voltage plateau or the pin state.

Revised plan:

• Charge until the USB meter shows the current has tailed off to ~0, and note what D25.read() says at that point
• Clean boot - and this time verify it actually took (blank screen, no clock, no widgets)
• Bluetooth left on, as you suggest, since sub-60μA won't distort the result and I can't read anything without it
• Leave untouched for 24h, then connect and read D3, releasing the pin afterwards

I'll report the voltage against your 4.15V threshold. Thanks for the patience - both of you have caught several things I got wrong.

[enaon]

One thing I'd like your read on, because my watch seems to behave the opposite way round to your description. Right now it's sitting on the charger and I have a USB meter (FNIRSI FNB58) inline. It reads 0.024A flowing into the watch - so it's definitely still charging, that's far above the ~150μA idle draw. At that exact moment D25.read() returns false.

So on my unit, false appears to mean "charging in progress" and true "charging finished" — which is the reverse of what you described. That also makes my earlier reading consistent: I got D25 = 1 (true) at 4.178V at the end of the previous charge, which under this polarity simply means the charge had completed, rather than my charger having given up.

Am I misreading something, or does the polarity vary between units/firmware? I'd rather get this straight before I use it as a stop condition, since "charge until D25 goes false" would mean "charge until it starts charging" on mine.

sorry about that, I didn't check very carefully, I have it in my code and read it the wrong way, you are correct, D25 will go hi once the ic on the watch stops the charging process, so what you get is correct, sorry again.

	Bangle.on('charging', function(charger) {
		let s={};
		let r=D25.read();
		if(charger) {
			s.state=r?false:true;
			s.full=r;
		}else{
			s.state=false;
			s.full=false;
		}
		ew.is.ondc=s.state;
		ew.sys.emit('charge', s);
		let hexString = ("0x" + (0x50000700 + (D25 * 4)).toString(16));
		poke32(hexString, 2);
		
	});

Right now it's sitting on the charger and I have a USB meter (FNIRSI FNB58) inline

nice, this way you will make sure the charger will not shut down, so full charge is settled. It is not very important I think, anything over 4.15 will do, but this is even better.

I will report the charge state on my test bangle when you report yours, so we can compare.

[spkocz]

Rerun done properly this time. Result is well below your threshold.

Setup

• Charged until D25.read() returned true, confirmed with the USB meter showing the current had tailed off
• Clean boot - verified it actually took this time
• Bluetooth left on, as suggested
• No Power Manager, no other apps running
• Watch untouched on a desk for the full 24h
• D3 released with poke32 after reading

Readings

Time	analogRead(D3)	~V
23.07 17:00 (start)	0.31616	4.184
24.07 17:00 (+24h)	0.29761	3.958

So 0.226V lost in 24 hours, idle, from a confirmed full charge. Against your "should still be above 4.15V after a day" criterion - and your own watch reading 4.19V after the same period - this isn't close.

At this point I've run out of software explanations - everything that was polluting the earlier measurements (Power Manager, the unreleased D3, the failed clean boot) has been eliminated and the result didn't improve. So, would you say this is enough to call it a hardware fault?

[gfwilliams]

The voltage reported comes from a potential divider which varies slightly between watches, so I'm not sure you can infer from a single day of discharge that there is a problem (there's actually a calibration option in settings)

Also lipo batteries need a few cycles with full charge from new to get to their full capacity, and that hasn't been done yet.

As I understand you've still never actually run the battery flat? If the bangle is fully charged, on default software and runs flat after a week then yes something is wrong, but right now we don't know that.

Please can you at least give it a few charge cycles first and see how long it takes to run flat before we start jumping to conclusions?

[spkocz]

That's fair, and thanks for the reality check. So I'll stop jumping to conclusions and do this:

• Just start using it normally on the default clock face (GPS off, since that's the biggest single draw and would only add noise)
• Give it 2–3 proper full charge cycles to let a new cell form - charging each time until the current has actually tailed off (D25 true / meter near zero), not to the first 100%
• Run the battery calibration in Settings once it's fully charged, to deal with the potential-divider variation you mentioned
• Then, once it's had those cycles, run it down to shutdown and report how many days it actually lasts

I'll come back with a real runtime figure rather than voltage extrapolations. Might be a couple of weeks before I have that. Thanks again for the patience, and to @enaon too.

[enaon]

The voltage reported comes from a potential divider which varies slightly between watches, so I'm not sure you can infer from a single day of discharge that there is a problem (there's actually a calibration option in settings)

Your point is valid when comparing readings between different watches without a known base reading. But on the same watch, if the resistors on the divider are a bit off, they will be off both on 4.18 and 3.96, they cannot be off in one the two readings. 4.18 is normal for full charge, there is no ground to doubt 3.96.

Also lipo batteries need a few cycles with full charge from new to get to their full capacity, and that hasn't been done yet.

lipo cells do not need charging cycles to reach full capacity. It makes some sense if a gauge meter ic is present, but only as basic consumer info. On the cell level this is a myth, not technical info.

The thing a new battery may have is a higher internal resistance, they do that to prolong shelf life, but it only affects instant readings under load, not stabilized readings. If the battery reads 3.96 at rest the energy is lost, internal resistance or charge cycles has nothing to do with it.

Your code assumes that 20% of 180mAh is on the 3.95-4.2 range, 36mAh that is. I do not agree, I measured 45mAh on the same range on a latest batch bangle from berrybase, but say 36mAh is correct. This is 10 times more that it should be when on clean mode and the watch is still, 1.5mA instead of 150μΑ over a period of 24 hours.

As I understand you've still never actually run the battery flat? If the bangle is fully charged, on default software and runs flat after a week then yes something is wrong, but right now we don't know that.

Please can you at least give it a few charge cycles first and see how long it takes to run flat before we start jumping to conclusions?

I understand that you need to filter out false claims somehow, but he has an excellent usb meter, he could be asked to measure capacity intake from 3.96 to 4.18 and be done with it.

In any case, the test I proposed is not in the "jumping to conclusions" realm.

[gfwilliams]

Thanks @spkocz!

I'm assuming you probably started looking into the battery usage because it felt like you had high battery drain when you got the watch, and so it could well be a hardware issue.

There's no rush though - if there is a problem we'll definitely make sure you get a new Bangle. It'll just be a pain for all involved if we get you a new one and it behaves exactly like the old one!

@enaon interesting about the cell performance - I have always assumed that there was a reason that most manufacturers ask you to fully-charge products before using them, but it does appear that the initial charge step should be being done at the battery factory. I don't think you're right about constant voltage charging not adding anything though - Googling it, it looks like the constant voltage step could add 15-30% more charge into the battery, and this is exactly at the point where we saw the high voltage drop in the first day.

Your suggestions are really helpful, but it just feels very counter-intuitive to write off a device before it has even been fully discharged. The cost to me of replacing a watch is pretty high, so I want to make sure I get it right!

[enaon]

I don't think you're right about constant voltage charging not adding anything though - Googling it, it looks like the constant voltage step could add 15-30% more charge into the battery, and this is exactly at the point where we saw the high voltage drop in the first day.

I think there is a misunderstanding here regarding what "CV stage" refers to. The chat is talking about voltage readings with the charger connected. It maybe that it assumes the user does not differentiate between rest voltage and voltage under load, it assumes that once the charger is disconnected the voltage will drop rapidly to the 4volts range. The 15-30% you are referring to, resides in the 4.0-4.20 range. Over 4.15 for the batteries like the ones on the bangle, a 2-5% exists, it is the perfect spot to detect hi idle consumption, it is the reason I proposed this test.

so to be clear, the CV stage starts at ~4 volts from a cell perspective. The reason the chat says CV starts at ~4.20, is that it assumes a charger is connected, it is referring to the Constant Voltage(CV) the charger is outputting, not battery voltage at rest. I did not want to say that the CV stage is irrelevant, I meant to say that one can terminate the CV stage at 4.15 (battery voltage at rest) with no real capacity lost.

The start point of the CV stage is actually pressure related, it can start from 3.95 to 4.10 battery voltage, and it is constant, 4.20 in our case, but this is a bit out of scope. The important thing is the user had concluded the CV stage already in any matter that makes a difference, there is no doubt about that.

Your suggestions are really helpful, but it just feels very counter-intuitive to write off a device before it has even been fully discharged. The cost to me of replacing a watch is pretty high, so I want to make sure I get it right!

I really respect the engineer in you, I understand you sometimes need to be a sales person too, this triggers me, but please excuse my lack of diplomacy. I have used this method in many battery operated espruinos this year, it replaced my PPK2 testing with excellent precision, do not dismiss it, it could help your internal QC.

I will explain why I posted in this question. When I got my first bangle I was very involved in the community, I wanted it to be a good watch, I was planning on investing time on it for free. Instead I found out the battery was nowhere near the P8, I complained a bit on the old forum, was told that maybe I did something wrong, I ended up dismissing it and placed it on a drawer for 3-4 years. Some time ago I got some more bangles for different reasons, I did proper measurements on all of them using the PPK2, and realized my original bangle was idling at 1.2mA instead of 150μΑ. Your initial responce in this thread brought back memories. I am sorry, maybe I overdid it.

[spkocz]

Quick update. Since my last post I stopped fiddling with it, no top-ups, no measurements, just left it on the desk and picked it up now and then. It discharged over about 4 days 3 hours, then died.

Left it on charge overnight, ran the battery calibration, and it's now back on the desk. I'll glance at it occasionally to see how it goes - not quite normal wrist use, but enough to get a feel for this cycle.

@enaon - thanks again for the test and the P8 comparison; both were genuinely useful.

@gordon - I'll post the runtime after each discharge from here on, so we've got a few formed cycles.

[gfwilliams]

It discharged over about 4 days 3 hours, then died.

Ok, thanks - well, that sounds pretty definitive then. Thanks for bearing with me and for doing all that testing!

It definitely rings true with everything @enaon had posted for comparison - I just wanted a quick sanity check :)

Please can you send a message to contact@espruino.com with your address and a link to this thread, and we'll get a replacement out to you.

[spkocz]

Before I send you my address - the picture has changed since my last post.

After that first cycle ended (4d 3h, flat and switched off) I said I'd carry on with a few more. I charged it fully again, and it's now ~24h into the new one:

• 12h: 4.139 V (raw 0.31323)
• 24h: 4.136 V (raw 0.31250)

The only thing that happened between the two cycles is that the cell went flat - a real power-on reset rather than a soft one. I had done clean boots during the first cycle and they changed nothing, which would fit something outside the nRF.

That's a hypothesis, not a conclusion - it could equally be an intermittent fault that happens not to show right now.

I'd rather not have you go to the expense of shipping a replacement for a symptom that may have gone. Assuming the offer still stands if it comes back, I'd like to give it more time - I'll let this cycle run and let you know how it goes.

On a personal note: I run Linux on a Framework laptop and I'm slowly migrating a Fairphone to /e/OS — open hardware and free software are the reason I dropped my plan to buy a Pebble and went for a Bangle instead. Whichever way this unit turns out, the response in this thread has more than justified that choice. Thanks for the time all of you have put into this.

[enaon]

After that first cycle ended (4d 3h, flat and switched off) I said I'd carry on with a few more. I charged it fully again, and it's now ~24h into the new one:

• 12h: 4.139 V (raw 0.31323)
• 24h: 4.136 V (raw 0.31250)

very nice. It could be the acc or the TC chip that needed a hard reset. Maybe they were to blame for the idle leak. It now looks as it should. So @gfwilliams was correct too.

We will have some fun trying to figure out how this could come to be the case on a new watch, this is a very nice thread :)