My battery last 5 days top

Hello,
I am having an issue with my Banglejs 2...
Most of time I have to charge it every two days.
The battery dies super quickly, compared to the estimated 2 weeks said in the kickstarter.
I dont really have a lot going on in it, just the pedometer. Not really anything else running in the background...
Except for the Android app/integration that has been really spotty for me since the start. Could this be the issue?

Anyone else having bad problems like this?

What are your "unlock/light up LCD" settings?

brightness 1
timeout 10
wake on btn1 on
wake on touch on
wake on face up off
wake on twist off

I've pulled down my LCD brightness to .5 (I find that more than enough with the reflective display), changed Twist threshold to .2 and Twist timeout to 500. Using the Anton clock and always connected to Bluetooth. I don't use any pedometer or heart rate monitor stuff. Easily lasts 7-10 days wearing it from getting out of bed until going back to sleep.

Edit: maybe the wake on touch turns on the display a lot, unintentionally. That could really impact battery life.

I think the wake on touch draws a bit just by the active sensor. Try wake on twist and button.

I get about 10 days with only wake on BTN1 enabled. All other wake ons disabled, No heart rate monitor and LCD set to 0.1.

I changed the settings as suggested by users but I still get the same poor battery life.
I think there is something bluetooth related that is draining the battery and it may be the same culprit as my other problems, because it lasted for like 8 days the first time

That is surprisingly low battery life - have you installed a clock face other than Anton? It's entirely possible that some clock faces really drain the battery hard.

But the bluetooth connection could hurt battery life - usually it goes to a low power mode after 1 minute of inactivity but if stuff is being transferred it'll stay in the high power mode which draws 1mA, which would fit well with you 8 days battery life.

It is possible to force it to use the low power connection mode all the time, so I guess that could improve things for you. If you install the 'Custom Boot Code' app and paste in:

NRF.setConnectionInterval(200);

That'll use the low power mode - BUT subsequent app uploads will be WAY slower until you delete the app again.

Just a slightly related note, after I drowned my Kickstarter Bangle JS2, I found an Alibaba link to a ~$50 sample Q3 from SMAWatch. I think I got lucky because the price at the link promptly went up to $200 after I placed my order!

Anyway, the sample I got is lucky to run for 16 hours before it dies. My drowned KS was still operational, except for its button, so I was able to compare the two. The new sample one was running about 3degC hotter according to the internal temperature sensor, and had a case temperature about 1degC hotter according to my FLIR camera.

That 1degC difference remained even when I turned them both off. Although that test was a mistake because now I can't turn my KS Bangle back on because the button doesn't work :-(

The shorter run time is clearly due to excessive power consumption by something. It's possible (probable?) that production variations are behind this issue.

That 1degC difference remained even when I turned them both off.

I've experienced with my FLIR that such a small temperature difference can be caused by different reflectivities of the surface or even the thermal reflection of the own body or the angle of it to the object.

The shorter run time is clearly due to excessive power consumption by something. It's possible (probable?) that production variations are behind this issue.

I'd guess it is far more probable that it is some software issue. 16 hours vs weeks is not due to 'production variations'. 16 hours means it does not sleep.

You had to unlock it via SWD and flash your custom build. Which tool you used to do that? Using SWD via openocd keeps the debug hardware powered so it will die soon unless you know the magic to power it off at openocd disconnect time (nrf52.dap dpreg 4 0x04000000 ; shutdown).

Or does it happen again even after battery is completely drained? Then it is not SWD but something else (probably still software). Can you share full hex file you flashed after unlocking that has this issue?

I've experienced with my FLIR that such a small temperature difference can be caused by different reflectivities of the surface or even the thermal reflection of the own body or the angle of it to the object.

I'm aware of that. I was looking at the two watches side-by-side and the temperature difference was always apparent from different angles.

BangleJS2 Flir

Regarding my 16hr Bangle, it doesn't charge up to the same voltage reported by my KS Bangle (readAnalog(D3) is 0.293 vs 0.341). The original Swawatch Q3 firmware never reported above 56%.

Can you share full hex file you flashed after unlocking that has this issue?

I flashed the following using the drag-and-drop functionality of the nRF52DK:

espruino_2v11_banglejs2.hex

Edit: The watch has now fully discharged several times (easy when it only lasts 16hrs), so I don't think it's the SWD.

Regarding my 16hr Bangle, it doesn't charge up to the same voltage reported by my KS Bangle (readAnalog(D3) is 0.293 vs 0.341). The original Swawatch Q3 firmware never reported above 56%.

so you still have two watches that you can do same steps with (mass erase, flash same hex) and they behave in different way?

anyway, looks like bad battery to me if it does not charge fully and drains fast. or something wrong on PCB that leaks current

also as you previously mentioned if you do power off from bangle settings menu which puts cpu and other hardware to deep sleep and it still dies in this poweroff state in a day then maybe you indeed got a lemon from SMA

then maybe you indeed got a lemon from SMA

This is quite possible I'm afraid. While I'm sure there are some Q3s kicking around especially in distributors, right now I believe I'm ordering pretty much everything SMA have - to the point where they're not even able to ship out the full amount of units that I ordered.

I'm actually quite surprised they are still offering them.

@Gordon - I'm wondering under what configuration you were confident to say 4 weeks battery life on the Kickstarter. I've never seen more than 10 days.

My config is usually

• A clock that only updates every minute (Pastel, Pebble, Rebble, Anton, Simple)
  
• Android / Messaging - quite often I disconnect from Gadgetbrige in the evenings as it conflicts with the IDE and AppLoader.
  
• Pedometer, Battery Widget (themed), Lock Widget
  
• LCD brightness set to 0.1
  
• Only Wake on BTN1
  
  

Should I be turning BLE off in order to get the 4 weeks ?
What other things should I consider to maximise battery life.

Does BLE power make a lot of difference ? I notice I can have 1 Bangle 2 upstairs in the bedroom doing nothing but I can setill connect to it via my chromebook in the living room 2 floors away. Definitely going to experiment with turning bluetooth off.

I wondering if the BLE boot power level could be a setting in the settings menu?

I know GPS draws a lot, so if I use that I expect to have to charge it same or next day. As an experiment I ran my battery down in 6 hours by leaving the GPS on.

Is there a list of what current different things draw ? Eg a 1 minute clock draws x, a 1 second clock draws y, connection to gadgetbridge draws zMa, Is it possible to measure the current draw externally without openning up the watch ?

An anecdotal observation:
The battery in one of my Bangle 2 lasted 27 days until I charged it. It still had 9% left then so might have survived still a day. Pretty much default settings and apps. Bluetooth was on, wake on twitch on, BUT ~90-95% of the time it was lying idle on a table. Rest of the time I used it for gesture related activities and uploading/downloading data, relatively exhaustively. If I’d used it on my wrist all the time, the battery would most probably not have lasted that long.

If I’d used it on my wrist all the time, the battery would most probably not have lasted that long.

I agree, waking up the hardware with a motion and logging it (to Flash?) will cost a lot of energy.

Is it possible to measure the current draw externally without openning up the watch ?

Maybe if you measure the current of a fully charged watch with a precise USB amperemeter. I will give it a try later.

Is there a list of what current different things draw ?

https://www.espruino.com/Bangle.js2#powe­r-consumption

Some impressions of current measurements with an USB Multimeter UM25C

• charging: 109 mA
  
• fully charged: < 3 mA
  
• Bangle.setLocked(0) : 5 mA
  
• Bangle.setGPSPower(1) : 24 mA
  

4 Attachments

Thanks @HilmarSt - good to know. ANy chance you could do a meaurement of installing Apps through the Loader or the IDE ? I would be interested to know what that is like even of it is for a 60 second period. I do a lot of installing , deleting, changing apps.

Thankd @Poolitzer - I wonder what the NON idle accelerometer draw is ? IE walking around.

The accelerometer's data sheet ( https://kionixfs.kionix.com/en/datasheet­/KX023-1025%20Specifications%20Rev%2012.­0.pdf ) states on page 8:
Current Consumption
High Resolution Mode 145 uA (0.145 mA)
Low Power Mode 10 uA
Standby 0.9 uA
As "Acceleration sensing is based on the principle of a differential
capacitance arising from acceleration-induced motion of the sense
element" I don't expect a current change in the device itself when in motion - only if switching from Low Power Mode to High Resolution Mode when a motion is detected.

ANy chance you could do a meaurement of installing Apps through the Loader or the IDE ?

I tried it, but it can't be measured with my USB multimeter - an oscilloscope is needed:
The download is usually very fast (<2 seconds) and the backlight is instantly turning on with a big current draw. I suppose that the energy for downloading and flashing is minimal compared to the backlight's.

Does BLE power make a lot of difference ?

I see no current change on my USB multimeter when connecting the watch in the IDE or in the App Loader - maybe Bluetooth Low Energy needs very low energy ;-)