Bangle.JS 2 flashlight mod

[enaon]

While waiting for Bangle.JS 3, I added a flashlight on the 2.

So I got different LEDS ,both 3mm and 5mm ones, the 5mm were a bit large for the case of the bangle, so the 3mm I tested are:

Optospupply:

1.warm, 60 degrees, 30mA
2.cold, 60 degrees, 30mA
https://www.tme.eu/Document/889801426481e6ebd1de2bda4ad4c4e2/OSXXXP3761B_VER.D.1.1.pdf

3.cold, 30 degrees, 50mA
https://www.tme.eu/Document/2622980f6559e36b045ff56bec4834a9/OSW54L3131P.pdf

Aliexpress
4.warm, 120 degrees, ?(40)mA - aliexpress
5.cold , 120 degrees, ?(40)mA - aliexpress

the Aliexpress ones I got just for reference, and I did not really like them, too warm and too cold.

I went with the warm optosupply, that is not really warm, more like a warmish neutral color. The cold optosupply is not cold either, more like a coldish neutral, maybe it is better suited for a fashlight, but warm is a personal preference. The 50mA oversupply was a bit taller, I needed to trim it, it then dispersed light more like 120 degrees.

The 60 degrees angle is the sweet spot, very similar to my phone's LED. Brightness on full (30mA) is like the phone's LED on low (1/5) scale, usable.

I also added an AO3400 for driving te led and two resistors, 30Ω for limiting, and 10KΩ for keeping the uart TX (D43) pin low.

I used the PPK2 to make sure the pulses do not go over 45mA at 4.2V (30Ω), it does not get warm even on prolonged usage on 100% duty.

There is not space on the top (gps antenna) ,or over the button (BT antenna), so under the button was the only place available. It is a convenient placement, usability is nice.

[enaon]

Some photos

[enaon]

video,~1.5 meters away from the wall. 50% intensity or 18mA, which is about the same as the max brightness of the screen, is OK for walking in the dark.

vd.mp4

[gfwilliams]

Wow, that's very cool - some great work managing to fit that all inside!

[devsnd]

Wow, that's really cool! Would you mind sharing schematics? That's a fantastic mod and maybe I'd dare to do the same to my watch 🤔

[enaon]

sure, the sematics are simple, you need a small N-type mosfet, a 10K resistor between gate and source, and a 30Ω resistor on drain (or between Batt + and LED), like so:

They are a bit small, but you can use the 2 razors technique, placing a razor over the component so that is locked, and then solder, like so:

also a 0.1 wire is much better than the one I have of the opening post, space iis no problem when using this wire

It is even possible to add 2 LEDS for extra power, I did it on one of the bangles just as a test, it is nice but not really needed, one is enough.

the hole for the led is the only part that is tricky I think, so better not use a drill. I used a small soldering iron tip to mark the hole, a small screw driver to open it, and a round shaped filer to get it to the right width.

you will also need some T-7000 glue.

[enaon]

@gfwilliams , I think we talked about this in the past, but I had no photos to explain. This is what I was doing for the button on the P8, I did the same on the Bangles, the button is identical. I filled the button pocket with T-7000, let it dry a bit for 3 days, then pressed the button to help it extend and form a sleeve, then let it be some more. This results in a better feel on the button, more solid, as the dead space gap is filled with t-7000 and the metal ring actually pushes the T7000 rubber not the actuator, and waterproofing. I used Logo Silic before, a silicon based glue, so this a test for the T7000, it may become too hard with time.

as a side note, this 4euro strap from aliexpresss is nice if one prefers metallic bracelets. It helps the HR sensor getting stable values and protects the pin holders, as it is flexible.

[gfwilliams]

Thanks! That's really interesting - definitely holds the button in nicely and maybe having it totally sealed there will make it very unlikely that water can ever get in

[enaon]

@gfwilliams, if you have some time can you explain the following? Not related to the flashlight, just curiosity.

I used the PPK2 to get some measurements

the bangle stays at 75μΑ when idle, BT on, not connected. But every ~40 seconds there is a spike at 40mA, so the standby is actually ~100μΑ. Do you know what those spikes are?

And another thing. When the bangle is moved a bit, the idle current jumps to ~250μΑ for some 35 seconds, like the acc is entering a higher power mode?

[thyttan]

but when I use the internal event I do miss some taps, and it looks like it happens when the polling is slow.

If your onto something here it may be relevant to espruino/Espruino#2702 as well.

[gfwilliams]

@thyttan I'd be pretty sure that one's not related - the event handling (eg from connected/disconnected events) is very different to the accelerometer

Could it be that another event came before the acc was read, if the bangle is in slow poll mode, so some events get lost?

Yes, that's almost certainly it! Normally because we use the chip for double-tap detection, it's not a problem as the way it's configured it'd be very hard to get two close together events as they'd just be classed as a double tap - but if you're configuring it for single taps that's quite likely.

You could just force it into the higher power mode with [Bangle.](https://www.espruino.com/Reference#l_Bangle_setPollInterval) and Bangle.setOptions({powerSave: false})` and see if that fixes your problem first?

[enaon]

@thyttan I'd be pretty sure that one's not related - the event handling (eg from connected/disconnected events) is very different to the accelerometer

Could it be that another event came before the acc was read, if the bangle is in slow poll mode, so some events get lost?

Yes, that's almost certainly it! Normally because we use the chip for double-tap detection, it's not a problem as the way it's configured it'd be very hard to get two close together events as they'd just be classed as a double tap - but if you're configuring it for single taps that's quite likely.

You could just force it into the higher power mode with [Bangle.](https://www.espruino.com/Reference#l_Bangle_setPollInterval) and Bangle.setOptions({powerSave: false})` and see if that fixes your problem first?

Thank yo Gordon, this makes sense, even single tap is quite slow on defaults because it waits to see if another tap will follow, in order to register a double tap. The way I use there is no delay, I can tap it as fast as I can and all taps will register. I will test your recommendations

out of curiocity, do you have any insights, maybe an idea of how many people are using dtap to wake up the watch? I feel it is very unreliable, the second tap being directionless, it results in too many false positives for me.

the same goes for the twist to wake. It is very easy to register if the hand is down. I don't want to sound rude ,but maybe this needs some more work? I did this, as a basic axis filter, maybe someting like that could be embedded by default?

Bangle.on('twist', function() {
let tilt = Bangle.accelRd(0x10, 2);
if (tilt[1] === 32 || tilt[1] === 16 || (tilt[0] != 8 && tilt[0] != 1)) return;

[gfwilliams]

do you have any insights, maybe an idea of how many people are using dtap to wake up the watch?

I'm afraid I have no idea at all! I guess I could in theory upload everyone's settings.json file to find out stuff like this, but it feels like it really goes against the whole 'you own this'/'no tracking' ethos of Bangle.js!

[enaon]

do you have any insights, maybe an idea of how many people are using dtap to wake up the watch?

I'm afraid I have no idea at all! I guess I could in theory upload everyone's settings.json file to find out stuff like this, but it feels like it really goes against the whole 'you own this'/'no tracking' ethos of Bangle.js!

yes I agree. It could be me just me, but I eally feel I am not that strange , I really believe those function do not work very well. I do not really mind, I take joy in customizing my bangle, but maybe if you could add a check on the loader, the same way you ask for installed apps, maybe it could help you get an idea of true usability of those functions. In any case, thank you, my problem is solved.