I finally got the 2300K LEDs and tried them in my rechargeable tealight candles. I absolutely love it, it’s the closest I’ve been to a real candle light, and in a light holder of any kind it looks convincing.
From left: Original monochrome orange 3mm LED, 2300K LED and 3400K LED. Camera @5000K
Here they are with their included diffusers, looks pretty good. Can’t wait to solder all 12 of them and place them around the house.
I still use real candles on my living room table due to the cozyness and smell, but these are a lot safer to use unattended around the house, especially with cats around.
I hope you meant to say “10 in parallel to an AMC7135”. But why do you choose AMC7135 and PWM? :facepalm: The Arduino could be used to sense current over 3x sense resistor paths, precisely tuning a driving MOSFET VGS in each of them for accurate, PWM free linear current control.
No, that’s fine.
For 10 in series you need 32 29 Volts.
For 10 in parallel you only need 3.2 2.9 Volts, like from a single Li-Ion cell.
The AMC7135 will eat up a minimum of 150 mV.
In any case, driving leds in parallel could have some current sharing problems due to Vf differences between emitters. This can be minimized by binning leds with a multimeter, i.e. set them all extended over a table in a room with stable temperature for a little while first (acclimating emitters, they're sensitive to temperature) while laying a multimeter close, then use the multimeter to read and note down every emitter's Vf systematically at least once. Use the results to sort the emitters in Vf packs: 10 lowest, next 10, …, last and highest Vf 10-pack. This would minimize the above problem.
There are many ways to drive emitters. Arrays of series emitters will display closer average Vf sum, this for example means it would be far better to drive a 10-pack in 10S or 2P5S arrangement. Watch out I said 2P5S, this means 2 parallel 5S arrays, which is a lot different than 5S2P. There are cheap boost DC/DC converters which can output that voltage from li-ion cells easily:
The MT3608 boost module is going to be reviewed by Henrik soon I think, sent him a 5 piece pack long ago. Now that I recall this, it was the 5th of August, he should have received the package sometime in September and we're almost in December already.
Regarding driving LED’s in parallel, I think we can make some basic deductions from djozz’s data:
If you’re driving 10 at 350mA, that’s 35mA each.
From djozz’s graph, that’s about 2.95V.
If you had one LED in the group with a Vf that was lower and therefore hogging the current, then from the same data, I’d estimate the resulting current in the worst-driven LED as follows:
0.1V lower = 55mA
0.2V lower = 70mA
Since one of his survived an isolated test for 12 hours at 100mA, the over-current does not seem to be a problem. However, rngwn observed that if packed together, they heat up and may eventually run away:
So for the idea of running 10 on a 7135, I think heat is the main concern.
Yes, heat can be a problem in those 3xAAA 20 or more LED lights.
Especially those in the middle of the PCB can get hot on the crappy stock PCB they’re soldered on.
Good idea to epoxy a heat spreading disc on the back.
The LED that’s in those bullets is a 0.5 Watt 5mm LED similar to this:
I bought a copper bullet branded Astrolux (i think TB-01 ?) with pretty powerful 3x LR41 inside plus 3x LR41 spare batteries a few years ago from Banggood.
I estimate it’s around 15 Lumen with fresh batteries, maybe even 20 Lumen.
Pretty impressive for the size.
Those Mateminco branded ones are crazy expensive.
Here’s a stainless steel one: https://www.aliexpress.com/item/32844630381.html
You can see the brand name is photoshopped away, but i bet it says Astrolux.
Here’s an aluminium one: https://www.aliexpress.com/item/32839012681.html
Either way, yes it would be nice in high CRI warm white, but Djozz tried it with a Yuji 5mm LED iirc, and it was much much less bright.
These LEDs here will do a better job, but still a lot less than the stock LED.
.
