I’m building a thrower flashlight with the goal of sustaining maximum output for the entire runtime.
The LED is an SFT90 (much cheaper than SBT90.2 in case I burn one). Driver is some random buck converter from AliExpress.
5 Thanks
A bit more design and prototyping later, I’ve fit a reflector for the Convoy L21B for now. First outdoor tests are very promising.
Ignore the digital ammeter - it’s not even close to accurate, I suspect for EMI reasons.
There are some power issues (AliExpress converter can’t put out rated current and USB C trigger board isn’t requesting enough current) so it’s only running at around 12 amps to the LED for now.
The fan alone is about ten watts. At only 65W USB C, 12 amps max seems to track.
I’m able to hit 14 amps with a stronger USB C wall power brick but the LED driver starts having flickering/blinking issues after that.
1 Thank
The little fan is from a Dell 1U server so it’s very loud and fast, all in a tiny package. I might consider switching to a larger heatsink with a less power hungry (and obnoxious) fan later, but this is what I had on hand.
I tried using a Fresnel lens to collimate/focus the light, but it just converged instead of being collimated.
I’m definitely doing optics wrong - feel free to enlighten me on why flashlights don’t use Fresnel lenses.
I’ve got a little convoy 3deg TIR optic, but I’m not sure if that will be tighter or wider than the current L21A one.
KD huge reflector is on the way.
1 Thank
This project definitely needs a better name. Open to suggestions…
I love the way that thing looks!
1 Thank
Ive wondered if a pc atx or sfx psu would be ok for something like this. They have 3.3v a 5v and a 12v rail that can handle higher currents up to and over 1000w. Power the emitter, fans etc. Has active cooling also. Cheap as chips anyway.
Maybe even a peltier or cpu water cooler for the heatsink. Those things are designed to handle 100+ watts.
Overclocking emitters… lol. Next step, liquid nitrogen cooling 
Very cool project. Could house it all in one of those super old school mega chunky Eveready lanterns and be retro… but not.
As for name…flash bang might be a bit corny. Unless you are into transformers. Maybe: “Project Sól”
Sól being the norse goddess of the sun
1 Thank
PC PSU is definitely an interesting idea… getting a mains supply to it is a limitation though.
I have daydreamed of using LiPo “drone batteries” which have a huge current capacity / power density to power some ridiculous LED.
How much heat does an LED actually produce? I’d hope that most of the energy I put in leaves the front as light. I guess this is calculated from the efficiency numbers.
Besides being able to stay on “turbo”.
What are the ultimate goals here?
Distance ,run time? Beam? How large of a hotspot at 100 m? How much spill at 100 m?
What do you see as your limitations right now? Do you want to supply it with a USB supply with 65 Watts 100 w or direct battery at what voltage? Let’s see an actual beam shot at 100 m of what you’ve got right now.
You can also get an inverter attachment for a milwaukee 18v power tool battery that will deliver a maximum of 175 watts at 120 v.
Its kinda doubling down on inefficiencies tho i suppose.
But on the drone battery idea… a remote control hobbyist lipo pack is very affordable. A 1s, 8000mah lipo pack will deliver up to140c for around 20-30 usd for cheaper brands. They go up to and beyond 6s configs. Better brands will be 50-100usd
You can readily get various sizes in various configurations for various remote comtrol hobby vehicles. From tiny little thumnail sized lipo… to beyond 8000mah like this one:
This one pictured is 93.mm x 47mm x 18mm. The holes are the + and -. Lipo tech and power requirements of modern rc vehicles is pretty extreme. Lipo out performs liion by a lot in terms of discharge. But not in terms of density and cycles.
Design goal is to prioritise intensity/candela/“throw”, contrary to many Youtubers’ “100000 lumen flashlight” in a similar form. This drove the choice of emitter and reflector (the biggest one I can find is coming from KD). Essentially, I was disappointed by many SBT90.2/etc. flashlights stepping down from their maximum output.
Not sure on exact specifications/parameters. Maximise distance (i.e. small hotspot, minimal spill), beam quality/CRI/run time mostly irrelevant.
Main limitation now is power supply issues. The Aliexpress PSU can’t supply 20 amps, and the USB C trigger board doesn’t ask for 5 amps for some reason.
I will consider using direct battery in the future. I didn’t know about those rc vehicle 1s batteries.
Thats just the 1s pack. 4s and 6s packs will be much larger and more readily available. Ive seen 4s packs over 10,000 mah. But they can deliver huge amps and are affordable for what they are. Power tool packs are also just stacks of 18650 or 21700 cells with bms. Some are in excess of 10ah @ 18v. Makita make 40v batteries have probably the highest sustainable output of the bunch at about 80amps. Which is a shade over 3500 watts.
You have plenty of options for high output batteries once your design is down pat.
Agreed. So actually the main bottleneck is finding a suitable driver (especially if I use multiple emitters in future).
I used USB C initially because some of the places I build this don’t like li-ion batteries being used.
As it turns out, accepting the OP Reflector substitute from KD was highly unwise. Despite the larger size, the throw is absolutely killed by the non smooth finish. Should have seen this one coming.
My phone luxmeter reports only 1/4x the hotspot lux at the same arbitrary distance, compared to the smooth smaller reflector. It was also evident with how much more diffuse the beam was.
Time to figure out if I can exchange/return the Reflector to Kaidomain.
@Ruckusmattster has inspired me to look into Fresnel more with that excellent spotlight build.Ruckusmattsters 13th Old Lumens entry - machine made
I’ll get the right sort of lens this time, as well as choosing a larger focal distance to reduce how much light I lose to reflections.
1 Thank
I have access to my stash of heatsinks again too. Given that I’m currently using a 40x40mm tiny one, I’m imagining some monstrous multiple SFT90 (or other) builds using the huge PC heatsinks. Alternatively I could just make this thing quieter…
Any problems with putting multiple SFT90 in series? Perhaps with Vf drift causing thermal runaway?
i found for the sft42r LED that the ideal focal length is somewhere between 70-80mm for a 200mm diameter lens.
you want to check from each edge of the LED how wide the emitted light will be considering the LED has a emission angle of 120 degrees. you can do trig where the opposite is radius of your lens and the adjacent is the focal length.
i would aim to have a slightly too large focal length so that the LED doesnt cover the entire lens, if you go slightly too small like me then you have to deal with the internal reflections (if fully enclosed) or just stray/wasted light that doesnt hit the lens
I’ve ordered the 200mm lens.
Also forgot that I was trying to design this into a projector so the big lens might not be as useful. However, for pure throw I’m curious what it can do.
Because of the lens size it needs a redesign. I’m thinking of using aluminium extrusion and (laser cut?) panels.
Alternatively (if I’m lazy) I will just chop up my bits of threaded rod and keep doing that. Will be hard to keep everything parallel/square though.
Power issues were because my power bank was never capable of 100 watt PD output - only 120 W “xiaomi special sauce protocol” !
As much as I want to move to Lipo/Li-ion, the lab I work in doesn’t like bare batteries being used.
I’ve made a little PWM board to throttle down the fan. Hopefully getting its power consumption below 5W will let me allocate the rest of the power to the LED (neglecting converter inefficiency).
Have you considered using power tool batteries? Hundreds of watts are no problem for many brands’ 18/20v batteries.