I spent some time last night collecting data and parsing the datasheet for the CXB3590 and noticed some interesting things.
This LED is extremely efficient around 30w. It hits around 190lm/w and produces ~6000lm.
As input power increases the efficiency falls off but even at 12000lm, it’s still operating around 170lm/w
I did some research on DB Custom’s modded Fireflies E07, and he’s hitting 10000lm at something like 100w input and of course, this is only for a few seconds as the whole thing overheats and the cell starts crying for mercy.
I seriously think that we could build something like a Convoy L2 with 2s or 3s 26650s and maybe more heat sinking on the head that could do 10000lumens fairly sustainably. It would require a custom driver though as I don’t think the hyperboost can handle the input current that this would require.
Have you looked at the LEDiL Stella lineup? Thought of doing something like this myself and I figured, with such a large heat sink, that cutting the size of the reflector down would be good thing. You would also gain some level of waterproofing and be able to switch lens according to you needs.
I saw the LEDiL Stella mentioned in the optics thread. The model above looks pretty good for my purposes. 90mm is kind of huge but I guess it is what it is…
If Arrow had it, I’d buy it right now. Gotta love that free overnight shipping.
I only just realized the CXB3590 is ~50mm diagonal.
The hyperboost is good up to 7A max input I believe. But you’d need to find a way to shoehorn it into an L2 host. Methinks one of the 4s soda can lights, along with LEDIL optics, would make for a pretty cool host
The 6A~7A input current limit means that you’d need a minimum of 5s to get close to full output. This isn’t really a problem for my Navy Battle lantern retrofit project but for a “hand cannon” it probably is…
A soda can light would have a very hard time dissipating the ~100w of heat. I had a similar idea but put a big heatsink and fan between the LED and the batteries.
That driver would need a minimum of 5s as well. 4s could drive the LED at full output when the cells are fully charged but it would drop out of regulation or exceed the input current limit fairly quickly.
I’m actually curious if either the Hyperboost or the Led Treiber Xtreme have input current sensing. I suspect they do not and so it would probably be safer to run both off a higher input voltage so that the input current never gets high enough to damage the driver.
I think I want a driver that can handle about 20A on the input. I would also prefer something that can not generate voltages above 50-60vdc as above 60vdc can be seriously dangerous or lethal.
I was looking at the LT8391 last night but I’m not an electrical engineer and so electrical things I design typically just follow manufacturer design guidelines.
I redid my spreadsheet a little bit and so if we assume we’re going from 4.2v to 3v per cell the hyperboost would need 7s and the Xtreme would need 6s.
For testing shenanigans, I just happen to have some 7s4p NCR18650GA battery packs. I think one of them would run the COB at full output for something like 3 hours.
Looks like they work with the solderless COB connectors… Which is convenient… Might have to pick one up to try out for the back yaard light. I do have a cheap 50W COB floodlight that was given to me because of a bad driver. I replaced the drive with a DC boost from MeanWell (about $10) so I can run it off a LiPo to make it portable.
I’ve got all their datasheet for CXB3590 compatible optics open right now. It is really fascinating how the different LEDs produce different beams with the same optic/reflector attached. For instance, the 30 degree reflector produces a 27 degree beam.
I’m probably going to get a FN14074_STELLA-HB to use and a F13380_ANGELA-M to play with.
Although the F13325_ANGELINA-S also looks interesting.
Cool, would love to see some beamshots when you get them in…
I suppose the different characteristics of each manufacturer’s COBs - the thickness of the LES in relation to the board surface, and in relation to the mounting ring - have something to do with the effective beam angles…
I was looking at some “commercial” drivers last night and didn’t really see anything all that interesting. I did see some buck drivers from MeanWell but no boost….
Have you seen this?
It’s a monster and not too expensive so it probably actually works.
I had an idea for heatsinking on the battle lantern but after looking at some heat sink performance data, I don’t think it will afterall. Passively dissipating ~100w is hard.
On the other hand, there are some twin radiator heatsinks designed for ThreadRipper TR4 and EPYC that look interesting but they’re also ~$50-75 and HUGE.
The Dell T3610 heatsinks look like a decent option for a hand cannon as they appear to have a good bolt pattern/mounting plate, cheap, available, relatively compact and the battery pack could probably be mounted to the other side of the fan shroud.
I don’t even care much how it cools - it LOOKS freakin’ cool! NOFAN’s claims about the first fanless PC system though are dubious - they’ve been making fanless computers seemingly forever…
