This is probably hijacking your thread a little, but the knowledgeable folks who I have conversed with over the centuries are here, and the topic is UV LEDs.
Despite all the recent claims of 4 watt, 5 watt, overdriven, turbo Convoys (7135 drivers), I believe there ain’t no such animal. The engineers in our fluorescent mineral group have watched the ridiculous power claims climb over the past three years with amazement At the power levels folks are claiming today these little lights should literally be burning your hands. They aren’t - this blog post explains why, all comments welcome: Convoy false power claims
dude the 7135 is a linear regulator
Sure is, but I don’t understand your point
I'm not sure what you're getting at. You may be correct regarding the high Vf of the UV LEDs preventing high output on the UV Convoy, but the Convoy hosts are capable of dissipating over 7W of heat. I've personally run a Convoy on a fresh cell for hours at approximately 2A. Not much sag on that cell, but I'll be generous and assumed 2A * 3.5V for the power dissipation. For short periods, they handle significantly higher power well - not just the stock 2.8A that the white Convoy's come equipped for, but so-called hot-rodded Convoys can pull currents in the high teens.
Even your own chart shows the Convoy doing 3 Watts at 3.9V, which could be sustained for a while before the battery got that low - definitely more than just a few minutes.
700mA is very little current, it won’t burn your hands (but it could damage the Nichia LED if the heat transfer to the star and pill is not good enough as that LED is quite small, the LG LED can handle more current but has more visible light). Even 1400mA shouldn’t get too hot. 2100mA (6x7135) and 2800mA (8x7135) can indeed get quite hot to painful levels if left unattended. There was a temperature test of the 8x7135 driver: How hot does an S2+ get at 2.75A? Comparing an anodised Convoy S2+ with a fluorescent yellow powdercoated Convoy S2+
I guess that a 4.35V cell like the LG ICR18650E1 would be better for extended usage of the Nichia version.
Edit: you are making your claims based on a ‘typical’ 2200mAh cell. Nowadays 2200mAh would be either an obsolete or cheap cell, or a cell optimized for very high current for power tools or vaping (20A-30A) rather than capacity, not a typical 18650 cell. A modern cell 4.2V cell like the NCR18650GA can keep the voltage much better. In fact a 4.35v cell isn’t a large improvement over the 18650GA. From HKJ’s battery comparator :
It means a 7135 based driver will maintain a constant current until it can’t sustain it anymore. So a 1x7135 (standard bin 350mA) driver will maintain 350mA , nothing more, nothing less until 3.6V ish, and shortly after that turn off. It will not dim from 4.2 down to 3.6 unless you apply PWM.
Not too worried about heat, just driving voltage. But that said, UV LEDs are much more sensitive to heat - just look at any graph, derating starts at 40c on most of them. Of course hobbyists can do all kinds of neat things, drive them harder, etc. But production flashlights to the masses, probably a bad practice.
As far as voltage goes, drop below 4 volts and the current to the average UV LED drops drastically. 3.7v (rating for lithiums) and it really sucks.
Don’t think so. The driver just goes into direct drive mode. The LED will light (in the case of an LG or Nichia LED) as long as the battery voltage is high enough >3.0V~, but will progressively become very dim as the voltage drops from 3.9v to 3.0v (or so). PWM another story, I’m talking about single mode drivers, no smarts. edit - keep in mind that UV LEDs expect ~4.0v
and I’ll repeat, 7135’s are not direct drive
Maybe a little better but according to the graphs it still drops below 4.0v pretty quick it seems. But in our market we have a bunch of people who have no clue about batteries and buy them off Ebay, your local vape shop, Amazon, etc. No idea what they’re gonna get (except the cheapest). In fact, folks selling Convoys to the mineral market on Ebay and at shows are including really junk batteries.
I’ll dig into the one you linked to see if we can get a little better run time - thx for the link. But sadly, won’t help the general population.
…and I’ll link this: AMC 8*7135 3A CC driver is in fact direct drive
“AMC 8*7135 3A CC driver is in fact direct drive” (but I’m always eager to learn)
Linear driver are never in direct drive.
As long as the battery voltage is high enough, constant current will be sustained.
The problem is that many lights don’t have an optimized current path. It’s not much of a problem with current low VF LEDs…
But it is a problem with high VF LEDs.
Voltage drop over the springs, over the switch, and the wires can be solved.
Using copper alloy springs, thicker 20AWG wires and a thick mechanical/FET switch, we can get max regulated runtime on UV LEDs.
The perfect driver for UV LEDs would be a linear/boost driver.
Right so since UV emitters require 3.8V it’ll sag once it hits that point, but it’s NOT what you make it out to be, certainly not what’s in your article. With a 2*7135, you never get 4.2v, it’s 700ma @ 3.8V. To compensate, you can just use more 7135’s.
I guess that’s my point exactly. UV LEDs require a boost driver to run off 3.7v and lower, a buck/linear at 4.2v. But problem solved if you go to a two battery solution and driver, but not feasible for a Convoy as marketed. (Please let’s not get into CR123s - whole ’nother set of problems).
UV emitters range from 3.7v to 4.2v. Out of the hundreds of LG parts I’ve used, the ones I have tested are all 4.0v. The Nichias I have tested are also 4v. I wish I could get the lower voltage bins, but not easy, and probably pricey? 7135s will drive the LED at its Vf as long as you supply a tad more at the input.
Assuming 4V, even at 3.3V you’re looking at a worst case scenario of 18% sag. Visibly would be difficult to see. You’re really splitting hairs for your conspiracy theory and again, the current is never 700ma @ 4.2V
Not splitting hairs. Measuring UVA output with a meter. The drop is significant. Not a conspiracy theory - simple design. But I’m worried about the mass market for flashlights sold from China, not the great lights that y’all build here and fine tune for every last photon. Love that part of this hobby, but I have to design for the masses, and it has to be a reliable product. LG parts must be derated starting at 25-30c - expecting the Convoy to reliably dissipate heat well enough for one of their UV LEDs driven at 5.8W is wishful thinking (IMHO).
Have you tried the new not-so-new LG LED which is the subject of this thread? The Vf is 0.3v lower than previous offerings, so it should keep in regulation a longer time in a linear driver. However it won’t be almost a point source like the Nichia.
No problem hijacking this thread pirateo, it is good practice on BLF 
I think something should be noted about the Convoy UV flashlight: since Convoy changed from the Nichia 276A to LG 365nm leds, the leds are reflowed on copper DTP boards which makes an immense difference in heat shedding of the led. It improved 3 things:
1)the 276A seemed a very sensitive led to damage while the LG leds seem to handle abuse better (this is my impression at least)
2)the aluminium board that Convoy used for the 276A had about the worst thermal properties that a ledboard can have. I tested it against my own custom (non-DTP) board and my board was miles ahead (link)
3)DTP boards have proven to shed heat way way better than non-DTP boards.
I think that there are still many 276A Convoy UV flashlights in shops around the world but the newer ones with LG leds should hold up much better (and not the least because of the ZWB2 filter)
That said, if the led should be kept under 40degC for keeping performance in the long run, my estimate is that in an S2+ host with DTP board the current should be kept under 1 A. But for occasional users that may not be needed and they can run it at 2A and have a bit more output.
Btw, at 1 A, the LG led of the OP has a voltage of 3.55V, assuming 0.3V loss in the circuitry at 1A current, the battery should deliver 3.8V to keep the 7135 chips in regulation. A modern but cheap high currrent cell like the Samsung 30Q can keep 3.8V at 1A for about half the drainage of the battery, the other half will show slow dimming of the flashlight.
FYI, I recently bought a Nichia S2+ and it definitely uses a DTP board