Burned out 3 boost converters trying to power a 100W LED... at my wit's end... please help.


Joined Feb 8, 2018

Heatsink - sounds good. 60-70°C is perfectly acceptable for long and reliable life. The boards made on aluminum are really good for high power surface mount since they spread heat way better than anything that can be done with the foils on a standard epoxy-glass substrate.

Eyesight - Ha! I used to be very near-sighted but now have plastic eyeballs (just plastic lenses, actually) and absolutely fixed focus eyes. It has a lot to do with why I don't actually do electronics anymore. My brother used to be a competitive marksman (bench rest, mostly).
I've done a lot of switcher design and used those green and blue cores extensively. They are quite a good compromise of performance and cost.

Board differences - a problem with rather a lot of stuff from China is that something decent gets copied - sort-of, and then maybe the copy gets copied, and often the copies are second-rate approximations. Until you get something, examine it and test it, you just have to hope for the best. From what I can see in photos, your board looks pretty good, but looks only count for a little. For example, for ceramic capacitors of the same nominal value and voltage there is often a choice of two or three or even more types, differing in the ceramic used, which affects the electrical behavior. The better type might cost 2 cents more, so someone who doesn't understand the difference buys the cheaper part that is inadequate for the task. (One of my clients once showed me a bunch of caps they bought to use on a board designed - I called them crapacitors.)
It takes so bloodly long for things to arrive from China that buying something and deciding you want to try a similar something is really really frustrating. I order a few things from Amazon marketplace vendors early last year and they were still drifting in three months later.

Stuff on perfboard: It may not be a problem, but I really do recommend some extensive testing with the original trimmer on the PCB. I wish we could scare up a schematic for your board - it would be very helpful in assessing sensitivity of that bit of circuitry to noise. Often the part of the circuit that measures the current is operating at millivolt levels, so it doesn't take much noise coupling in to scramble things. It's usually surrounded by circuits that switch high currents and voltages at high speed. I often say of switchers that everything is in conflict with everything else. It isn't much of an exaggeration.
I've seen and spent many many hours tracking down and fixing noise problems in switchers and may seemingly trivial changes that made things much better or much worse.

FET failure - I have no reasonable explanation or suspicion; on paper that FET looks very very good for the job
FETs (most semiconductors, actually) almost always fail short-circuit, though if that leads to really high current for long enough they'll blow open. I've had lots in packages much bigger than your FET blow the plastic paclage apart with a loud bang. But that was work on a 1200 watt AC powered switcher. They know lots of ways to frighten you. I'm like a nervous cat when I work on some of that stuff. I literally used blast shields sometimes - acrylic or plywood to shield me from bits that might go flying.

Best way to drive the LED: I think what you are trying to do is quite reasonable given what is available off-the shelf. There are some commercial high-power LED drivers that can be pulse-width modulated for dimming, but that isn't something you want for photography, since the PWM frequency is usually low enough that you could have a whole exposure at full brightness, a whole exposure at zero brightness, or shot with half & half if you're using a focal plane shutter. One drawback to changing the current to change brightness is that it does have a very small effect on color temperature, but precious few eyeballs on this planet would notice. (little aside on that - I used to do photography a long time ago. I was doing some GUI design about 3 years ago and set up what was supposed to be a neutral grey background. It looked too brownish to me. I dug out an old Kodak neutral grey test card, looked at it and concluded my memories of neutral grey were wrong, since it also looked brownish. After I had my eyes repaired, the brownish cast was gone. I'm actually really surprised color memory could be that good. For awhile I had one fixed eye and one bad eye. The color differences between the two were astounding.)

Dynamic response: For a fixed load like a LED and with a well regulated input supply, slow response shouldn't be any kind of an issue. Dynamic response comes into play for rapidly changing loads and/or source supplies.
From what I've seen, most boards like this kind of cobble the constant current circuitry into what is primarily a constant voltage regulator. Sometimes the CC circuity is made extra-slow, either because the designer isn't very good or there really is no practical choice that isn't quite expensive and complicated. If you turn up your brightness pot and the LED overshoots in brightness and then settles, or similarly if you turn the brightness down, it does suggest response that is slower than it needs to be. But for your ap it shouldn't be a concern, though it might be somewhat annoying.

I'm new here and not sure I'll hang around for long. It's been kind of fun, but I could sink an enormous amount of time here. If I go away I'll try to remember to drop in for a look at this thread from time to time until you have achieved success or frustration has sent you to climb a clock tower with a ...

Good luck!


Joined Apr 28, 2012
I have burned quite a few MT3608 and SX1306 boards, trying to drive LED chips from 12V.
Though well within specs they dont seem to like the 12V battery.

From 4V lithium its no problem.

I tried various things but after burning more than 10 I gave up.

With the well known 150W boosters I didnt have this problem.

But I havent tried 100W chips, theyre pretty low Ohms, so could be theyre damaged when starting up or shutting down,
by some abnormal condition.Could be the FET is not what it says or infact is killed by some transient.

Mind the LED resistance is dynamic, and also a large chip has some capacitance, I dont know what is impedance of this capacitance.
This was one idea I had why the modules burn out.

You could try changing the diode and FET on the module, probably inline a small resistor, or actually use a LED driver

If the LED forward voltage would be 60 to 70V probably a CFL PCB would work, this is the burning voltage of a tube.
Pretty cheap and theres 100W kinds

Also theres new LEDs for mains direct connection since a while.
These are chips with integrated driver! Saw them recently online.

Thread Starter


Joined Feb 5, 2017

[...] I'll try to remember to drop in for a look at this thread from time to time until you have achieved success or frustration has sent you to climb a clock tower with a ...

Good luck!
Hahahaha, you're my favorite. But, it would be more of a "untill frustration makes you grab hold of the exposed mains solder connections on the AC-DC converter" sort of thing.

Also, plastic eyeball lenses? What in the hell... fixed focal length eyes!?! (Note to self, be kind to EPB for the day when he ascends with his robot brethren and becomes our new robot overlord.)

Now, in regards to what you said about chinese manufacturing, and noise, and interference, and sensitives, i suppose that all makes sense. What I'm supposed to do about it to prevent future failures, beyond twisting the cables around each other... i don't know. If all that is being said by you and by others about sensitivities in the circuit is true, then WHY oh why am - I - the only one having these issues? Tons of people on youtube have built the same light, with the same components, and seemingly the same circuit, and none of them have the same problems (though, to be fair, they're all using different power sources).

I suppose i will just have to buy new boards, wait the month and a half it takes for them to get here, and try again, both with the original trimpot on, and with my modified circuit. That pisses me off though.. seems like im just throwing money to the wind, and I have to postpone my photography plans for more months. Alas.

And yeah, like you said, while I know a PWM circuit would be easier, it doesn't work for photography for the reasons you listed. This also unfortunately invalidates Mains/Integrated Driver LEDS, since they operate with a flickering system to cope with the AC electricity going through them.

The only other thing i guess i could use your direct opinion on, then, is which boost converter i should use. See, i could order new 250W boards for like 30 bucks a pop (even though they seem to be universally sold-out at the moment, across all websites), or i could go with a different one.

There's this one, which is only rated for 150W, and says it needs increased heatsinking if its dissipating more than 90W of power,

or there's this one, which is rated for 600W, but might be too big to fit into my enclosure,

or there's even this one, rated for a ultra-overkill 1200W.

Or should i just bite my tongue, accept my losses, scrap everything, and try rebuilding it all with a dimmable LED driver... if i can find one.


Joined Feb 8, 2018
Plastic eyes: My meat lenses were chopped up with a laser (pattern looked very much like a round waffle, two heavy cuts so the lens could be broken into quarters and smaller cuts to make the quarters easier to liquefy with an ultrasonic tool (phacoemulsification) and suck out. Plastic lenses were implanted to replace my meat lenses. I actually ponied up $6000 for "premium" implants (Crystalens by Bausch & Lomb) that should have given me some degree of ability to focus, but I got zero benefit from them beyond what I'd get from an ordinary low-cost implant. (this is cataract surgery; my left eye was pretty much functionally blind and my right seemed like it was starting the very rapid decline I experienced with my left; I also have floaters and flashers from vitreous detachment in both eyes, but they are just really annoying sometimes - vit. detach. can cause retinal detachment, which is a whole lot worse). Enough whinging.

If others are succeeding with the method, it is even more baffling why you should be having such problems. You don't happen to live near a radio or TV transmitter that is blasting out many thousands of watts, do you?

All of the boards in your links look reasonably well made, but again looks don't tell a lot. The 600 W one uses an input power jack that is pretty dubious for the amount of current. Curiously, all of them use (if it is what I think it is) a more expensive lower-loss core for the inductor and two use a winding design that reduces power loss in the winding, but those are efficiency matters, not reliability (there is a reliability issue with the green/blue core - if it runs too hot for a very long time its characteristics change for the worse; depending on design that could mean the useful life is only 30 000 hours instead of 300 000 hours; the LED likely begins to weary of life at less than 100 000 hours, depending on temperature). Without actually testing the various boards or at seeing schematics for them, I can't see any compelling reason to change.

Thread Starter


Joined Feb 5, 2017
"You don't happen to live near a radio or TV transmitter that is blasting out many thousands of watts, do you?"

Wait... do you not?

I guess that's why im the only person I know who can tune a radio or fly a drone with my thoughts alone..

"Without actually testing the various boards or at seeing schematics for them, I can't see any compelling reason to change."

Fair enough. I suppose I'll stick with the recommended board then. My only guess as to whats going on is based on what some previous people have said: if i switched off the AC-DC adapter before i switched off the LED, the voltage being supplied to the boost converter would taper off from 12V to 0, and as the input voltage drops, but the LED demands the same output voltage and current, the working current through the board would shoot up asymptotically (to infinity??) as the input voltage drops to 0... Same if the switch to the LED is left open before the AC-DC adapter is powered up... the boost converter would be asked to supply 30v @ 3 A, but would have only 0, then 0.1, then 0.2, then 0.3 V to work with as the thing comes to life...Sounds like a recipe for disaster but i guess then the solution is to just only ever switch the boost converter and LED on AFTER plugging in the AC-DC adapter, so that full power is always available. I wish there was a way i could build a redundancy into this so that i couldn't forget to throw the switch in the right direction before plugging or unplugging the thing.. hm.

Anyways, thanks again EBP, i really appreciate all the help and commentary.

PS. Lasers took your eyes, so i say its time to get laser eyeballs. Swap out those plastic lenses with flat laser emitting diodes, and begin your overthrow of the world government. Vitreous fluid should act as a decent heatsink...


Joined Aug 12, 2014
I wish there was a way i could build a redundancy into this so that i couldn't forget to throw the switch in the right direction before plugging or unplugging the thing.. hm.
You might be able to do something simple to insure that the LED circuit is broken before the boost converter starts working too hard.

I'm just thinking out loud - nothing solid, but you can get SSRs that take mains voltage input. If you wired it up such that the mains power plug controlled the SSR and the LED line ran through the load side of it, then when you unplugged the main power cord, the SSR would immediately break the LED circuit, hopefully faster than the time it takes the two power supplies to decay and start fighting each other.

Theoretically, if the LED side isn't drawing a bunch of current anymore, the boost converter input shouldn't try to draw destructive levels of input current, right?

Or maybe the SSR should be switching the power from the 12V supply into the boost converter, not switching the output of the converter to the LEDs. I think either would work, but I'm not really sure.


Joined Aug 12, 2014
Variations on my idea above could be done with a mechanical relay (which might be ok, or might be too slow) or with a home made SSR (much cheaper, but requiring more expertise to implement safely.)