At what current do you kill LEDs - which driver was used?

Curious what emitter currents you kill various LEDs at when using them in flashlights.. Wondering if some consistenly are able to push LEDs higher than others (assuming their reading is fairly correct.) And if there is a pattern between the driver that was used (current spikes) and if the LED was domed or dedomed.

For people who share their setup, im only interested in emitter amps. No guesswork based on tail cap readings and such.. And please, do not let these deaths put you off from killing your beloved LEDs pushing the limits...

XP-G2- RIP list

RaceR86, 5,2-5,3A. Dedomed. Driver: STL-V6/TR 3T6 driver with resistor mod. The one where you can change between two groups (3/5 modes). Copper mounted LED, secured with screws to copper "pill". Less than 1 hour of total use.

Gtaamazing, 5,2A. Used an SST90 driver. Reference

XM-L2 - RIP list

ImA4wheelr, 7A. dedomed. MF TR3T6.

RaceR86, 6,3-6,4. Dome. MF TR3T6 (artic silver 5, screwed down to ZY-T08 pill. Less than 15 minutes of total use. )


Other info:

I can mention I have never had issues with XP-E2 up towards 3A. MT-G2 at 10A, and Nichia at 2A. There is really no point in pushing those emitters further (or in some cases even that far). I dont see the point in pushing an XP-G2 above 5A either (based on graphs). But the XM-L2... It just keeps getting brighter until it dies. We need more info about that sweetspot, and if some drivers are better suited than others..

For the record, I have never fried an XP-G2 or XM-L2 with the FL-2 or similar driver at 5A.

Here are some of the crash testing by djozz. XM-L2 and XP-G2

High current drivers


Some pictures of my latest LED killer..

LED 1. I assumed it fried due to lack of pressure towards pill. So I put in another one..

Nicely secured...

Stopped working properly after some use.. When used on high it shuts off after 1-3 seconds (its the LEDs fault)

Slight sign of burnt wire?? Either way its damaged and only working at "low currents". I consider the LED broken/junk..

Not sure what my run time was. Probably less than 1/2 hour total. I believe I was close to 7 amps actual as I ran short soldered 18 gauge wires directly from emitter and driver into DMM.

Other info. Noctigon, reflowed with silver bearing solder to 3/8" (9.5mm) copper bar. Copper bar was about 1" (25mm) long through solid aluminum pill.

Did you adjust down to around 6,5A? If so, how is that working for you?

I haven't yet and now I'm worried about trying it after hearing your result. I have already fried 3 emitters in this light. I may go down to 6 instead now. Well, when I decide and do it, I'll try to keep track of run time and report back.

EDIT: Those are crazy amazing macros in the OP.

I wonder if its the amps or if the (modified) TR-3T6 have some current spikes that we are not seeing... I though 6,5A should be okay..

Ill aim for less than 6A on next attempt...

I burnt 2 xml emitters on noctigon pcb's. Both dedomed.

First at 8,4 amps and second at 8,1 amps. Both died when the light was dropped or bumped hard. They had both been working fine for many months before that. Cannot accurately say anything about runtime.

EDIT: The pcb is NOT Noctigon. It is a DX copper pcb with the dielectric layer under the center pad scraped off. I just remembered. It is sitting in a Trustfire X7 where it is soldered to a second, much bigger copper disc that is tightened down inside the head with several screws and DX thermal glue (not Fujik, the other one that hardens). I have been wanting to change it to a noctigon but... It seems I only think about that when I need to use it. Not when I'm done :-)

I think you are on to something about that driver you both have been using, Mine were running off stacked 7135 drivers with 2 sanyo inr cells in parallel. Both times it happened within a few minutes after turn on with freshly charged batteries.

I changed to XML2 but cannot push the amps that high with that setup. Same as everyone else it seems.

Also burned a load of the XR-E emitters of various bins. About 2,1 amps comtinously in small light hosts. Mostly zoomie lights that have many, many running hours on them.

A copper mounted XR-E R2 direct on copper pushed 3,1 amps for the longest time but is now lingering. The dome has fracture lines in it. Maybe from the heat. And the phosphor layer has started to look decidedly brown in the middle. I consider that led dead.

Interesting thread :-) good to see what really happens inside flashlights. I'm actually positively surprised how well those high amp emitters survive over time in general.

I have not fried any emitters inside a light yet, the highest current I use is 5A for a xpg2 and 7A for a mtg2, with no serious runtime.

I never fried one from high amps, but I'm a bit confused here... If you fried an XM-L2 at 6.4A on a well mounted Noctigon, how are all those TN31vn's working at supposed 6.5A to the LED? I haven't heard of any that failed yet. I know they are on a much bigger copper star, but? I've also done several TN31's the same way (jumpered the resistors and have a K40 done the same way), but could only measure 5.5A to the LED, and my lumens and throw seem a little lower than the TN31vn's, but not by much.

I'm not doubting anything here, and I think there's a plausible explanation, but I'm also very interested in the same thing - finding that magic best high amp configuration and hoping to get a bit more than a 1/2 hour of a lifetime from the emitter... Smile

My latest thought on a high end mod is to take a 3 parallel light like my 7G9 and drive it crazy high on a super stacked Nanjg (or 2 Nanjg's wired up), and see if I can get to over 7A using Sam 20R's. Suppose this could all be tested first on the bench. I was thinking the LED would go poof over 8A from djozz's tests though. Looking over djozz's graphs, I'l thinking I can't get the voltage up high enough to obtain 7A though... Frown I suppose the best way to get to high amps with an XM-L2 is buck driver with cells in series.

The TN31 driver is something more than just a CC driver. It also regulates voltage somehow. An MT-G2 can be driven by the 3T6 driver whereas the TN31 driver can't even turn one on dimly. So maybe the TN31 driver prevents voltage spikes better.

Myself, I can't seem to get my TN31 into 6 amp territory. I haven't tried to pull all the stops though as it delivers 400kcd and I feel I should leave good enough alone.

EDIT: I also have trouble getting consistent emitter current readings on the TN31. It fluctuates for some reason.

Well vinh did a ton of TN31's and K40's by bridging the resistors, and I've done a few TN31's and one K40, and no reported problems, so far that is... I measured 5.5A at the LED, and forgot, but maybe 3A or higher at the cells (3 in series), so there's a lot being lost somewhere, and not greate runtime on high for sure - just the way we like it Smile.

I did same as you probably recall. Did most of the mods to reduce resistance too. I couldn't get reliable readings, but I would say they centered around mid 5's for me as well. I haven't had problems yet either.

So RaceR86. I know this thread is about LED deaths, but for sure there are folks running dedomed xml2's in TN31's at 5.5 amps to the emitter without failures. Supposedly there are others running in the mid 6's without failure, but I know of no confirmable cases yet.

Ledsmoke. You seem to live up to your name. Great info! :)


In terms of extreme output to an XM-L2/XP-G2 direct drive/linear driver is like hotlapping. You can get very good first lap/qualification. But you will never perform good in a race..Your second lap will be slower.. In the middle of the race you will be slow compared to the fast guys... And with these high currents, it does not take that long to get to the middle of the "race" if you get my drift. It does not seem like a direct/linear drive setup poses a threat to an XM-L2 and especially not to an XP-G2 even at peak.. 2-3 cells in parallel helps a lot on peak and maintaining output, but still the batteries will always be "fighting the Vf requirements"..

The TN31vn are the main reason why I have gotten the perception that 6,5A should be just fine. So I always assumed 7A should probably be okay... But I have never seen any emitter measurements showing 6,5A on those lights. So either those 6,5A are a bit inflated, or I would assume that the TR-3T6 driver have some current spikes that kills the LED since both me and ImA4Wheelr are playing mass murderers at what our DMMs says should be 6,4-7 amps. Makes we wonder if peak current should be lower than what I used to assume...

The main reason to know about LED deaths is to know where the limits are.. So confirming what works is also helpful... 5,5A to an XM-L2 should work if everything is done properly.. it seems... So how about 6 amps? and is 6,5A really too much?? Can 7 amps work nicely in a light? How does max output effect emitter life, and where is the sweetspot where you can be "safe" and avoid death?

We need more information before drawing any conclusions.. Best way to learn more about this is for various people to share their experience so keep em coming.. :)

Thinking vinh sold about 30-40 TN31vn's, saying they are 6.5A to the LED. Several posts here on BLF say they are 6.5A but I don't know of anyone that confirmed vinh's measurements. I have emailed with Wayne (vinh) and he did on one early unit, measure 6.5A, but not sure if he confirmed that measurement himself or not - I'm sure he didn't spend the time to check every single unit, because he must have gotten out 20 in just a couple of days. I think I got a good, solid 5.5A I checked and double checked - of course I was disappointed because I thought the mod was 100% exactly like what Wayne did, but dunno for sure I suppose - something could have been missed or forgotten something. I did a lot of tweak'n and checks to try to get it higher, but all attempts failed...

Hope I'm not derailing the thread, but it's all kind of related. If the TN31vn's do run at 6.5A and have been doing so for weeks and weeks now, and in volume, then maybe there's something to that super-sized copper star in those beasts that's saving the LED from getting a beating.

Edit: Ohh, and truthfully, I think the 6.5A is real because to get over 1,700 lumens from a de-domed XM-L2 (and a throw of 500 kcd) is unexplainable any other way.

Relic38s testing show basically zero difference between 16 and 20mm copper star. So I don't see how 20mm vs even larger should make a difference. Pill could make a slight difference at those extreme currents, but Ima4wheelr had a good pill it seems.

There is really no reason to doubt vinh.. But something does not add up. Again, makes me wonder if the 3T6 driver have some current peaks or something, which was one of my main reasons for starting this thread..

All information around the subject is useful information. :)

A start-up or even shutdown spike could be happening that could be damaging with those drivers. Those TN31 and K40 drivers are pretty amazin.

Ohh - vinh also did maybe 20-40 K40's the same way, so lots of units in the field, and I don't think they are all getting pampered. Wondering if he's getting reported failures - I'm sure he'd hear about it, but I bet most guys wouldn't go post about it - just common courtesy... My small handful of units out there are probably not getting stressed too much...

Boy, if vinh doesn't read this thread and had time to post, maybe I'll email him - would be good to know because he has the biggest volume of high amp lights out there by far, I'm sure...

My super shocker mods were measured at 5.5A to 6A, but at the carrier, not at the LED's, so can't really compare, plus they lose amps pretty quick, while the TN31 and K40's are truly regulated.

Yes true about the equivalent results between 16 and 20mm stars, but think he measured just output, and not "stress" or lifetime - pure speculation on my behalf...

I used this driver http://www.fasttech.com/products/1612/10005161/1404708-tr-0124b-84v-5-mode-led-flashlight-driver-for-lumi which I think is a TR-3T6 copy, thanks to the RaceR86 Ima4Wheelr 10ohm pot mod to a XM-l mounted on aluminum mcpcb thermal glued to a large computer aluminum heat sink and it died at a little over 6 amps. I then tried a XM-l mounted on a sink pad to the same heat sink and it died at 6.5 amps.
.
I could try killing another XM-l and see if there is a current spike with the TR-3T6 driver. The 87V has a min max function that records the min and max. Only thing is I don’t have any spare XM-l’s to serve the death penalty on at the moment, already served the sentences. R.I.P.

Stock defiant super thrower led on the standard pill (l use this for testing drivers) blew in a milli second at 7.5 amps.

I did a decent job on my pill, but that TN31 base and head is impressive. The base is not only much larger than a 20mm Noctigon, it also thicker. I have to be honest and say that my setup isn't even in the same ball park as the TN31. That being said, I do agree that the driver is probably the problem.

Side note, but maybe related. I see a thermal pad gap in many Noctigon pictures. That can't help when we're pushing emitters this hard.

If I can figure out how to mod an SST-90 driver to <7 amps I'll be driving an XM-L2 in that range for my current project. I'll let you know how it turns out.

Which SST-90 driver and in which light?

Im considering to modify the driver from my lustefire 3xU2 a bit and use that instead of the MF TR-3T6 (I think its similar to a certain type of SST-90 driver) I should be able to make it work in a ZY-T08 and make it deliver the amps I want.. Its not on the top of my "mods to do list" though. So while I take my time.. Feel free to share.. :)