Emitter data and talk. A look at Vf, amps and some crash testing (Updated with more data 10th of December)

Flashlight Modding and DIY Parts
1

Im going to share a lot of my story and testing around emitters lately. Warning, its not short.

But first, some info about the table and test setup presented in the spreadsheet.

All emitters were on Noctigon, a few on Sinkpad. When I first started this testing it was just for myself. It was not supposed to be super scientific. I just wanted to find some answers and learn more. There are some shortcomings to the testing, but it is what it is. :p Hopefully you can extract something useful out of this thread...

Test setup 1 did not have a light meter included. I did not have a capacitor on the output. In some cases the emitters were not secured as well as they could have (with screws). In test setup number 2 which were the setup I used for the majority of the emitters I took care of those things. Capacitor could have been wired better, but I choose a simple and convenient setup.

Ripple being the reason for fried emitters?

Could my powersupply have any issues with ripple? As far as I know, not much likely, but on my test setup number two I used a capacitor that should help or eliminate potential dangerous ripple. I can also add that I have not seen any patterns that indicated issues with my PS. I have had emitters above 8A powered directly from it before. And in the cases where I have recently fried emitters at lower currents (certain XM-L2 U2 and U3s) it seems other members have fried similar emitters at similar currents.

The driver I used for killing several of the emitters was the H6Flex. Ive tested many of these drivers, and have pretty much concluded that there are no issues with the output signal from any of them. If I have a capable emitters, the driver will push them to 6,6A if desired. The problem have been finding capable XM-L2 emitters.

Clink the link below to see some data from my testing:

https://docs.google.com/spreadsheets/d/1Un6aQIqoyfXQYb_wRz-3lNpsv06Q6fO9TFSMBzTnkwY/pubhtml

Ive got videos of all the emitters thats been tested and documented in the spreadsheet.

Here is a random example video where I killed an XM-L2 U2 1A from Aliexpress/Convoy/Simon to give you an idea:

My light meter results are just to see that output is increasing. Its not comparable from video to video and was never supposed to be.

If you want to see any specific video. Let me know, but they main data are in the spreadsheet.

Comments about the various emitters along with additional data:

XM-L

Just wanted to add one for comparison. No surprises to see that this emitters Vf was lower compared to most of the XM-L2s. Some strange behavior, this emitter did not want to go above 7A. If I increased voltage more, amps went down. This probably came from some China light. Either way, im not familiar with pushing these into the 7A range..

XP-G2`s

These had both been de-domed, and the phosphor was not perfect. I suspect that the R5 emitter I tested was damaged.

The S2 binned emitter I tested had very high Vf. Even higher compared to the XP-L. So it may not just be the top binned XM-L2 emitters that shows some signs of high Vf. I stopped at 6A. Output increased all the way, but not by much near the end so I stopped there.

Ive put a few dedomed S2 and many de-domed R5 1A in lights lately. Once the lights are properly finished I hope to verify if they show signs of similar behavior to the emitters I tested. Basically, see if the S2 emitters have much higher Vf and also compare output. For now, the data from the two not so perfect emitters will do.

XP-L

I threw in one XP-L as well. As expected. This had higher Vf compared to the all the XM-L2s. Despite the high Vf it handled 7A/ 4,27V. I did not feel like pushing it further. Im curious to find out how the higher binned V6 compares. I have ordered one several weeks ago..

XM-L2 CW «T6»

In the table you can see 4 different «T6» emtiters. I say «T6» because who knows what they really are as they came from typical "china lights". All of these handled high current fine. One had a bit higher Vf compared to the others, but that did not affect how far I could push it. My typical bench test emitter that I pushed to 7,5A have seen 8A+ on PS and through driver circuit several times before.

XM-L2 U2 1A

Issues with blow U2 emitters were the main reason for all this testing. Before I did some official testing I had killed several of these. I could not even push them to 5,6A. As testing shows that was the limit for many emitters. I have tested a lot more compared to what is seen in the table.

Lets go through some.

I ordered 5x XM-L2 U2 1A from Simon/Convoy @ Aliexpress. Ive blown three in various circumstances. Two of these are in the spreadsheet. I suspect all 5 have similar behavior.

I was sent 3x XM-L2 U2 by a member who bought them from RMM/MTNelectronics. Not sure if these are the exact same as the ones from IOS, but maybe RMM can verify. Either way. None of those 3 were capable of high amps either. One was documented in the spreadsheet.

Ive bought some TK75s. My first and oldest from around 6 months ago. All 3 emitters capable of high amps.

Ordered 2 more TK75s not that long ago. Ive blown one emitter from each one. I suspect the two last emitters in each light would not be capable of high amps either.

The only reason why I was able to come across some U2 emitters capable of high amps was because I had bought a 3rd TK75 used. This light was probably older. It had emitters that were capable of high amps. Data from this is in the spreadsheet. When that is said, the date on the TK75 mcpcbs were all the same.

In summary Ive tested roughly 20 XM-L2 U2 (mostly 1A) emitters from completely different places and probably from several different batches. Ive blow a lot of them since they have not been capable of high amps. I can add that Ive talked a bit with MrsDNF and been in contact with a few others as well in the past weeks, and Im not the only one who have had "issues" with the U2 emitters and high amps. I can add that MrsDNF have ordered some from Cutter, and did not get above 5,6A before another emitter was gone.. Seems like no matter which corner of the world they come from, similar issues can be experienced.

XM-L2 U3 1A

Ive only had 5 of these. I put 4 of these in a resistor modded Fenix RC40. Based on output I believed they performed better compared to XM-L2 U2 emitters, but I upgraded from 1st gen XM-L and I also changed current so not that easy to say. Due to the high Vf of the U3 emitters, the stock driver was not able to stay within maintain peak current regulation.

In my bench testing the U3 emitter had the highest Vf Ive seen in an XM-L2 emitter and it was not capable of more than 6,4A before id died.

Rambling, thoughts and such:


How will these emitter Vf differences affect direct drive lights and lights with a linear driver?

Higher Vf means lower peak amps, or that the light will fall out of regulation earlier depending on driver/battery/led configuration. I don't think that even the emitters that died at around 5,6A would have had issues in a direct drive light, simply because peak current would have been a bit lower with those. But the differences in Vf can easily effect peak amps when doing direct drive builds.

What about properly regulated lights (say with a good buck driver)?

As seen from the testing. If the driver is good for 5-8amps. There is a clear risk that the wrong emitters will die no matter how good the driver is.

How does the Vf differences affect actual output? Amps isnt everything..

Hard to say. I have not focused much on that bit when I did my "technical" bench testing. I wanted to try and get a better picture of what was going on with amps and voltage so that I could find the maximum limits for various emitters better. If I can push them to the max, then I can also push them to max output. At least that is the theory for XM-L2s and XP-Ls.

Everything was better in the old days..

..old days, as in 6+ monts ago.. Maybe they were better, things seemed more predictable then. I had not noticed as large differences in Vf (I had seen some tendencies though) nor had I experienced blown XM-L2s below 7A. I don't doubt that output from some of the latest emitters are very good and can exceed older emitters, but something seems to be going on. This thread has been on that topic as well.


Is there a good way to find out which emitters are capable of high amps?

If we look at the U2 emitters I have tested. At 3Amp there is not a lot of difference between the Vf of the emitters that died below 6A. But if we look at 5A, the emitters that died all had higher Vf compared to the one(s) who handled 7A without issues. With my setup and numbers there is a chance I can just look at Vf at 5A and determine which XM-L2 U2s are capable of much higher amps. If its around 3,7v or lower then Im probably good. If im seeing around close to 3,8V or higher then chances seems slim based on what I have seen so far with the recent U2`s.

Your setup may be show slightly different numbers. Enough to not use my data as reference. If you don't have a powersupply, then direct drive from a battery might give you and indication. Assuming you learn to interpret the numbers from your test setup and get a consistent way to test emitters..

Djozz have done crash testing as well

Looking at the XM-L2 U3. We both killed our emitters at quite similar output levels.

Djozz`s graph and data show: 6A and 4,2V (reported that emitter fried between 6-6,5A. Tint was 2A)

My data shows: 6,4A and 4,28V on moment of death for my emitter with 1A tint.

Id like to add that djozz have made some really good crash data in the past too when testing XM-L2 T6 and XP-G2 R5. My XP-G2 R5 data is not that reliable due to the emitter that was used, but our data on the XM-L2 T6 seem to match quite well. Which probably means that my XP-G2 S2 does have a very high Vf compared to the typical R5 binned emitters.

I think is fairly safe to say that data from certain emitters with certain bins may not be representative for similar type emitters with different bin and tint. (I hope that sentence made sense)

Please contribute

This is mostly just my data. Take it for what is (its not the greatest testing), and feel free to evaluate it yourself (look away from most of what I have written).

Just thought it might be useful and interesting to some. Hopefully others can add more data and experience to this thread so that we get a better picture of what is going on with various emitters. Or post about it in other suitable threads...

2

Thanks RaceR86. Allowing for differences in equipment both U2 and U3 leds I have tested from cutter and CNQG, I did not test any U3 from IOS to oblivion as I've spent enough and wasted far to many expensive leds, both had voltages around the 3.9 volt mark at 5 amps. None of the emitters I had in these batches were any good for more than 5.5 amps. I started measuring light output at the same time with new measurements but have stopped. l will have to do this when I have time and more leds from different known batches.

RaceR86 approached me awhile ago whether I had had any problems with the taskled driver and it turned out we both had similar problems. Again as RacerR86 has asked the more data we have gives us a bigger picture to look at.

3

Thanks for taking the time to put this together. :beer:

It would seem that maybe a recent design/manufacturing change/process has brought the overheads that we exploit closer to the manufacturer’s rated ‘maximum’.

4

My thoughts exactly and maybe cost cutting as well if the above is founded.

5

'Elementary, my dear Watson'

That was a huge undertaking RaceR, I would never have had the time to do so many emitter tests! Thanks for the hard work on this. So it looks like Cree has introduced a manufacturing change for the worst recently.

A question: the XM-L2 in my BLFX6 died with intact bond wires, that was the first dead XM-L2 I have seen that does that, I always saw fried bond wires before. How about the many dead XM-L2's from your test?

So we have to start hoarding old-production XM-L2's for our overdrive builds now I guess

6

Thanks for comments guys.

Djozz, what happened to the emitter in your BLFX6? You also mentioned that you had fried some XP-Ls over in the XM-L2 U3 thread? Any more info from your adventures?

I have seen various types of damage on fried emitters, everything from super obvious like the one in this picture that you posted:

To obvious burnt wires (old picture I had of one):

To the very subtle that you can not see by eye. (Old pictures of an XM-L2 U2 I believe)

In my two last pictures I believe its the slightly dark area on the bond wire that is damaged.

I did not bother to study the ones I fried this time. They quickly went in the trash without any pictures. I noticed that some looked like in your picture and some I could not tell by eye they had been damaged. I probably covered all the common types of damage when I was on my killing spree. :D

7

Thank you for the testing Racer. This must have taken quite some time to do!

A shame we are as of yet unsure about the output. A while ago I read somewhere that Cree was implementing production changes that should increase output approximately 20% but that they were not going to change the binning system or generation numbers since it would be across the board of current production led's. Maybe the side effect of that is giving us headaches like the above. I will just add that I never pursued the information to verify its source. Eg. I don't know if it was from Cree's homepage or some lesser source.

8

Thanks, I will have another look at the BLFX6 emitter, perhaps the bond wires were damaged after all, in a less obvious way like you show above.

I fried the BLFX6 one by foolishly trying out 2x18350 cells in series (surprisingly the driver survived). The bond wire damage may be different than I am used to because in a emitter test they fry while already warmed up, and at just over the maxiumum current they can handle, and not way too much all of a sudden.

The XP-L's were two 7A1 80CRI ones from intl-outdoor, I tested them out before use in a flashlight (they never came that far :-( ) because I wanted to repeat the test I did earlier with a cool white XP-L that showed no output loss after dedoming. But they both died under 7A, so no dedomed test was possible.

9

Oh no. Some XP-Ls cant even breach 7A now.. :_( I had hoped the XP-Ls could be a safer option after having so much issues with some of the XM-L2s. Did you get any voltage data when you tested it?

I hope my incoming XP-L V6 2C will not get the same fate as I plan to test it.

10

BTW, last year I specifically bought a descent old-tech heavy lineair power supply because of rumours that switch-mode power supplies could suffer from voltage spikes (which I do not believe is true, but I'm an electronics noob, who am I to claim anything ), so the chance is minimal that spikes are a reason for fried leds in my tests.

11

Good thread. So for our high current builds, we should record Vf around 5 amps (for xm-l2), then report here on the measurement and ultimate result (fried, didn't fry & x amps, etc)? Would it be hard for you to set up a collaborative spreadsheet like the LED database?

RMM reported frying like around 20something xml-l2 working with the HX-1175b driver. Not sure if he has any data of the emitter Vf's. I think he did see issues with the driver's output on a oscilloscope. So it may not be relevant.

12

It was suggested to me it could be voltage ripple ImA4Wheelr. I could not confirm this.

13

So where is the best place to buy old XML2 emitters :frowning:

14

MRsDNF wrote:

It was suggested to me it could be voltage ripple ImA4Wheelr. I could not confirm this.

Yes, my understanding too. Just wondering if this apparent emitter failure issue was part of the problem too.

15

RaceR86 was using the same driver but a different power supply in testing. Using his results some leds will run higher than 6 amps with his power supply where as some wont. My gut feeling for what its worth is that the leds have changed not allowing the higher current and this has come about with the new generation of XML-2s and 3s that have the higher forward voltage requirements. This is unsubstantiated though.

I went through a heap of leds before I found 3 that would run at 6.6 amps before RaceR86 contacted me.

16

Just measured a XM-L2 U3 1A and 2A.

3A: 1A, 3.34v. 2A, 3.64v
4A: 1A, 3.50v. 2A, 3.85v
5A: 1A, 3.68v. 2A, 4.10v

Both on 20mm Noctogons, 18 gauge wires, sinked to a 4x4x5 chunk of Aluminum. Power supply: Kenwood PD18-20D.

17

This is where I get my XM-L2 T6 emitters from I have good luck with driving their emitters very hard. I don’t have the testing equipment to say exactly how high, but high enough that it will kill many XM-L2’s.

18

Thanks Racer that is some fantastic information.

I have a 7amp driver showing up soon for a TK61. I’ll try the original emitter that is in it first and if it kills that one I’ll put in one of the XM-L2 T6’s I get from aliexpress. I have had good luck with them.

I have killed many emitters mostly from shorting them but also my fair share by overdriving them. I was wondering why I could get some that would die a quick death while others seemed to handle whatever I threw at them (within reason).

I might just have to stock up. In fact now I wish I had bought more when they had them on sale for 11/11

Thanks again Racer, great stuff!

19

We need to show that these leds in question are putting out more light at the same forward voltage as the older style XML-2 if that makes sense. All may not be lost if this is the case if running in a light with two or more batteries in series. They just cant run at the higher current but will have more light output. Ok, I'll be quiet now. I know what I'm trying to say.

20

How about dedoming and then putting something on the bond wires to cool them or to reduce their resistance? Oil if one can find one with a high enough boiling point, or something like Arctic Silver. Best would be something metallic that would reduce resistance as well as conducting heat well. There does exist paint with very high silver content so that it is a good electrical conductor.
As cars became more thoroughly engineered they were harder to hot rod.