Please help me determine vF for XM-L2

I just received 50 XM-L2’s from our good buddy Hank and need to weed out the 10 with the highest vF.

I dont have a bench power supply and wonder if I can reliably determine vF with a wall wart that delivers 2.847V resting. Ive used this small power supply for ages to test LED’s to make sure they work before dropping them into a flashlight… and its sure saved me a lot of aggravation.

With this power supply and the following two tests, can we reliably determine which LED has the lower vF?

Emitter 1

Emitter 2

As you can see, there is voltage sag when connected. The tests results were repeatable. If this isnt a reliable way to derive conclusions, what would be the next best and easiest method?

Ultimately, these will end up in my light bar project. 40 emitters will be driven in two strings of 4S5P at 16.9A each, so 3.38A per emitter. Do I need to test each emitter at 3.38A and record their voltages to know for sure which ones are the lowest vF performers?

I would do it the other way 'round- give them a fixed voltage and then measure the current.

Vf is a curve and not a straight line, so one that's good at very low current might not be at higher current. Test each one at several different points to get an accurate idea of what's going on.

i agree with fixed voltage because you dfon’t have to deal with voltage sag issues, but unless the OP has a fixed voltage source or can get one cheaply
testing them at diff voltages is great but will vastly multiply the work involved

Thanks, but I think you’ve got that turned around. You’d want to set the PSU at a specific current and measure voltage to determine vF. Im still not convinced that my initial tests wouldnt be useful in helping preclude the highest 10 vF in the 50, especially since they are all identical emitters/mcpcb’s.


I’ll see if I can find someone with a bench supply they could lend me. Grrr! This is going to be a pain!

Current at a given voltage, or voltage required for a given current, it's all the same. Take two LEDs, give them both exactly 3 volts, the one that draws higher current is the one with lower Vf.

That be my thoughts on trying to find the lower vf.
How are you testing them flashPilot? On a Heat sink, mid air, thermal paste. Are they all measured at some exact time after being connected. I would think that differences in time or some of the other factor’s could play a small part in vf differences.

Thanks for the replies. For testing I put the emitters on an aluminum plate and test with dmm probes. The test current is extremely low (check post 1 for results). I triple checked and get the same results each time, so external factors havent seemed to cause any variables.

Would it be safe to say that Emitter 2 has the lower vF? If so, would it still have the lower vF at high current? Keep in mind that all the emitters are identical and presumably from the same reel.

Ive searched for information on vF curves tested from several identical XML’s to see if the vF plot would remain linear for each emitter in the batch… not the same but linear. I couldnt find anything. My ultimate goal is to pick 40 of the lowest possible vF emitters @3.4A and use the remaining 10 in other projects. I dont have a way of taking precise measurements at that high amperage right now.

Thank you all for reading.

What about 1 or 2 7135s and a 5v/1A wall wart? Record both V & A for each LED (with enough voltage overhead the current shouldn't change, but the voltage to achieve that 350 or 700mA will vary depending on each LED's Vf). I would really want more than one data point even if it's a lot more time consuming; if it matters enough to sort through them, you should want to be sure your data isn't misleading you.

Just a question of a noob, when it comes to VF or whatever:

what kind of difference would it all make? Would the eye perceive the difference?

Thanks comfy. If I knew that vF would remain linear throughout the curve, Id have my answer. But since I dont, it looks like Im going to have to come up with a way of testing low and high currents. Your ideas are good for testing low current. I’ll put the word out amongst friends and see what I can come up with for a bench psu. I was trying to avoid working with high currents so I wouldnt have to deal with other variables, such as heat and momentary electrical contact issues. These will certainly increase the chances of recording errors. I may drop by the other place and see if anyone has any experience with identical XML emitters before I proceed. Im also waiting for the opinions of the driver engineer… he’s a lot like our own pilotptk in his knowledge and enthusiasm about the subject.

ChibiM, the lower the cumulative vF in my emitter strings, the lower the input voltage can be to maintain regulation with the linear driver I will be using. Since this will be in an automotive application and power supplied by an alternator, I want to stay below 15V input, ideally 14.4. Considering the total current load, this is going to be quite challenging since the alternator wont start producing any real output below 1500 engine rpm. Every little bit of resistance lowered in the system combined with the lowest possible vF at the emitters will put me that much closer to reaching my goals… keeping the driver in regulation.

Thanks for explaining FlashPilot!

I understand now you are looking for the lowest possible vF, to have it work with your new project...
But all the other info is still a mystery to me... (it means, I don`t understand it ;) )

With LEDs in series it doesn't matter much about variations in Vf among the individual LEDs, but (in EE-world, at least) with LEDs in parallel (or multiple series strings in parallel) variations in Vf will result in the higher Vf path drawing less current, and the lower Vf drawing more. In parallel, all LEDs (or all series strings) get the same voltage, and the voltage applied determines how much current will flow to each LED (or each series string).

Say you have 10 LEDs arranged into two strings of 5 series LEDs, the two strings connected in parallel. By some random chance string #1 happened to get all the low-Vf LEDs, and string #2 got all the high-Vf LEDs. The difference between any two individual LEDs is small, but add 5 together and the cumulative difference can get large enough to cause problems.

Have you measured them all with your wall wart method? I'm curious how much deviation there was.

As Comfy stated since they will be in series strings, the string Vf is what matters. Maybe test them in as strings in series and see what the deviation there is.

Frankly, since they are from the same reel I think you should be ok just randomly hooking them together. The probability of getting 4 high Vf or 4 low Vf all in the same string is exceedingly small.

Here is a graph that I made using djozz values. Vf is the blue line. I would say that Vf is linear(ish) after you get past 1 amp. If you want to do a valid test of each emitter, I'd just hook up a 18650 and Qlite and cycle all of the LED's through your setup. Put them in 3 or 4 groups according to what you find.

Testing with a battery is a Bad Idea, if you want the data to be meaningful. Even with a very short run for each LED by the time you get through all 40(!) you'll have a very different power source than the one you started with.

I don't think I'd be able to live without those neat little LM2596 converter boards... one of those plus a 12v wall wart can do wonders for jobs like this.

i was thinking about the battery too, as long as he recharges it after a few minutes of running he should be fine, say 5 minutes of battery power time

Doing this kind of curve with more resolution looks a bit different:

Place the 50 stars on aluminium, wire them all 50 in series (small solder job ) and find a way to feed the string 150V DC, you know that every led then receives exactly the same current (about 1A). Then measure the voltage of the individual leds. Oh, don't forget to take pictures

and update your fire insurance?

I dispute that statement. Just because the battery voltage sags does not mean that the test cannot be meaningful. It MAY not be meaningful. But it could also be just fine. We are using a driver after all.

I decided to test my theory. I did two tests.

Test 1: LED on constantly for 6 minutes.

Samsung 20R

Battery Pos to Qlite spring. Battery Neg to Qlite outer negative ring.

Multimeter hooked up across the XM-L2 LED in parallel with leads from Qlite.

Voltage across the LED drops .081v over the 6 minutes. That would be significant, but could probably be accounted for. It also assumes a long time with power applied for readings.

Test 2: Discrete measurements with power briefly applied, removed, reading recorded, and a short (5-10 sec) rest before the next reading

Setup: Same as first.


Vf varied a total of .02 Volts over the 40 tests from 3.31 down to 3.29. That is not significant. I read the first two decimal places when the 2nd place was steady. If it flashed 3.30, then quickly to 3.29 I used 3.29 as my reading. All readings were probably under 3 seconds under power.

I conclude that the 2nd test is an acceptable way to test Vf over 40 LED's for this purpose.