Thanks. Mine died prior to the reflow. They came pre-mounted on sinkpads from MTN. I only tried a reflow after they had half-died.
This is generally a sign of the LED overheating, since it worked up until it was almost at max power the LED was ok and it is not to blame for this, unless it just happened to be a weak LED.
If you have a good thermal path to the L6 it is strange this is happening. You are sure that it is getting full coverage to the L6? You are clamping it down tight? Not too much thermal paste?
I would check the tailcap current as you increase the power next time. Maybe the Vf is lower then normal and it is pulling too much power?
Was the L6 getting hot?
This was just the initial function test, both times. Thermal grease was complete coverage of the bottom of the sinkpad, with very little to no oozing around the edges. Screws were tight, holding the pad to the shelf.
Failure was within a few seconds of turning on, so the body and head didn’t have long enough to heat up. The second time it happened, it did get warm from being on lower settings before I ramped it up.
I also didn’t think to measure amperage since it was the first light-up.
I tried one more time, with another new led pre-mounted on a sinkpad. Worked on low power, ramped up a little, good, ramped up a little more, good, got warm, ramped up a little more. Poof. One side lit. No purple, this time, just stopped on the one side.
That’s 3 bad xhp70.2s, in a row. XHP70 is still going strong. On two separate MTN drivers. One of which runs the XHP70 just fine.
I ordered some replacement LEDs to reflow onto the sinkpads and will go for round 2. Possibly mounting to a cpu heatsink to test first, and rule out poor cooling.
Was the reflector and bezel installed to help press the mcpcb down?
Strange that this many would die like this, I assume it is the normal P2 bin xhp70.2 that most use?
Bench testing on a CPU heatsink is a good idea.
Yup, installed and tightened.
p2 bin 5700k emitters
Yeah, bench is the way i’m going next time. I’m down 3 so far.
I don’t know if this is what happened nkresho, but sometimes the emitter shelf is cut for a particular MCPCB and a “SinkPAD” with it’s 6 sided shape can hang up in the very corner of the wall and shelf. If this does happen, even when screwed down tight the bottom of the MCPCB won’t be seated against the emitter shelf.
Before using the thermal paste and all, make sure the MCPCB sits flat on the emitter shelf, even has a bit of play side to side. It would almost certainly be a problem with heat getting out, as I’ve built several of these as have other folks and they do 18-22A at the tail with well over 9000 lumens and are still running fine. The last one I built was done quickly, used a 20mm MCPCB and Richard’s driver, it made 9500 out the front lumens. It’s owner loves it!
I say it was done quickly because I usually cut the emitter shelf wider on the lathe to accept a larger MCPCB, didn’t have time for that this time though.
EDIT: When using a 20mm SinkPAD I “round” the corners on a file so it’ll sit down onto the shelf cleanly.
EDIT II: A pic to help clarify the rounding of the corners…
This was on the quick, so my apologies for the quality…
nkresho, it’s also possible the bottom of the emitter shelf has a slight radius and the sharp square edge of the Sink Pad mcpcb is hanging up on it, not allowing you to fully seat. Might need to bevel/chamfer the bottom edge of the board back far enough to make full contact with the shelf.
Other users in a facebook group are talking about this same problem today on a different host and several people are saying sideways pressure on the LED from the reflector and not having insulating gasket installed right can cause this. Seems less likely on an L6 since it’s made for this LED and should have a proper gasket though but food for thought.
Gasket was seated well on each try, so i don’t think that’s it, but at this point, i’ll check anything.
Thanks for the ideas so far. I checked all three dead leds on the sinkpads in the host. They fit well on the shelf and there is a little side to side movement. They also appear to be lapped pretty flat.
Worth noting, as well, is that this happened in two separate lights, and the one ran an xhp70 just fine as a replacement ( back to the oem emitter it came with).
Im rigging up an intel cooler with a fan to test out the new ones when they come. I’ll tap it to take 2 screws to mount the board under pressure and test with 5v power supply power, before dd’ing a pair of lower drain 18650s, eventually making my way to a pair of 26650s wired to it.
If it works, i’ll try again in my light and see what happens.
Hopefully they’re all just a bad batch. All are from mtn and ordered within a few weeks of eachother. If it works with the new emitters, i’ll reach out to mtn and see if others have had issues.
I tend to pick the simplest solution… user error on my part. Although, knock on wood, i’ve hsd some good luck with s2+ lights, triple setups, and a few others. Going to tweak a q8 this week too, following in a lot of y’alls footsteps.
Very mysterious, nkresho. From what you’ve said it seems like the LED should be well enough heat sinked. I remember several people reporting a similar problem with the XHP50.2; half the dies going dark. Do you remember which pair of dies died, and was it the same in each case?
Just so you are aware, it working with an XHP70 doesn’t really mean a whole lot. The xhp70 has a much higher Vf and pulls much less power (almost half the power in many cases).
It is a very strange problem for sure, I have torture tested several xhp70.2’s recently and had no issues at all, not even loss of lumens when retested at low output.
5V power supply to run a 6V emitter? Hmmm……
I will check the side that died and compare. And about the power supply, I was planning to give it 5v as an initial test, moving on to more voltage as it passed.
edit: checked the emitters that failed. With negative towards my left hand and positive towards my right, “sinkpad” text readable upright… First one, left side working, second, right side working , third one (to my surprise, both top and lower left working, just one burned out. So just one quadrant burned out.
Odd that one of four is dead, since they’re wired 2S2P. That again would indicate it’s not an over-voltage issue but a heat related one.
I haven’t tried to run an XHP-70.2 up with a power supply, but my power supply can be set for constant voltage or constant current… I use it in constant current and the voltage is regulated by the need of the Vf of the emitter. I myself have seen over 19A to a 70.2 for 9200+ lumens in my own L6. I didn’t get an emitter current reading on the last one I built, but it was making 9500 lumens out the front using Richards driver. I know Kawiboy has one being pushed in excess of 22A, so I’m fairly confident voltage and/or amperage is not the culprit here, also indicated by the tint shift to blue, almost always an overheat issue.
It is odd that one of four is dead. In the first generation XHP70 the dies were arranged as two 2s strings in parallel, so if any one die died (became an open circuit) that 2s string would not pass any current. But with this second generation it’s possible they electrically connected those middle junctions, so if one die died the other other 3 could still see current (becoming a 1p-2p arrangement).
My guess at this point is that these emitters are part of a bad batch in which some of the chips are not well attached to the package. We have seen these symptoms before with the XHP50.2.
Not really possible EasyB, because if the emitter is connected as a 6V emitter then 2 dies must be run in series to provide 6V usage, if one of those breaks a bond wire then both remain unlit. If the emitter is in 12V configuration, any one die with damage kills the whole thing.
Yes, I forgot about the 12V option. The fact that there is the 12V option means it can’t be connected the way I described.
Testing on a heatsink tonight went well. I reflowed two emitters to replace two of the three I killed. Did the pan on the electric cooktop method, slowly warming and cooling.
Drilled and tapped two holes in an intel heatsink (aluminum stock cooler) to match the cutouts in the star.
applied as5 to the surface and screwed down the emitter
first, tested with 5v, low amperage, from my power supply. Worked fine. didn’t measure amps, but it was not terribly bright.
then I put on my sunglasses and proceeded…
then, a pair of keepower 18650s (protected) measured ~10a with my clamp meter
then, a pair of sanyo unprotected 18650s measured about 12a with the meter
then, a pair of button top 18650 30q and measured about 14a with the meter.
no failures, no tint shift
even with the sunglasses i’m still getting a bit of a dark spot in my vision :disappointed:
ran about 30s and kept my finger on the heatsink to make sure it didn’t go supernova on me
heatsink got warm, but not uncomfortable, at 30s
next step, I think i’m going to lap the bottom of both stars. They wobble ever so slightly from being a bit convex on the bottoms. I’ll likely polish them up to 1000grit, or so, for good measure.
Then the big step, re-mounting in a light and seeing if I made another (how many steps was that?) mistakes
wish me luck, and thanks all for the assistance and ideas so far.
Well, that kind of confirms it was a heat issue, which is honestly the only thing that makes sense.
Very strange that all of them would of overheated though. Do you have some 26650’s to try?
Yup, i can try them too. Solder blobbed liitokala blacks and purple efests.
I’m a little apprehensive to put the emitters back into the host, but I’m going to try that if the 26650s pass.
I confirmed, again, that nothing is holding the corners of the star off the base inside the head. There is slight wobble though if i press a corner down.
Going back to as5 instead of the kryonaut too. I don’t see a lot of others using it. Although it’s the best for cpu blocks, it may not be up to the higher heat a light makes.
I’ll test out the 26650s after i lap the bases. I’m using a piece of glass with automotive sandpaper on it. This has been my cpu waterblock lapping technique for years and passes the razorblade test well.
edit:
not yet lapped, but tightened down on the heatsink to ensure contact.
15.2a with the efests, and 17.2a with the liitokalas both tested with two 26650s
I think i’m ready for the host.