It’s not obvious from the data but I tested two LEDs. The second had a few percent higher output, but identical relative output and Vf. The first suffered a burn in the center of the phosphor from the 7+ amp torture. The second sample was then sliced with a razor blade using 1.1mm shim against the PCB. Output dropped 13% at all current levels and Vf and peak current didn’t change within the noise of my measurements. These things turn blue in a hurry. I put a triple in an FW3A and turbo is insta-blue with an aged red HE2 @ 4.06V. Under both the stock Carlo frosted optic and KDs clear spot optics, the beam is quite pleasing and tint is acceptable down low and wonderful above the mid ramp flash. The periphery of shadows against a white wall still show the tint shift, but that’s a detail only the sickest of us would notice, let alone care about. I will say it’s a total NO-GO in a reflector with the dome on. Terrible green corona. Sliced at 1.1mm it’s unobjectionable, but maybe not the best either. Maybe on par with a domed SST20. I have some crappy pics that I’ll post later, but for now take my word.
I put the data in a google doc and will share it once I clean things up. I’ll be adding other emitters as time goes on.
As for the test setup, I put it together in a hurry from scrap parts laying around. LED was mounted to a CPU heatsink with a fan running. No screws/clamping method were in place because it wasn’t tapped and I don’t have time right now for that. Future improvements will include that as a top priority. I compensated for this by using Arctic Alumina and spinning the mcpcb with downward pressure until a ‘vacuum seal’ was created. Additionally, I was careful about my test method and how long I let it run at high currents. Basically a lux meter (HS1010a) inside a styrofoam cooler with a rats-nest of baffles/diffusers made of other foam pieces. I worked on it for a while to get it mostly insensitive to beam pattern and angle of the light source. It’s far from anything I’d be proud of, but it wasn’t supposed to be. I compared the values of this test to Lumileds datasheet and it seems spot on honestly. Vf is reported from the Power supply digital readout, then compensated after the fact for the measured 174mOhm of resistance. A shunt resistor and DC clamp meter both verified the calibration of the current readings.
Sounds like how I started testing, and to be honest, the simple hacked together first setup gets you almost all the information you need about a led. Later I made more sophisticated test-rigs that were more accurate and better calibrated, but the extra information that it provides compared to the first setup is limited. You want to know the voltage, a rough idea of output, and at what current the maximum output is, and that is what you did.
So I did something I’ve been curious about and had theories forming on, but have never seen data for. I felt that many emitters, especially warm white and high CRI versions were having their overdriving capacity cut short by the phosphor and/or silicone matrix, and not the underlying blue chip. I feel it’s true in the vast majority of cases that the LED turns blue before dying via burnt bond wires, etc. The way the HL2X behaves and the fact it’s a bond wire free design, I figured it’d be a great candidate to start with. So here it is…
I did this quick and dirty as well. Something that I observed that I don’t have captured in the data is the lack of thermal sag up to around 5 amps. Turn on to 30s was pretty stable in ranges that the stock emitter was already maxed out.
Great experiment and very clear result. It is consistent with the white and green WF that djozz measured. The green LED peaked at a higher current and we were speculating it could be the different phosphor, since both had the same blue die.
I got my 4000K 90CRI emitters earlier this month. I just reflowed one into my seeknite Titanium AAA light and… Well, all I have is disappointment. Even in the TIR the thing is just hideously green. Maybe these are good for someone with red-green colorblindness.
I cannot overstate how disgusting the tint on this thing is. I’d rather have installed my old bad 5000K 219Cs, or 6500K SST-20s. I’d rather use XP-G3.
If anyone wants the other nine that I ordered, I’ll send them CONUS for free.
I just got my 4090s and 5790s two days ago. Haven’t had time to evaluate them, the 4080s were green, but nothing worse than a 351D. But they had terrible tint shift with the domes on. Sliced it wasn’t all that bad honestly, night and day difference.
Terribly disappointing overall though. I guess that’s where all their lm/W comes from.
I had preordered a bunch of these and had them sitting around for a while but after reading the above impressions I didn’t bother with them. I finally decided to test one. I’d go ahead and stay away from these.
This is the 4090 variant (P/N L1HX-4090200000000). Tested in an S2.
Oh boy, dunno. From all these results, I still like them in a triple TIR of the EDC18 - one of my favorite triple TIR light tints. Weird, maybe I don't see the green that others do. Should take white wall pics of the beam. The green in the SFT-40 was not really noticeable until I took whitewall pics of it, then I could see it when I look for it.
