Texas_Ace should get a couple of same leds in a week. I would like to see dead test. I know they are good under 1A (better than XP-G3?) but probably far away over 5A.

Nice test! Thanks!

Thanks, those are great tests to evaluate how good these leds are!

Thanks for the test, indeed 1200 lumen at 4A is really good.
Strange things sometimes happen with led tests. I have done tests on the XHP70.2 that have a maximum at much lower current than TA and Koef3, I blamed that on for my (for that immense power) insufficiently cooled ledmount, but it could have been something else too, I never really found out. Perhaps in your case the thicker phosfor layer which also has a different composition is to blame.

Quote:

There’s just a hint of green on the highest mode, but I wouldn’t complain.

I don’t know if it is a “fair” comparison, but how are these comparable to the XP-G3 5000K 90CRI used in the Wuben flashlights? Are they less green or so-so?

Thanks in advance!

maukka wrote:

My review of the TO10R and TO46R: http://budgetlightforum.com/node/60934

With the XP-G3 the center of the spot might be fine, but the tint is all over the place and severely green on the edges. This can be remedied with d-c-fix and a minus green filter. The LH351D is much better out of the box. You don’t lose output but throw does decrease, larger emitter area = larger spot.

Thanks for the answer maukka! I have DC-Fix on my TO46R due to those shifts of the XP-G3 (even if they are better than the CW 70CRI from the stock lights). The DC-Fix did help to improve a lot. And that was the reason I was asking, to check if this would be less green. And you confirm!

This said, I am still fascinated by the 4000K one you’ve shown above in comparison
Let’s see if the FW3A will have it

The differences in Vf between the 70CRI and 90CRI are interesting and substantial, even at low test currents, where resistive effects are minimized.

That suggests to me that the 70CRI are using a shorter wavelength die to emit higher-energy photons. I’m not sure how commonplace this is for other manufacturers and emitter families.

maukka, thank you for taking the time to do all your testing. I must admit that I need to learn how to interpret your results. What must seem obvious to some is lost on me. Can you point me to a website that will help me decipher your CRI charts and graphs?

Well this is interesting. Did a test on an LH351D 5000K from MTN and got some impressive results quite different from Maukka’s here and very much like TA’s results on the 70CRI variant. I think Maukka’s test LED really must have been a bad reflow or poor performing mcpcb or overall bad sample of this emitter because what I’m seeing is output increasing all the way up to 8A. Since for now this is my only sample and I want to use it in a light I’m not going to ‘crash-test’ it. But here are my results. If you were to even remotely believe my calibration numbers, we see decent alignment in the <2A current range, but beyond that, there is significant separation. Almost looks like a non-DTP vs DTP comparison. I trust my relative numbers mostly. The only thing I can’t verify at this point in time is my meters accuracy (HS1010A) vs wavelength. If the CCT is rapidly climbing (which it honestly doesn’t appear to be), and my meter overestimates blue, that could skew my relative curve to suggest slightly higher peak currents. Comparing these numbers to TA’s values, it matches up quite well at ~80% the flux output and Vf about 50mV lower across the range. However compared to the datasheets relative output chart, my results overperform in the upper current ranges by 10 or 15 percent.

Interesting, I was thinking that they could have sent you a 70CRI one by mistake, but I see that they don’t sell 70CRI, so that’s unlikely.

MCPCB, reflow, meter… maybe a bad/very good sample ?

I don’t know. By eye, R9 is significantly worse than my Getian FC40 and SST20s, but then again most are compared to those two. I sliced it down to 1mm now and sanded it smooth to 1000 grit. I can compare to a <70CRI XP-L in an Armytek, but being better than that isn’t saying much.

MTNs reflows always seem pretty darn good to me, and I feel the thermal performance of my setup is very respectable. My test method doesn’t allow heatsoak and full stabilization at every current, but rather I run it at max rated current until values stabilize and then sweep low to high from there, moving as fast as I can record the values. Someday I’ll hookup a DAQ and collect more data with fast (2-3s) sweeps of the full range and then do that at 25C increments or so.

EDIT: If you look at the Vf values from Maukka’s test and compare to the 70CRI he tested, my 90CRI (I trust MTNe), and TA’s 70CRI, there is something odd about his 90CRI sample. Heat could explain it in the upper currents range, but it’s more of a straight offset across the range. Besides, I_ think_ Maukka used the same mcpcb for the 70 and 90 versions.

EDIT2: Oh and in terms of current density, it seems only right for the emitter to handle these currents. It has a die that’s larger than an XPL/XML/SST40 and has no bond wires to fail. It’s very nearly the size of an XHP35. Consider the XHP35 with a 4S/12V config tops out around 2.5A in a 3535 footprint, then 10A should be a reasonable top end for the 351D.

I am almost certain that it puts out more light at 7A than at 5A – i.e. it’s not declining past 5A. I don’t put much weight on my lumen values, but since they lined up so well with maukkas at <2A I considered them reasonable. If I’m determined I can get some spectrometer readings on this emitter, but I haven’t created a fixture yet for testing LEDs consistently, and when I played around with it once, the readings were extremely sensitive to LED position and surrounding environment. I won’t be able to spit out all the CRI data you guys do without post processing in some software that may or not exist (free), but eyeballing the curve should differentiate between 70 and 90 CRI.

I’m going to try a different setup in the next week or so where I do a non-integrated 0° intensity reading through an opaque aperture, eventually setting up a luminance test rig. We’ll see if it behaves the same there.