Luxeon Q, observations and testing

There has been some talk about this newest Luxeon because it has the 3535 footprint so it is a direct swap for xp-boards. Mouser started selling these about two weeks ago, so I borrowed my girl-friend's credit card (I do not own one, and she is a very kind and understanding person :-) ) and did my first (and only) Mouser order. Ordering from Mouser was even more expensive than I already thought because dutch taxes appeared to want their share as well :-( So no Mouser orders for me from now on.

But on to the led:

The test-led was a 5700K 70 CRI Luxeon Q . The die size is 1x1mm, placing this led in the 'xp-e category' , so I will directly be comparing this led to the XP-E2 R3. However, when the leds are visually compared, the LuxeonQ die appears larger than the XP-E2 (but smaller than XP-G). This is caused by the phosfors being further away from the die (part of the light comes from outside the die-area), this is because there is a transparant layer between the die and the phosfor, so a significant part of the light goes sideways. But perhaps also the optical properties of the dome plays a role. (picture shows xp-g top left, luxeonQ top right, xp-e2 bottom, very low current, actually in series so they see the same current)

the dies from above picture next to each other:

Close-up of the led showing how the light is going sideways as well:

If you can 'parallel' your eyes here's a stereo pair, for 3D vision.

The die from above shows quite a lot of blue, but when the led is illuminating things, the tint and colour reproduction is ok.

Of course what is most interesting for us flashaholics is the performance, so I did an output test, the same way as I did a few other led-tests before (which is: not overly precise but I'm confident that the general picture of the test is correct) . The led was reflowed onto a 16mm Sinkpad (mind that the performance at the higher currents on a traditional aluminium board will be poorer than I measure here) the board was mounted on an abundance of aluminium, a reflector placed on top and it was directly connected to my humble power supply. Current readings came from the power supply, voltage was measured over the (thick) led-wires by a DMM, ceiling-bounce lux-readings were converted to estimated lumens coming out of the reflector, so the numbers can actually almost be achieved in an actual flashlight. (the picture shows the reflector next to the led, during measuring, the reflector was placed over the the led).

For comparison I did the same test with a XP-E2 R3 (from IS) on a Sinkpad. I could have compared the results with the XP-E2 data from match, but I thought it would compare easier (and fairer) if I compared them with the same set-up. The XP-E2 results I got compare quite well to match's results by the way (the lumens that I measure are less because I do not measure bare led lumens in a integrating sphere like match, but out-the-reflector lumens, so there's been already a light loss there)

Here are the results of the LuxeonQ and XP-E2 measurements:


-the output of the LuxeonQ up to 1.6A is very similar (a bit lower) to the XP-E2, but is much better at higher currents, peaking at about 4A, the thermal management of the 1x1mm die is excellent! But the performance is significantly less than a XP-G2 (see my XP-G2 data from a few months ago). So everything, 'virtual' die size and output, is a bit in between XP-G2 and XP-E2.

-the led voltage is low, lower than both the xp-g2 and xp-e2 at 3A, and also climbing slower than the xp-g2. So the led is easily regulated at all currents with 7135 current regulators and a single li-ion.

-although already far above the current with maximum output, the LuxeonQ survived 6.1A without noticable damage. The XP-E2 died at 5A, only very little output was suddenly leftover and a gloomy dark spot appeared in the beam on the ceiling. The bond wires were intact, but the die only conducted very little current with hardly any output and showing dark areas:

Conclusion sofar is that the Luxeon Z has a output performance in between the XP-E2 and the XP-G2, but because the virtual die-size is also in between those, it does not outperform them, at least not much.

This would change however, if the LuxeonQ can be dedomed and have its superior performance over the XP-E2 straight from its 1x1mm bare die (the luminance of the die at 4 amps should be unsurpassed). That way it could significantly outperform both the XP-E2 and XP-G2 in throwing capability .

So off I went trying to dedome the thing. First I tried the gasoline method, because that is the most common way to dedome (at least here at BLF :-) ). I used petroleumether, but that does the same as gasoline. Pictures at 0 hours, 6 hours and 30 hours (that does not look right ):

So at this point I pried the dome off with tweezers. Here's the carnage in 3D:

One led down :-(

I usually do hot dedoming, but as you can see, it looks like the phosfor is embedded in the silicone, so I would not expect the result to be different.

So there was only one method leftover: slicing the dome off with a razor blade, I used a paperclip placed around the led as a guidance and spacer for the blade (the paperclip had the right thickness to leave a very small layer of silicone (0.2mm?) over the die and slice the top off the dome. I had one C8-pill leftover so I reflowed the led board in there, with a qlite driver (3.04A worth of 7135 regulators).

I put the pill into my Uniquefire UF-T20 aspheric, that does 108klux at 1meter with a dedomed XP-G2 on a qlite. I measured the throw, and ouch , only 55 klux at 1 meter, and what an ugly hotspot (in contrast, the beam from the LuxeonQ in a reflector light is very nice):

I expected a lot of light reflected to the side without being reflected back into the phosfor, so I decided to slice the dome back further, just into the phosfor area, to do this I sanded the paperclip a bit thinner and sliced another bit off the dome, this time with a scalpel (because the led-board was soldered in already I could not use the razor blade):

And the result: exactly the same ugly hotspot, and 66klux at 1 meter. It hardly improved.

At this point I believe there is no way you can improve the throw of the LuxeonQ much with any of the available dedome tricks.

Conclusion of these tests:

In my opinion the Luxeon Q is interesting for flashlight use because it performs as well as the newest Cree's with a slightly different die size that could be just what you need. I have not checked, but because of the bit fuzzy light I expect a nice smooth beam, even using a smooth reflector. But new breakthroughs in throw it will not give because dedoming does not work.

And now for the unexpected giveaway for anyone who endured to read up to this point (a fun new custom at BLF :-) ) : I have 5 bare LuxeonQ 5700K leds leftover (no boards!), and will send them free (worldwide, no tracking, just in a plain letter) to 5 enthousiasts for this led who respond to this thread. If more than 5 people want one, I will do a thing to pick out the lucky receivers.

Thanks for reading!

Nice tests, thank you for this!

… I’m not up to reflowing, so I’ll have to pass, but it’s nice to know there’s choices other than cree/nichia.

Fantastic testing!

Thank you for sharing and all the time and money put into this.

Considering the easy availability of XP-E2 and XP-G2 combined with low price, I doubt this will become popular. Most XP-G2s are 70CRI. As far as I know most XP-E2s are too. When that is said, I would not mind experiencing this LED for myself, so I would like to be in on the giveaway. :)

Trim off the leftover phosphor & silicone that's over the substrate, leaving only the square patch over the die. Helps a lot to narrow the beam on MTG2s, should work the same here.

I have to say great pictures! :beer: :beer:
Can I / should I be in if I’m not sure what I will do with the Luxeon? :ghost: I love BLF’s new hidden giveaway custom. Keeps the riff raff at bay?

Thank you once again for a great emitter review and for the hidden giveaway opportunity.

Thanks for putting in the lab time and those are awesome photos :bigsmile:

Count me in for the giveaway! I have a bunch of P60 reflectors, I’d be curious to see how the LuxeonQ does in them :party:

I’d like to play with one of the Luxeons… I can toss it into the Sphere ’o Many Mysteries…

Interesting thread .

I'm not in for the hidden giveaway because I have lots of emitters

nice testing :slight_smile: I could use another interesting led so please count me in!

Cool test and I dig the 3D Photos! Those are fun to look at, how do you make them? I am out of the giveaway. No reflow skills here. If oldlumens doesn’t enter you can give my entry to him if you want.

Thank’s for the test, djozz. I would very much like to try one of theese. I have bare boards for xpg types.
(PM sent)

djozz. You have seriously got me worried about what you are doing. For this I thank you. Very nice work indeed. Happy xmas. :beer:

Thanks everyone for the nice comments. I really like this exploring unusual emitters, I never know what comes out because the way every different factory make the leds are pretty different and the factory specs are so conservative and what we do here is way over specs. Of course I want a new emitter to perform better than the ones we are used to, but apparently the Cree leds are pretty hard to beat. I have a Luxeon Z that I will try to test sometime (1x1mm, no dome), but first I must find a way to mount it, the footprint does not fit any of the for me available copper boards and it is soooooooo tiny.

@RaceR : you are right, the xp-g2 and xp-e2 come at 70CRI minimum, also at cool temperatures, while I thought that was 65CRI, I should have checked that. So the LuxeonQ is not special CRI-wise. I corrected the OP.

@comfy: removing the phosfor at the sides of the die will produce a narrower beam, but the illuminance of the die (and thus throw) I expect not to increase. If you remove the surrounding phosfor straight up to the die there will be blue light leaking from the sides (should be cool looking in a reflector light :-) )

Xp-g2 and xp-e2 are 70CRI typical,Luxeon Q is 70CRI minimum,slight difference.

Your lumen results for xp-e2 are close to datasheet,but for luxeon Q they are too low(it should be around 300lm@1A).Maybe wider emitting agle causes more light hits reflector-more loss?You should send one to texaspyro,Int. shpere will be more precise.

Just a note, luxeon Q is totaly different technology from Luxeon Z,Z ES,TX etc. Vf and thermal resistance are worse.Vf is higher because chip is much thicker-longer vias,similar to cree sic chip.So this isn't true TFFC anymore.Thermal resistance is worse because supstrate is ordinary alumina,Luxeon Z(ES),TX,M are mounted on Aluminum nitride supstrate(thermal conducitivity is sky high compared to alumina).

And one more thing:I noticed your "death" current for Xp-g2 and Xm-L2 is much higher than what other people measured.
I killed two xp-g2 somewhere between 5 and 5.5A,others also reported 5.2-5.3Amp.Which multimeter do you measure current with?

My lumens are not bare led lumens but 'out-of-reflector' lumens, with the LuxeonQ at 1A I measure 245 lumens, the datasheet states 305, so I measure 20% less which could be typical for the light loss from a reflector. The XP-E2 R3 at 1A is 266 lumens minimum according datasheet, 20% reflector loss leaves 212 lumens, but I measure 240 lumens which seems higher than the datasheet. But my set-up with the led on copper and the board on a block of aluminium cools very well, so at 1A I expect the junction temperature to be closer to room temperature than to 85degC at which temperature the datasheet numbers were obtained. But then, the LuxeonQ should be cooled just as well, but still performs close to what the datasheet tells at 85degC, perhaps the wider emitting angle does indeed spoil the output more than the XP-E2... Well, I don't know, this is what I see with my limited equipment, And ok, I will send one to texaspyro so he can do some proper measuring on it.

I have hoped for some time for someone more accurate than me and with better equipment to stand up and do this led-stuff for BLF, even though I enjoy this playing around with leds very much. Nevertheless, I still presume that my -admittedly makeshift- set-up only suffers from relatively small inaccuracies that influence the numbers with say less than 5% . But I do not rule out that something is more off, that is why I always describe how the measurements are done so anyone can judge for themselves how trustworthy the method is. Thusfar I have had no reason to doubt the numbers I get to a great extent. Something I perhaps should not trust so much is the current reading from the power supply. The XP-G2 and XM-L2 measurements were done with a Peak switching mode power supply that can deliver 40A max. It has an inbuilt current reading that I used (I do not own a DMM that I trust the current reading from any better). This power supply is certainly on the budget side, but I would not expect a current reading that is 2A off at 8A .

Thanks for the information on the other Luxeons, it makes me more motivated to try to mount the LuxeonZ somehow and do some measuring on it :-)

On graph I see lux Q is around 240lm@1A,but Xp-e2 260lm?But you said you measure 240lm for xp-e2 which does make more sense.

It is known that cree drops more at higher temp. so you are right,this is one more important factor.

Well,it's time to buy one decent multimeterI wouldn't trust cheap power supplies,especially switching types.

I tested Lux Z ES,and voltages are much lower than xp-g2,so I expect the same with lux Z vs. Xp-e2.But nonexisting thermal pad is problem.

I'm sorry, I keep being inaccurate ;-) , I did measure 260 lumens at 1A for the XP-E2 R3, like the graph said. But mind that these measurements are done with my other power supply (a Kert, which I think is an ok one) that is difficult to set to the desired current, so a 1A reading could have been anywhere between 0.97 and 1.03A, this also creates some (minor, but everything adds up) noise in the numbers.

In theory, there are reasons enough to think of to distrust every bit of my amateurish set-up, but on the other side, the fair amount of measurements on various leds I have done using this set-up do produce numbers, trends and graphs that do make enough sense to suspect that it is not working that bad overall. For instance, if the currents that I read out would be off from a certain value on, I would see that reflected in a bump in the output-graph and the voltage-graph. But I don't see any of that. So apart from everything not calibrated well, and all numbers surely being off by a few percent I think that I sort of trust what I am measuring.

But it may be time to buy a very decent multimeter (instead of the ok multimeter that I have, it is not the worst of the worst, i paid 60 euro's for it at the time) for current measurements, so that I do not have to rely on whatever the power supply in use tells me. It is asked before on BLF, but does anyone have a suggestion for one that is not a 200+ dollar Fluke or Agilent (I can afford up to maybe 60 dollars), and does measure up to 10A accurately without introducing too much resistance? (I can make some monster-size leads myself :-) )

You have 60€ multimeter and you don’t trust it?
I have UT-71C and it was best price/performance at the time (maybe it still is,don’t know what’s popular these days). For <60$
this one looks good:
Good precision+true RMS.

hi, i currently work on…let’s say “dedoming device” so i’m in for giveaway, it will be interesting to try and slice something other than xml… :stuck_out_tongue: