White LED lumen testing

[saabluster]

The main problem with doing it the way you pictured is that I can't put the LED on my test jig to get the beam profile.

Thats why i posted a picture. If you can give me a picture of your setup I'm sure I can make something that will fit.

The heat pipe won't make the heat sink "better".

I think i may not have been clear enough. My intent was that you still use your heatsink. I would not classify what I'm trying to do with the heatpipe as a "heatsink" per se. Consider it as a heatspreader which would still be in need of a good heatsink.

In fact, the heat pipe will add a small amount of thermal resistance into the equation as opposed to just mounting the LED on the heat sink as I always do.

I have the utmost respect for your knowledge and methods but I have to disagree very strongly on this point. As our old buddy Newbie has on his site(and here too I'm sure) many heatpipes have thermal resistances 2,000 times lower than copper and 4,000 times lower than aluminum. When you consider that the vast majority of heat coming from the LED die is concentrated in a very small spot and is not significantly spread by the ceramic substrate thermal resistance(and the need for a heat-spreader) becomes a very big deal. Heatpipes are in an entirely different league when it comes to spreading heat.

Love your "freezerator"! I can not imagine how many hours went into that. What do you use it for?

 

[jtr1962]

Here's three pictures:

I need something where the LED is mounted to a plate with screw holes spaced 0.75" apart. Note the 5/8" aluminum spacer between the LED and the heat sink in the second picture. Originally I simply had the LED right on the heat sink, and I had the heat sink mounted such that the LED sat where it does now (right at the pivot point of the test jig). I later added the spacer and moved the heat sink back 5/8" because some light past 90° was being blocked. Not much, but it could affect the results by a few percent. I'll be removing the aluminum spacer for these tests. Basically if you make a plate with holes drilled for 4-40 screws, space 0.75" apart, solder a heat pipe to the plate, and then bend the heat pipe in such a manner that the LED is sitting roughly 5/8" off the plate, and centered between the screw holes, I should be OK. I can always bend the heat pipe to make any small final adjustments once it's in my test jig. And solder the LED the heat pipe such that the power contacts are oriented vertically when everything is mounted to the tester. This will make everything consistent with the way this test jig works.

Here's a rough drawing (side view) of something I think will work:

Love your "freezerator"! I can not imagine how many hours went into that. What do you use it for?

Testing electronics devices mostly. I sometimes make things for the taxi industry which must function in temps as low as -40°F. I'd rather test them before the get into the field. Besides that it was just a fun project. Yes, literally months and perhaps $1500 went into this. Note that the housing for the power supply and controller is an old AT computer case I found! If I had to make another chamber I could do so much faster and better based on what I learned. Also, we have nice heat sinks available which I didn't at the time. I actually made the cold sink myself by soldering copper plates to a copper base. Now I would just use a couple of microprocessor heat sinks instead.

 

[saabluster]

Thanks for going through the trouble of posting those pictures. I have also rethought slightly where I would mount everything. I used your picture and tried to make a fairly accurate representation (layout+scale) of what it would look like. I based it on your cpu heatsink being 2" across. The thickness of the copper plate and heatpipe together is 1/4". My thought is ditch the aluminum spacer as you mentioned and replace it with some 1/4" thick aluminum that I have. This will have the LED die very close to the 5/8" you currently have. I think putting the LED over the area of the copper plate will allow even faster heat extraction than if the LED was mounted on the heatpipe far away from the copper plate.

 

[jtr1962]

That will definitely work so long as it's not too much trouble for you to make it. My only real concern here is will soldering the LED to the heat pipe cause any damage to it? I'm guessing you've already done this with good results. If so, then we're all set to go as soon as you have a chance to build what you envisioned.

Nice photoshop job on my picture! :twothumbs

 

[saabluster]

That will definitely work so long as it's not too much trouble for you to make it. My only real concern here is will soldering the LED to the heat pipe cause any damage to it? I'm guessing you've already done this with good results. If so, then we're all set to go as soon as you have a chance to build what you envisioned.

Nice photoshop job on my picture! :twothumbs

Great!:twothumbs I"ll get started on it. And yes I've figured out how to keep the LED from being damaged. I just need confirmation on the width I have to work with. I scaled it at 2" in width. Is this correct? Should I make my setup just under that maybe 1 3/4"- 1 7/8"?

 

[jtr1962]

Great!:twothumbs I"ll get started on it. And yes I've figured out how to keep the LED from being damaged. I just need confirmation on the width I have to work with. I scaled it at 2" in width. Is this correct? Should I make my setup just under that maybe 1 3/4"- 1 7/8"?

It's slightly more than 2" but making the plate about 1⅞" should work out fine. Just try to orient the LED so that the power contacts are vertical as in my drawing. It doesn't have to be perfect as long as it's close.

I look forward to doing these tests!

 

[saabluster]

Anyone want to guess the amperage it blows at and the peak lumens?

OK. Here is my official prediction.

[email protected] with the heatpipe
[email protected] chilled

 

[marschw]

I'm definetly outclassed on the engineering involved here, but I thought heat pipes were only useful when they draw heat upwards, since they rely on gravity to let the condensed fluid fall back down to the hot end of the pipe. Thus an arch-shaped heat pipe, as in the drawing, would only be as effective as an equivalent mass of copper, since the fluid would condense on the cold side and just sit there?

 

[jtr1962]

I'm definetly outclassed here on the engineering involved here, but I thought heat pipes were only useful when they draw heat upwards, since they rely on gravity to let the condensed fluid fall back down to the hot end of the pipe. Thus an arch-shaped heat pipe, as in the drawing, would only be as effective as an equivalent mass of copper, since the fluid would condense on the cold side and just sit there?

My understanding of the way heat pipes work is that the condensed fluid returns via a wick. Therefore, the pipe can be bent, and orientation is unimportant. See here and here

 

[marschw]

My understanding of the way heat pipes work is that the condensed fluid returns via a wick. Therefore, the pipe can be bent, and orientation is unimportant. See here and here

Ahh, that makes sense, didn't know about the wick.

 

[saabluster]

It's slightly more than 2" but making the plate about 1⅞" should work out fine. Just try to orient the LED so that the power contacts are vertical as in my drawing. It doesn't have to be perfect as long as it's close.

I look forward to doing these tests!

Well it's ready! All that I have left is to drill the holes. I need to know how big you want the holes drilled out to. Also, just to double check, is it exactly 3/4" center to center? Of note too I did not get the LED centered perfectly left/right. From the the center of the LED die to the right edge(as viewed from the front of your device) is 1". Since you said your heatsink is a little over 2" wide I figure this is still barely within spec. If not let me know.

 

[jtr1962]

Well it's ready! All that I have left is to drill the holes. I need to know how big you want the holes drilled out to. Also, just to double check, is it exactly 3/4" center to center? Of note too I did not get the LED centered perfectly left/right. From the the center of the LED die to the right edge(as viewed from the front of your device) is 1". Since you said your heatsink is a little over 2" wide I figure this is still barely within spec. If not let me know.

1/8" should be enough on the holes. It just needs to be large enough for a 4-40 screw and yes, it's exactly 0.75" center-to-center. It doesn't matter much if the LED is exactly on center. I can probably play around a bit to get everything lined up once it arrives.

Just letting you know also that I'll be busy with some work-related stuff for another week and change, so I won't have time to do the tests until I'm finished.

 

[simple]

:bow:

 

[saabluster]

Just letting you know also that I'll be busy with some work-related stuff for another week and change, so I won't have time to do the tests until I'm finished.

No problem.

 

[saabluster]

Well I finally got around to sending it to you jtr1962. I know you've been busy so I have not been in a hurry. However as I was driving down the road today, thinking about the fact that I sent it, it occurred to me that I forgot to drill the holes. Hope thats not a problem. If so, just let me know. Should be there by Friday so the mailman says. It was sent USPS.

 

[jtr1962]

Thanks for letting me know it's on the way. :thumbsup: Nothing in the mail today, it'll probably come tomorrow. Also, no problem drilling the holes myself. I have a drill press.

 

[saabluster]

Thanks for letting me know it's on the way. :thumbsup: Nothing in the mail today, it'll probably come tomorrow. Also, no problem drilling the holes myself. I have a drill press.

I'm sure you already know this but you will need to use a little oil and go fairly slowly to avoid the two layers(aluminum and copper) from delaminating. They are bonded with Artic Alumina epoxy. Can't wait!

 

[saabluster]

JTR have you received it yet? Looking forward to the results. Thought I would mention I bought a power supply a couple weeks ago and did a little test on one of my Q5s. I took it up to 3.2A for about 5 seconds. Did not appear to hurt it at all. The tint didn't even shift very much. The LED was soldered down to a solid piece of copper. I did not want to blow the LED just yet because I'm using it for some driver and run-time tests but as soon as that is over I WILL find that Q5's breaking point.:naughty:

 

[jtr1962]

JTR have you received it yet? Looking forward to the results. Thought I would mention I bought a power supply a couple weeks ago and did a little test on one of my Q5s. I took it up to 3.2A for about 5 seconds. Did not appear to hurt it at all. The tint didn't even shift very much. The LED was soldered down to a solid piece of copper. I did not want to blow the LED just yet because I'm using it for some driver and run-time tests but as soon as that is over I WILL find that Q5's breaking point.:naughty:

I received it a while ago. Unfortunately I had flooding in my basement on June 14 so it's taking a while for things in my "lab" to get back to normal. I'll probably finally be doing the tests sometime in the next few days.

Good to know a Q5 can handle 3.2A. My plan is to increase current until either the LED blows, or the brightness stops increasing (no point going further than that since ultimate maximum output is really what we're after here.).

 

[jtr1962]

I finally finished the testing. First order of business after drilling the holes and attaching the LED to my heat sink was to determine the nominal output the usual way. My results were 122.97 lumens at 350 mA, slightly above spec actually, but then again the margin of error for my testing is probably a few percent either way. Vf was a somewhat high 3.37V. Efficiency was 104.3 lm/W.

Next order of business was to begin increasing current until either the LED blew, or it didn't get any brighter. For this round of tests I used passive cooling. The heat pipe and copper plate spread the heat to the heat sink, which in turn dissipated it. The results were a little better than my usual setup. Apparently the heat pipe, or perhaps the combination of heat pipe and copper plate, did something. At 2 amps the output was 3.35 times the output at 350 mA, compared to 3.21 for the stock setup. The heatpipe provided a small (about 4%) advantage here. As I increased the current past 2 amps I was crossing my fingers for the LED to hold together. Finally at 2.5 amps the output stopped increasing. At 2.6 amps it was a little lower. Peak output was passive cooling was 436.7 lumens. Vf at 2.5 amps was 4.22V, and total power input was 10.55 watts.

Obviously in the case of passive cooling you'll eventually reach a point of diminishing returns. However, it's a good indication of what a reasonably-sized flashlight body can do if you pump enough current into the LED. The next order of business was the ultimate output which the LED was capable of. To that end I set a large fan to actively cool the heat sink. Throughout the tests the temperature of the copper plate the LED was mounted on remained very close to room temperature. Once the temperature rise of the heat sink was factored out of the equation, I was able to get increasing output past 2.5 amps. The peak output was reached at 2.7 amps. At 2.8 amps it was a little less. The results at 2.7 amps were 453.9 lumens, Vf was 4.32V, and a power input was 11.664 watts. This is the absolute maximum which this Cree R2 emitter can give under ideal cooling conditions.

Wanting to continue this "insanity", I bolted the copper plate and LED to the cold plate of one of my thermoelectric assemblies. By the time ice was starting to form on the plate, the LED was giving me 502 lumens at 2.5 amps. Even though below ambient cooling was required, the 500 lumen barrier for a single emitter was finally broken. Even better, the emitter is none the worse for all the abuse it's been through. It didn't have to die in the name of science. :devil:

Here are some charts of the final results:

Lumen output using fan-cooled heatsink

Comparison of passive versus fan-cooled output