Luxeon Rebel LEDs

[James Jackson]

We were given a few samples of these a week ago, and I finally have gotten around to 'lighting one up'.

Initial impressions...

1) They're difficult to solder...

2) They're brighter than the sun! Yikes!!! Even at a low 350mA (they'll work at 1 Amp - if you can keep 'em cool!) they're bright!

I cranked up the current to about 750mA - just to see if it really does get brighter... problem is... I'm now blind. (joking)

I'm letting it run - at 350mA - and periodically check on it - testing the board it's mounted on - for heat (Bergquist). So far - so good. It's running cool as a cucumber.

I'll do more testing - taking better data - later on, and will try to report the light intensity - taken with a lightmeter.

I'm liking these LEDs... all except for the soldering... still have to work out that detail.

Regards,

James Jackson

 

[frenzee]

Thanks James Jackson for sharing your experience. What did you mount the rebel on if I may ask?

I just got a batch of amber and red/orange rebels. The performance (compared to LuxI and LuxIII) is a yawn IMO. Red/oranges appear at least twice as bright to the naked eye as the ambers, just as in LuxI and LuxIII. Thermal management at 700mA and above is a major challenge though. I tried a 4xRebel prototype circuit soldered to a 3/4"x3/4"x.5mm wetted copper plaque which was then clamped to a 1/2lb. aluminum chunck (I also a tried a copper slab and a finned extruded aluminum heatsink). What I found out was that at 350mA even on a much smaller hear sink, everything was just peachy with no heat build up of any sort. But at 700mA all the rules change. The heat buildup around the rebels is very quick and very localized - too fast for much of it to reach the other end of the heatsink. What was happening was that a hotspot would build up very quickly (in less than 5 second) within a few mm of the rebels to a point where the copper plaque was getting too hot to the touch. I didn't run this setup for more than about 10 seconds because I didn't want to damage the emitters. It may be that my interface between the plaque and the heat sink was poor. I don't know. I have yet to try this setup with the plaque soldered to the heat sink.

Another remarkable behavior I noticed was that the rebels respond fairly linearly to voltage changes. With LuxIs and LuxIIIs, once you reach around 3.2 Vf the slightest increase in Vf results in a substantial increase in If, whereas with the rebels, the current pretty much stabilizes itself around 700-800mA. I don't know if this behavior is by design or just my eyes palying tricks on me. At any rate, I didn't push the rebels past about 3.75V.

 

[James Jackson]

Good morning. Thanks for your response.

I wrote some numbers down that documented the Voltage & Current levels when I was playing with this LED yesterday. I will post them later, when I get them out of my computer bag.

I have the Rebel mounted to a slightly modified Bergquist sample - one that they sent out over a year ago - to those that requested it. It is credit card sized, and about 1/8-inch thick. It is basically aluminum with prepreg and copper foil on one side.

I modified it by scraping through the prepreg to expose the aluminum, and then used thermal epoxy to bond the Rebel to the surface. I also soldered the tiny SMT pads to a pair of pads on the top side of the Bergquist sample.

As far as handling the heat... I feel that the Bergquist did a marvelous job. It may have been my epoxying the LED right to the aluminum that helped. I ran the LED at currents of 350mA, 700mA, 750mA and 1Amp. In all but the 1Amp, it remained cool to the touch. Even at 1Amp - it was only a tad warmish after 15 to 20 minutes of operation.

I am thinking that the best range of operation for the Rebel is around 350mA, as that is where one gets the 100 Lumens. If one runs it at 700mA, one only gets about 180 Lumens, and the Lumens per Watt drops from 100 to 90.

I have also discovered the way to work with these white light LEDs without going blind. I have a pair of BlueBlockers - sun shades that will cut the intense white light to an amberish color that is a tad more bearable. Not that I go around staring into the white light... it's just necessary to be looking at the general area of the LED, and it helps to not have your retinas fried.

As with any LED... heatsinking is key to getting good performance.

Regards,

James Jackson

 

[James Jackson]

Here are the numbers that I got...

At 350mA - 2.4V
... 700mA - 3.2V
... 750mA - 3.3V
... 1.0A - 4.0V

This seems to be in line with their published data.

Regards,

James Jackson
Oztronics

 

[SemiMan]

The thermal connection on the Rebel is very small. I would not rely on anything other than soldering. Thermal epoxy, even Arctic Silver has a much lower (about 7-10x) thermal conductance compared to solder.

Semiman

 

[evan9162]

There's nothing wrong with using thermal epoxy, if you're careful with it. Apply a very thin layer initially, then wiggle and press down to squeeze most of it out. With some care, you can actually end up with quite a bit of metal-metal contact between the thermal pad and heat sink. With careful application, you'd be hard pressed to measure the difference between thermal epoxy and a soldered application.

 

[James Jackson]

The thermal connection on the Rebel is very small. I would not rely on anything other than soldering. Thermal epoxy, even Arctic Silver has a much lower (about 7-10x) thermal conductance compared to solder.

Semiman

Thanks for the heads up on this. I used Dow Corning 1-4173 Thermally Conductive Adhesive.

It appears to be adequate for this application. We have used it in the past for other high-power LED projects.

I feel that the important thing is to get the heatsink of the LED as close to the base metal (aluminum) as possible - which is why I removed the prepreg from the Bergquist material.

I looked up the 'Arctic Silver' material... it appears that it does not fully cure until after after 200 hours and many thermal cycles. Perhaps this is the reason that you are not having much success using it.

Personally - I do not believe that soldering the thermal pad is the answer - as this defeats the possibility of getting the thermal pad directly onto the heatsink - which is usually aluminum (although copper can be used, too - if large enough - just more costly).

I will probably take thermal measurements - to satisfy myself that I am not overheating the LED. But from my initial observations - I wasn't. The thermal characteristics of what I have configured - seems to be more than adequate.

Regards,

James Jackson
Oztronics

 

[BentHeadTX]

2.4V @350mA?
The Rebels are rated at 100 lumens at 350mA so does that not increase the lumens per watt considerably? That low of a voltage is 0.84 watts or about 119 lumens per watt!

Another added bonus is with such low voltage, single cell AAA or AA light regulators will only have to double the voltage instead of boosting by a factor of 2.5 or 3 times.

I have a Fenix L2D CE RB100 on order which I plan on running a single AA body. It has been reported that it will regulate at 390mA on a single NiMH AA cell. If it stays at that current level, my runtime should increase to well over two hours.

Is this the LED that use single AAA keychain light guys have been waiting for?

 

[LukeA]

Is this the LED that use single AAA keychain light guys have been waiting for?

It was for me. (See link in my sig)

 

[spaech]

What I found out was that at 350mA even on a much smaller hear sink, everything was just peachy with no heat build up of any sort. But at 700mA all the rules change. The heat buildup around the rebels is very quick and very localized - too fast for much of it to reach the other end of the heatsink. What was happening was that a hotspot would build up very quickly (in less than 5 second) within a few mm of the rebels to a point where the copper plaque was getting too hot to the touch. I didn't run this setup for more than about 10 seconds because I didn't want to damage the emitters. It may be that my interface between the plaque and the heat sink was poor. I don't know. I have yet to try this setup with the plaque soldered to the heat sink.

I've noticed the same - very little heat at 350mA and a hell of a lot of heat at 700+mA.

I think the problem you're having with heat dissipation is a combination of things: You've soldered them very close together, which is going to challenge any heatsinking setup, considering the heat these things put out at 700mA. Spacing them out will help a lot. Also, I would say your guess about a poor thermal interface between the copper and the heatsink is correct. A good thermal interface should mean that the copper heats up at about the same rate as the heatsink, and it sounds like that's not happening.

For reference, I soldered my rebels to rebel stars, then mounted them directly to an aluminum heatsink of similar size to yours. See here:
http://www.candlepowerforums.com/vb/showthread.php?p=2120588&highlight=rebel+mounting#post2120588

This setup took the heat away from the LEDs very well, but with 8x rebels at 700mA the ~300g heatsink was just getting too hot after half an hour of running. Active cooling would solve the problem, but I'd rather not rely on it. I think the only solution at these currents (using passive heatsinking) is to get a bigger heatsink with more exposed surface area.

 

[SemiMan]

Arctic Silver has about twice the conductivity of just about any thermal epoxy on the market. Even not fully cured it is still better than likely most others out there.

It is not a matter of not having success it is a matter of performance.

Dow 1-4173 = 1.9 Watt/meter-k
Arctic Silver = 7.5 watt/meter-k
Solder = > 50 watt/meter-K (depending on composition)

Some of the lead free compounds are near 60 watt/meter-K

Hence, that layer of 1-4173 needs to be 30 times thinner than the solder to be as effective.

Semiman

Thanks for the heads up on this. I used Dow Corning 1-4173 Thermally Conductive Adhesive.

It appears to be adequate for this application. We have used it in the past for other high-power LED projects.

I feel that the important thing is to get the heatsink of the LED as close to the base metal (aluminum) as possible - which is why I removed the prepreg from the Bergquist material.

I looked up the 'Arctic Silver' material... it appears that it does not fully cure until after after 200 hours and many thermal cycles. Perhaps this is the reason that you are not having much success using it.

Personally - I do not believe that soldering the thermal pad is the answer - as this defeats the possibility of getting the thermal pad directly onto the heatsink - which is usually aluminum (although copper can be used, too - if large enough - just more costly).

I will probably take thermal measurements - to satisfy myself that I am not overheating the LED. But from my initial observations - I wasn't. The thermal characteristics of what I have configured - seems to be more than adequate.

Regards,

James Jackson
Oztronics

 

[jirik_cz]

Here are the numbers that I got...

At 350mA - 2.4V
... 700mA - 3.2V
... 750mA - 3.3V
... 1.0A - 4.0V

This seems to be in line with their published data.

Regards,

James Jackson
Oztronics

Are you sure that these numbers are correct? Because evan9162 measured much different Vf on rebel 0100.:thinking: http://www.candlepowerforums.com/vb/showthread.php?t=169547

 

[koala]

2.4V is like lower than G bin... more like E or F bin which is impossible.

 

[frenzee]

Thanks spaech. I will try soldering it directly to the heat sink. I think I will try the S-bond material which has a lower melting point than regular solder and bonds to aluminum as well as a range of other materials. What I really wanted to try is nanofoil interface material from Reactive NanoTechnologies, but alas they didn't reply to my request for samples. I guess they're not interested in the hobbyist marketplace.

 

[coolwaters]

how do u run them at 3.4v? or even 350mA?

my rebels are at 3v cuz im using a AC adapter which only has 3v 4.5v 6v 7.5v 9v 12v. so its either 3v or 3.75v...and 3.75 is too much for the blue rebels.

 

[LukeA]

how do u run them at 3.4v? or even 350mA?

my rebels are at 3v cuz im using a AC adapter which only has 3v 4.5v 6v 7.5v 9v 12v. so its either 3v or 3.75v...and 3.75 is too much for the blue rebels.

The voltage varies with the level of applied current. As long as the AC adapter supplies under 1A and is in the neighborhood of 3VDC, you should be fine.

 

[lyyyghtmaster]

I have used all of one Rebel 0100 so far, to mod an Elly. I ground off the bottom electrical contacts and Arctic Alumina'ed it to a Star-sized piece of aluminum. I made my power connection by scraping the coating off the traces on the top (dome) side of the Rebel and soldering to these. (very carefully, and with finely stranded wire!!) The thermal interface from this "star" to the light body was by regular thermal grease. It starts at about 450 mA (at the LED) on a fresh Eneloop, and drops from there. I have no problems at all with the Rebel overheating with this setup, and I have run it for two hours straight at room temp ambient, handheld. I took special care to get the epoxy layer as thin as possible. I estimate its output to be about 100 lumens, by comparison with my L2D CE, after the light has been on for 5 minutes. This compares favorably with the published specs, although I have not actually attempted to measure the output yet.

I don't think, though, that there is any particularly easy way to use these LEDs.
Good luck!

 

[spaech]

I don't think, though, that there is any particularly easy way to use these LEDs.
Good luck!

Trust me, the rebel-->rebel star-->heatsink mounting method, using a stove hotplate to solder the rebel to the star, is quick, easy and reliable once you've done a few of them. You end up with a nice robust star-mounted rebel with good thermal management at significantly lower cost compared to buying it pre-mounted. And it's not permanently attached to a heatsink. Only thing is that it relies on 3rd party aluminium stars, but IMO that's far less trouble than mounting them any other way.