Led heat and temperature questions

[videoman]

Would a led that is running 25% less than it's rated max amperage but at the same time gets very hot ( heat wise ) due to small heat sink, display a cooler tint over time than rated ? I am running an XM-L at 2300ma but the heat sink temperature just next to the 20mm star is 114 deg. C.

 

[SemiMan]

Would a led that is running 25% less than it's rated max amperage but at the same time gets very hot ( heat wise ) due to small heat sink, display a cooler tint over time than rated ? I am running an XM-L at 2300ma but the heat sink temperature just next to the 20mm star is 114 deg. C.

Color binning is at the test current/temp, not at the max rated current/temp. Anything beyond that ... which you are on both current and temp, will result in a cooler tint.

Semiman

 

[AnAppleSnail]

Unless you're measuring with a thermocouple that is illuminated by the LED output, that is an alarmingly high temperature. If this is a correct measurement, then your LED is in great danger of thermal overload.

The XM-L package has a thermal resistance of 2.5 C/W. I believe the star thermal resistance is similar. Your LED runs at about (2.25A * 3.2V) = 7W, so it's (7*(2.5+2.5)C) hotter than your test point, or 36C hotter. This is too close to the absolute maximum sustained temperature of 150C.

Check your thermal path!

 

[videoman]

Thanks, that cleared up a few questions.
If I lowered that 114 degree C to 100, by a larger heat sink, would that be adequate for a continuous 2300ma to the led.? Again measuring as close as possible to the star. What I am trying to determine is the smallest heatsink size without compromising the life of the led.

 

[LuxTiop]

I'd have to agree with AnAppleSnail I would look to either active cooling ( a fan) or getting a larger heatsink. A note on determining your junction temperature is to ensure you use your watts in heat in your calculations, not total watts. For instance in your case instead of using 7.2W (3.2vf x 2.25A) you must take into accord the light output. Assuming 20% efficiency of producing light the system is producing 5.76W of heat (7.2W x 0.8). The thermal resistance between your heatsink and star is purely based upon what was used, Artic Silver, if applied correctly, normally has a thermal resistance close to 1 c/w. Thus if your heat sink is 115c the star should be reading around 120.76 ((1c/w x 5.76w) + 115), and your junction should be around 135.16 ((2.5c/w x 5.67w) + 120.76). With all this being said how close is the driver to the LED, and what sort of ambient temperatures are you dealing with? It has to be remembered that drivers are not 100% efficient (mid 90's at best in a boost only topology, and 70s can be seen in a buck/boost) so this might be adding to the thermal problems you see.


Regards,

LuxTiop

 

[videoman]

Thanks. Active cooling is out of the question as the final design must be absolutely quiet. The battery or power supply is not in the equation as it will be more than adequate to supply the needed power even with the driver at 70 % efficiency and the driver will be mounted separately so no heat from it will affect the led's performance. What I calculated that for the lumen output that I will be needing, the sweet spot is in the area of driving the led at 2300ma ( preferably when the latest XM-L2's are out). My main concern is the minimum size ( or surface area) of the heat sink that will not allow the led to fizz out at that 2300ma due to heat.. I will be using AA 2 part adhesive for potting. Thinking outside the box, a drop of water boils at 100 deg C. If I lay a drop on the back side of the heatsink directly behind the led after the led shines for 5 minutes at 2300ma, and does not boil out to evaporation seconds after I drop it on, can I assume the led can tolerate that for long term running since it will be less than 100C ? I am imagining that if the star's temperature is roughly 5 deg C more than the sink, then the junction temperature may well be less than 125 C area. Can the led handle 125C jT for long term ?

 

[AnAppleSnail]

Thinking outside the box, a drop of water boils at 100 deg C. If I lay a drop on the back side of the heatsink directly behind the led after the led shines for 5 minutes at 2300ma, and does not boil out to evaporation seconds after I drop it on, can I assume the led can tolerate that for long term running since it will be less than 100C ? I am imagining that if the star's temperature is roughly 5 deg C more than the sink, then the junction temperature may well be less than 125 C area. Can the led handle 125C jT for long term ?

They make very quiet fans these days. In general, the limit for a heat sink is its ability to put heat into the environment, not its ability to get heat from the LED. In numbers, the LED thermal resistance is about 2.5C/W. The star:heatsink junction may be something like 1-3 C/W. But the heatsink:air junction is more like 10-30 C/W with stationary air.

"Absolutely quiet" isn't a design requirement. 1 decibel? 10? They make fans you can only hear in soundproofed rooms.

Per temperature: The lumen output of LEDs seems to degrade mainly based on temperature of the air around the LED dome. The LED will survive at 125C. I don't like the idea of having exposed heatsinks of such high temperature around people, though. Above 60C it is immediately painful to touch most objects, and 100C leaves pretty instant blister burns. The LED will survive for a while. Will your cameraman?

Heck, if you have infinite size and capacity on your remote power supply box, use a water pump. Those can do powerful cooling for reasonable runtimes.

 

[uk_caver]

What about static water (or ice) cooling - having your heatsink being a metal plate on the side of a suitable container?

Unless I've screwed up my calculations,
1kg of ice could absorb 30W for ~3 hours just by melting (maybe worth insulating container).
1l of water could absorb 30W for ~2 hours by heating up by 50 degrees C, ignoring losses to environment.

That'd equate to an electrical input to LEDs of ~45W or so.

 

[videoman]

It will be for the use of a camera ( video) mounted light. It will be very close ( like 1 inch away, hence no fans) to a sensitive Sennheiser microphone that is also mounted on the camera next to the light. The color temperature of the led must be around but close as possible to 3200K as that is the color preset of indoor on the camera selector setting. A variation from 2900 to 3400K from 3200K looks acceptable, after all a white balance calibration can also be done on the camera during the shoot. I am trying to keep the metal enclosure or lamp housing as small as possible to sustain the led"s 2300ma current at long sessions of continuous shooting that sometimes last well over an hour between takes like long speeches or interviews. The led driver along with the dimmer and battery is housed in another case that acts as a counterbalance positioned behind the camera along the shoulder support. The led will eventually be the 90 CRI version warm white XM-L2 as the test unit. I am looking for the smallest aluminum housing possible that can be adequate in keeping the led within it's tolerable temperature range at 2300ma. I have found no chart relating to surface area needed by various metals or heat sink fin design parameters in relation to leds mounted on 20mm stars and ambient temperature variations. So I am just using an aluminum bar and cutting it in increments to determine the smallest I will need to get the job done. So far 6.5 inches seems about right from a bar of 2 inches high and .25 inches thick.

 

[AnAppleSnail]

It's tough to really make a good chart because there are so many variables. Heatsink temperature? External conditions and temperature? Airflow? and so on. It seems that you're limited by the heatsink's ability to put heat into the air, so let's work on ways to improve that.

How much variation is there from front to back of this bar? If not much, you might try a longer, thinner bar (More surface area). You can also get extruded, black-anodized (More professional look, possibly better for heat emission) heatsink stock that you cut to length. You'd cut to length and carefully polish the end the LED mounts to. Finally, adding very small fins very close to the LED mounting point should make a difference.

And in the end, well, the LED will be okay if you can keep it below 150C. It may lose a few percent output every hundred hours if it's close to the limit, but that's manageable. Pushing limits costs.

You might be best served with a few dozen extra grams of aluminum on the front and back of the camera, though.

 

[videoman]

Thanks. How does one go ahead and measure( unless there is a chart somewhere ) junction temperature at various currents as the led is lit ? Cree specs it as 150C max but the only graph I see from Cree is the percentage of flux vs junction temperature and that is at 700ma level. From that graph on page 4

http://www.cree.com/~/media/Files/C... Modules/XLamp/Data and Binning/XLampXML2.pdf

if I were to target 125C jT , then what would the current through the led have to be in order to remain that cool at 125C ?

 

[argleargle]

If your led is on the verge of frying in open air, it will run hotter in an enclosure (but maybe not.)

http://en.wikipedia.org/wiki/Thermoelectric_cooling
An active cooling suggestion that is silent is a "peltier cooler." They're sometimes used in small computers. It is basically a pad that takes current and devlops a "cool" and "hot" side. The hot side receives the heatsink treatment (your pill or whatnot.) They come in a wide variety of capacities. Good ones are properly labeled (hot and cold.)

The danger of a peltier is that if it is unpowered, it acts as a heat insulator. Also, it's possible to basically "put it in backwards and fry your build." They're also not very efficient. You said "silent," though. They are, at that.

Another suggestion is that if you physically roughen the exterior surface of the light, it will help it dissipate heat better than a smooth surface. Bead blasting and even knurling helps.

To add to the question of "how much current," place an Ammeter in series with the circuit and power it on. Don't guess. Know. :) I just said that because 2300 mah is a calculated value and those can vary plus or minus 5%-25% in the real world due to one thing or another.

A suggestion for measuring temperature is a remote thermometer. They have a nifty laser sight as well.

 

[AnAppleSnail]

That depends on the heatsink. Measure near the LED and estimate the thermal resistance between that point and the LED. So if you measured on the MCPCB metal, you'd see something like 3-5 C/W difference between that point and the LED die. So at 7W drive power, your LED will be 21 to 35C hotter than that measuring point.

 

[videoman]

Great stuff. Lots of good info, thanks to all.

 

[argleargle]

Best of luck, vm. I've subscribed this thread, and I'm sure others have as well. Feel free to ask more stuff and tell us how you've managed to win the war on heat.

 

[Esko]

What kind of optics are you going to use? None? Why don't you use a triple (three leds on a star)? Underdriving leds makes them more efficient which means considerably less heat.

For example (I use these numbers for the calculations):

1*XM-L driven at 2300mA: power consumption 2,3A*3,2V=7,32W; output 753 lumens
3*XM-L driven at 600mA each: power consumption 1,8A*2,9V=5,22W; output 798 lumens

If we assume that 750-800 lumens of light equals to something like 2W of pure lighting power, it means that the first setup generates 5,3 watts of heat and the latter one 3,2 watts of heat.

 

[AnAppleSnail]

Triple shadows on video are icky. Isn't 800 lumen pretty dim for anything but close work? I needed 1500 OTF in my last video project of decent quality.

 

[videoman]

I will eventually get to use the MT-G2 for a single shadow source and apply the same method of heat calculation. The heat sink will have to be larger as I will be running that at 3A and coupled with a Ledil Minnie wide http://www.ledil.com/node/2/p/3340. only problem is them newest MT-G2's can't be found anywhere in 3000K 6 volt versions. Should produce a nice wide beam to 25 ft.

 

[Esko]

Triple shadows on video are icky. Isn't 800 lumen pretty dim for anything but close work? I needed 1500 OTF in my last video project of decent quality.

The calculations still apply and single XM-L creates lots of waste heat. If 1 cm difference with 3 mm leds is too much, one can always make a custom board with leds side by side. It just can't be bought from led shops for a couple of $$ (afaik). Anyway, singular light source creates very sharp shadows which is not what people usually want. Extremely wide (180 degrees) beam angle often helps a bit, though.

 

[argleargle]

The War on Heat:

I just read that silent thermoelectric Peltier coolers have been basically defeated by a newer upcoming silent cooling tech. It's a laser. :) At the moment, the heat-sink would have to be cadmium-sulphide and they still don't have it working directly on basic semiconductors (yet.)

Too cool!

Published yesterday: http://www.nature.com/nature/journal/v493/n7433/full/nature11721.html