Is my cree too hot?

[Axkiker]

I have been experimenting with several cree leds lately. Im trying to figure out if the material which I have attached them to will work as a heat sink.

I have one hooked to a power supply and running at a constant 700ma. It is glued to to my heat sink with arctic silver. When I grab the led the little ring will begin to burn me after a second or two.

I assume this is too hot however dont have anything to check the actual temp with.

any ideas from those who have messed with these leds before if this is cooking it or normal.

For what its worth its running just as it should be. Just hotter than what I like.

 

[yellow]

When one cant touch something from its heat, the temp is about 55-60 deg C.
Burns --> higher and thus not good for the led (possible danger of reducing the lifetime from 100.000 hours to a few hours or even less)

* Your sink is too small
* use thermal paste on the slug, place the led and glue it in with epoxy around emitter (= skip arctic)
* is Your question in any way related to the 1st sentence? :thinking:

regarding material, there is no univ. degree needed
1st: diamond --> much too expensive
2nd: silver --> too expensive
3rd: copper --> too heavy
4th: aluminium --> making a perfect package


PS: mount the led to a plate that is as thick as You have space or want to accept the weight, now mount that plate with a press fit into Your Aliminium flashlight housing.
All the heat now gets out to the large surface and mass of the housing which acts as You sink.
Cant go better than that.

 

[R33E8]

Re: is my cree to hot

How thick is the layer of paste between the emitter and heat sink? Artic silver only has a thermal conductivity of about 4W/m*K and when applied thickly acts more like an insulator... You should try getting as much metal to metal contact between the emitter and heat sink so the thermal grease only fills in the small imperfections in the surface.. Also, how hot is your heatsink?

 

[JWP_EE]

Re: is my cree to hot

If your power supply is a constant current supply you can use the voltage drop across the LED to determine the junction temperature. From the CREE data sheet the voltage across the LED will drop 4mV for each degree C rise of the junction temperature when using a constant current.

You need to get a measurement as soon as the supply is turned on. There will be some heating effects that you will miss but if the reading it taken in the first second or so it will be close. That is the voltage at ambient. When the voltage across the LED stops dropping the temperature has stabilized.

Say the ambient temperature is 30C and the voltage difference is 200mV, then the junction temperature rise is about 50C. This give a junction temperature of 80C (ambient + rise) and at this temperature your LED will have a long and happy life.

 

[Axkiker]

Re: is my cree to hot

How thick is the layer of paste between the emitter and heat sink? Artic silver only has a thermal conductivity of about 4W/m*K and when applied thickly acts more like an insulator... You should try getting as much metal to metal contact between the emitter and heat sink so the thermal grease only fills in the small imperfections in the surface.. Also, how hot is your heatsink?

Well I believe it should be around .008" thick. I have the cree led soldered to a board which I etched that was .007 thick. I cut a hole in the board where the thermal pad of the led was so I could use arctic silver to adhear it to the heat sink. Im new to this stuff so maybe the artic silver just has too much resistance.

Someone mentioned the ides of using typcal thermal paste on the pad and bonding the led by the sides with epoxy. That sounds like a pretty good idea im gonna have to try that.

This stuff im using is not aluminum. its a composite material which I can vary the amount of aluminum in it. Im hoping I can get a mixture which provides enough thermal conductivity to do what I need it to do

 

[Axkiker]

Re: is my cree to hot

If your power supply is a constant current supply you can use the voltage drop across the LED to determine the junction temperature. From the CREE data sheet the voltage across the LED will drop 4mV for each degree C rise of the junction temperature when using a constant current.

You need to get a measurement as soon as the supply is turned on. There will be some heating effects that you will miss but if the reading it taken in the first second or so it will be close. That is the voltage at ambient. When the voltage across the LED stops dropping the temperature has stabilized.

Say the ambient temperature is 30C and the voltage difference is 200mV, then the junction temperature rise is about 50C. This give a junction temperature of 80C (ambient + rise) and at this temperature your LED will have a long and happy life.

Okay so lets see if I did all this right

I attached my meter to the led while at room temp and set the power supply to 700ma. I then allowed the led to cool back off to room temp. Next I turned it back on while cool and got a 3.47V reading. I then allowed the led to run a good while and checked the voltage again and got 3.19V

so with that said its a voltage drop of 280mv 280 / 4 = 70C

the room temp is at 21.111C so 21.1 + 70C = 91.1C for a junction temp

 

[blasterman]

Re: is my cree to hot

700mA requires fairly significant surface area for the emitter to safely radiate heat with passive means. The type of compound you use to attach the emitter to the board or heatsink makes no difference if there isn't sufficient area for the resulting heat to go.

Also, the actual composition of the heat-sink itself makes little difference if it doesn't have sufficient area to dissipate heat. It doesn't matter if it's stamped aluminum, copper, low grade steel, or unobtainium. The difference in thermal coefficients between various metals is insignificant compared to how much surface area it has to radiate heat.

I won't attempt 700mA without at least eight square inches of total radiating area.

 

[Axkiker]

Re: is my cree to hot

700mA requires fairly significant surface area for the emitter to safely radiate heat with passive means. The type of compound you use to attach the emitter to the board or heatsink makes no difference if there isn't sufficient area for the resulting heat to go.

Also, the actual composition of the heat-sink itself makes little difference if it doesn't have sufficient area to dissipate heat. It doesn't matter if it's stamped aluminum, copper, low grade steel, or unobtainium. The difference in thermal coefficients between various metals is insignificant compared to how much surface area it has to radiate heat.

I won't attempt 700mA without at least eight square inches of total radiating area.

Well thats the issue im not using an type of metal for the heat sink. This is all an experiment that im doing with a type of compound which conducts heat based on the mixture. So im trying to get a fairly acurate way to test or read junction temp so I know how well everything is working.


If that formula for figuring out temp based on voltage drop is pretty acurate them im not doing all that bad. I dont have a huge heat sink and am just beginning to mess with the mixture. So this could be pretty promising.

 

[JWP_EE]

Re: is my cree to hot

It sounds like you did it correctly. 91C is not too hot for the CREE. You should get about 50,000 hours out of it at that temperature. Take a look at this chart I posted elsewhere.

http://www.candlepowerforums.com/vb/showpost.php?p=2839215&postcount=6

 

[Axkiker]

Re: is my cree to hot

It sounds like you did it correctly. 91C is not too hot for the CREE. You should get about 50,000 hours out of it at that temperature. Take a look at this chart I posted elsewhere.

http://www.candlepowerforums.com/vb/showpost.php?p=2839215&postcount=6

Another experiment im gonna do is to drill a very small hole in the heat sink material right where the led will be mounted. Then I will use the sink paste and apply that to the bottem of the led thermal pad then glue the led to the heat sink with the hole directly under the thermal pad of the led. My idea is to then take a probe and check the temp of the led thermal pad from the back side of the heat sink. This way I should be abel to make direct contact with the led while its running and still connected to the heat sink. I know that it wont be a true reading since the hole in the heat sink will cause the led to run a little higher due to less surface area to dissipate heat from. However if I can get it within acceptable range with this test I should be certain im safe.

 

[JWP_EE]

Re: is my cree to hot

If you can get an accurate temperature reading of the LED thermal pad then you can calculate the junction temperature from that. The thermal resistance from junction to thermal pad on the CREE (XR-E) is 8 degrees C/W. With the LED voltage and current known the power is just the voltage times the current. From your previous measurements the power is about 2.25W. So the junction will be about 18C above the back of the thermal pad.

Don't you just love this stuff? :thumbsup:

 

[Axkiker]

Re: is my cree to hot

If you can get an accurate temperature reading of the LED thermal pad then you can calculate the junction temperature from that. The thermal resistance from junction to thermal pad on the CREE (XR-E) is 8 degrees C/W. With the LED voltage and current known the power is just the voltage times the current. From your previous measurements the power is about 2.25W. So the junction will be about 18C above the back of the thermal pad.

Don't you just love this stuff? :thumbsup:

yeah never imagined these new leds were this complicated

 

[blasterman]

You should get about 50,000 hours out of it at that temperature.

Provided the tested run temp is going to be constant along with the conditions it's tested in. Heat needs to measured with all covers on, panels closed, etc. Otherwise, when everything is put back together temperatures are going to be much hotter.