help with heat problem

[Stage X Tuning]

I would love to see some quantitative data on the materials you're working with to make CF heatsinks. Can you post some data sheets or pics of a heatsink? What is the thermal resistance of the end product? There are a bunch of folks who would be very interested in the development of thermally conductive composites and if you have a solution that can come close to the conductivity of aluminum at a lighter weight or higher strength then let's hear about it. Data...Measurements...:thumbsup:

We are doing a bit of work with it and can not give out too much as we are still filling out patent forms, but here you can see a similar use here

Hunter_Technology.pdf

ill get some photos up soon once i get it approved.

as to why you dont feel the heat of the flame and only being 3" away is the fact that it transfers heat as fast as can be put into it and the transfer of heat back into the are as fast as it if put in. Which is not true for metals they "hold heat". heat a plate of metal and heat a plate of CF, CF will cool down +4x faster then any metal. Titanium is close to the 4x most other metals are far more then 4x

 

[snarfer]

This is a really excellent idea, to run water through the composite case. another thing you might experiment with is to cast the composite in such a way that it becomes a flat heatpipe. I don't know if it is possible to cast accurately enough to make 1/32" wide channels, which is about the optimum width for the vapor to move the heat both up and down, but even if they were a little bigger it would still work. the hard part is finding the correct pressure i think. and your composite would have to be totally sealed. if you could make it work as a flat heatpipe then you wouldn't really have to worry as much about the thermal conductivity of the composite itself. The case probably has enough surface area to dissipate the heat, it's just a matter of spreading the heat around.

 

[Axkiker]

This is a really excellent idea, to run water through the composite case. another thing you might experiment with is to cast the composite in such a way that it becomes a flat heatpipe. I don't know if it is possible to cast accurately enough to make 1/32" wide channels, which is about the optimum width for the vapor to move the heat both up and down, but even if they were a little bigger it would still work. the hard part is finding the correct pressure i think. and your composite would have to be totally sealed. if you could make it work as a flat heatpipe then you wouldn't really have to worry as much about the thermal conductivity of the composite itself. The case probably has enough surface area to dissipate the heat, it's just a matter of spreading the heat around.

Yeah spreading the heat out of one of my first problems. The way I overcame this was yo actually cast the housing around a large piece of aluminum channel. This then spread the heat through the bottem and up around the sides.

Ill have to look into the channeling idea.

 

[Axkiker]

Haha well that makes a big difference.. Have you ever thought about casting or forging aluminium? If you are doing mass-production then it is very cost effective.. Just send it out to china and get it done for like $1 a piece..:thumbsup:

Well I guess when I mean mass production I mean me making several on my own. I dont mean thousands probably more like hundreds. I just dont think its feasable for me to cast them out of aluminum for an order of 100.

Im doing this on my own and plan on making several but probably not what most consider mass production. sorry for confusion

 

[Steve K]

Yes I would love to put a thermocouple on the led. I am trying my hardest to find a way to accuratly check the junction temp. I just dont have the equipment.

you might consider using a solid state temp sensor like the LM35 from National Semiconductor. You can read the output with a conventional multimeter and convert the voltage easily to the actual temperature.

As a bystander, I'd like to see how this all works out. Aluminum seems like the obvious solution, but I'd like to learn about a plastic or polyurathane with good thermal conductivity.

I've used some polyurathane that was filled with aluminum oxide to provide a very modest thermal conductivity, but it wasn't up to the task of keeping a LED cool. Plus, it's murder getting parts out of the stuff! That's only an issue when doing failure analysis, but it does get your attention. :-) There are also issues when potting electronic parts inside such hard stuff..... it tends to crack parts off of boards, so you need to coat the parts in something soft.

details, details! Let us know how this works out.

regards,
Steve K.

 

[Axkiker]

you might consider using a solid state temp sensor like the LM35 from National Semiconductor. You can read the output with a conventional multimeter and convert the voltage easily to the actual temperature.

As a bystander, I'd like to see how this all works out. Aluminum seems like the obvious solution, but I'd like to learn about a plastic or polyurathane with good thermal conductivity.

I've used some polyurathane that was filled with aluminum oxide to provide a very modest thermal conductivity, but it wasn't up to the task of keeping a LED cool. Plus, it's murder getting parts out of the stuff! That's only an issue when doing failure analysis, but it does get your attention. :-) There are also issues when potting electronic parts inside such hard stuff..... it tends to crack parts off of boards, so you need to coat the parts in something soft.

details, details! Let us know how this works out.

regards,
Steve K.

Well ill tell you im getting mixed results from this stuff. Here are the experiments that I have performed and the results thus far.

At first I made a small disk which was roughly 2" in diamager .250 thick with .125" groves cut in the bottem to kinda act like small fins. Next I grabbed a k2 and used some thermal epoxy to bond the little sucker to the composite heat sink and set the sucker to 1A and let it go. Well the led didnt exactly burn up but you could tell the heat sink wasnt doing a good enough job. Without having a definate formula to check the junction temp I cant say how much too hot. However you could tell it was too hot.

next I knocked the k2 off and cut a small piece of aluminum roughly ..75" x .75" x .125" thick. I then used some thermal epoxy to glue the piece of aluminum to the composite material then glued another k2 to the piece of aluminum. This seemed to really help matters out. It more evenly dissipated the heat throught a majority of the composite heat sink.

This all gave me the idea to incorporate the aluminum channel into the actual case. With the k2's actually dissipating the heat into the aluminum which is then dissipating the heat into the composite material it seems to wick off enough heat to keep things in check.

Now I can tell that you could not run this setup out of water for an extended period of time. I dont think you could even run them if the case was solid aluminum for very long out of water. The case is just too small and they run too hot. However underwater which is what this setup is designed for works fine.

The only thing I really wanted to accomplish was to make sure if someone were to run these out of water for a short period of time (15-20min) that they would not cook the bulbs. Past the 20 min time frame its just tough luck.

I will also add that the substance im using has ALOT of aluminum powder. When I say alot I mean alot. Well over 50%.

 

[Oznog]

the material im using is simply a casting epoxy with alot of aluminum powder added to it. Im sure it doesnt dissipate heat as well as a true piece of aluminum but it does seem to do it fairly well. So well that the entire case heats up fairly uniformly and keeps all the leds running. However as you said until I find a way to check junction temp its all guessing

I don't see how that's gonna work. Are you talking about powdered aluminum metal or powdered aluminum trihydrade? The aluminum-filled epoxies I know aren't actually all that conductive, not on the order required to transport the wattage outside the case.

Look up the stuff on how "thermal resistance" works and adds up and how you'd arrive at a die temp from it. You've got the internal-sink-to-internal-air resistance, plus air-to-case, plus through-the-epoxy, plus epoxy-to-air. Also, even if the case were conductive, do I understand you're using 1A * 3.4v * 6 = 20.4W of LED? Even if the case was aluminum, it could still overheat in a small fixture due to lack of cooling fins. If that's correct the total thermal path from device to air must be under 5C/W to keep it under 125C even- and realistically that's a terrible temp to run at you want it much lower.

 

[Axkiker]

I don't see how that's gonna work. Are you talking about powdered aluminum metal or powdered aluminum trihydrade? The aluminum-filled epoxies I know aren't actually all that conductive, not on the order required to transport the wattage outside the case.

Look up the stuff on how "thermal resistance" works and adds up and how you'd arrive at a die temp from it. You've got the internal-sink-to-internal-air resistance, plus air-to-case, plus through-the-epoxy, plus epoxy-to-air. Also, even if the case were conductive, do I understand you're using 1A * 3.4v * 6 = 20.4W of LED? Even if the case was aluminum, it could still overheat in a small fixture due to lack of cooling fins. If that's correct the total thermal path from device to air must be under 5C/W to keep it under 125C even- and realistically that's a terrible temp to run at you want it much lower.

Powdered aluminum... Im not even sure what aluminum thrihydre is????

You are correct I am running 6 k2's in series at 1 A

Well I cant say if it should or should not work. All I can say is so far so good. LOL

submersed in water its good to go. If I pull it out it does get hot and im sure would eventually overheat. However I think it buys me enough time to realize that they are on and out of the water.....

 

[Axkiker]

I would like to add that there are som new potting resins out which are supposed to be really good at dissipating heat. One is Cermacast. I would experiment with that but its just too darn expensive.

 

[HarryN]

For 100 each, you are at a bit of a challenge, but surprisingly, casting does not have to be an incredible number to be cost effective. It does not have to be done in China either, just not in downtown LA. It is such a fast process that it probably makes the most sense for qty 1,000 - 5,000 though.

As an example, here is an outfit that I have some experience with:

http://www.laaluminum.com/index.html

They use permanent mold casting. If that can work for you, it is not so much more than the cost of the aluminum. You will note that they used to be in LA, then moved to a small town in ID for quality of life and cost structure. Very reasonable and nice to work with.

You might also consider to use steel sheet metal for your housing. Stampings are just crazy cheap once you are running.

Last but not least, machining an aluminum housing is not that bad, esp. in qty 100. A decent cnc can spit them out pretty fast, and if the machining is not that complex, it can even be done manually with the right offer.

Drawings are helpful of course.

 

[Axkiker]

For 100 each, you are at a bit of a challenge, but surprisingly, casting does not have to be an incredible number to be cost effective. It does not have to be done in China either, just not in downtown LA. It is such a fast process that it probably makes the most sense for qty 1,000 - 5,000 though.

As an example, here is an outfit that I have some experience with:

http://www.laaluminum.com/index.html

They use permanent mold casting. If that can work for you, it is not so much more than the cost of the aluminum. You will note that they used to be in LA, then moved to a small town in ID for quality of life and cost structure. Very reasonable and nice to work with.

You might also consider to use steel sheet metal for your housing. Stampings are just crazy cheap once you are running.

Last but not least, machining an aluminum housing is not that bad, esp. in qty 100. A decent cnc can spit them out pretty fast, and if the machining is not that complex, it can even be done manually with the right offer.

Drawings are helpful of course.

Hey thanks for the info. If I get all this lined out and see that there is a fairly large need I will definatly look into the aluminum casting. It would definatly be ideal.

As far as stamping I agree that would be a great way but I think the cost of the die would be so great that I would need to stamp way too many to be cost effective.

I also thought about CNC machining but am just not sure it would be cost effective. I can knock these housings out here at home for next to nothing.

So now I just have to get a good accurate way to judge the temp. I just grabbed on of those IR thermometer from my dad. Im gonna see what I can learn from this. Im not sure if I can pinpoint it close enough though.

 

[snarfer]

I think it would not be too hard to hook up a temp sense resistor in your dimming circuit. Probably it would just be a matter of the resistor and maybe a transistor. the effect would be to lower the current if things got too hot. That would probably solve your problem right there. You probably don't have a lot to worry about when it is underwater.

You would make a circuit in which the NTC pulls the base of a transistor high (or low) in order to connect (or disconnect) the potentiometer that is setting the dimming level. Sorry I can't remember all the details of that driver. But the driver you're using is all analog. It would work.

 

[Axkiker]

I think it would not be too hard to hook up a temp sense resistor in your dimming circuit. Probably it would just be a matter of the resistor and maybe a transistor. the effect would be to lower the current if things got too hot. That would probably solve your problem right there. You probably don't have a lot to worry about when it is underwater.

You would make a circuit in which the NTC pulls the base of a transistor high (or low) in order to connect (or disconnect) the potentiometer that is setting the dimming level. Sorry I can't remember all the details of that driver. But the driver you're using is all analog. It would work.

That is a great idea that I am gonna have to look into. I really was trying to figure out a way that these could be run for extended periods of time out of water without harm however I just dont know if that feasable.

For the past few hours I have been doing a few more tests to try and get a better handle on how hot they run in and out of water. I borrowed one of those IR thermometers but dont really feel they give a true measure of temp since the area they gauge is so large. Anyways here is what I found

While submerged in water I am able to run them contionusly and still able to hold my finger on the aluminum heat sink right next to the led. It feels warm but by no means hot or uncomfortable. I made sure that each k2 had an ample amount of paste and each was completly seated onto the aluminum. With all that said I feel comfortable that since the aluminum was by no means hot to the touch that the leds are running at a proper temp. The case itself gets warm and will eventually start to warm the water which I have the case submerged in but it never gets warm enough to cause pain or discomfort.

Now when I remove the case and leds from the water its a totally different story. After several minutes the aluminum heat sink within the case will begin to warm and be so hot that it will burn your finger. The case itself will eventually get so hot that it will be hard to hold onto. I checked the temp with the IR thermometer and was getting readings at the heat sink in excess of 180+ F . Now I know that this is still within range however being that im checking with a very vauge IR thermometer I figure actuall temps are higher.

So what I think that these will be perfectly fine underwater. They will run fine out of water but only for a short period of time. I may still try to find a more accurate way to gauge temp but for what im looking to do I think the tests have given a good enough idea what is happening.