My main concern these days is trying not to set the cat on fire.
Speaking about the cat:
:bigsmile:
That’s a cool touch lamp!!!
How come you would not just use thermal adhesive such as Arctic Silver Alumina? I am sure the conductivity would be the same or better, no?
- Err, why are you “sure” of that? In any case, it’s not true. Common, decent quality thermal greases are better than decent thermal adhesives such as the one you mentioned.
- Thermal adhesive is also less sturdy than soldering, for what that’s worth. Surface prep also matters much less when soldering rather than using an adhesive.
- Soldering or grease is also much cheaper than using adhesive. Generally using an (epoxy) adhesive on single build will waste a lot of expensive adhesive due to the mixing process.
@PPDB22 - sounds legit to me…
That is why I said “,no?” In any case I just didn’t know that solder would be that much better of a thermal interface, as I have never had to use it, nor couldn’t as most everything I have worked on could not be heated to such high temperatures. I looked it up…
“Solder thermal interface materials with their 10 – 50x higher thermal conductivities over thermal greases and metal filled epoxies have much higher heat dissipation efficiencies and can are best for high power devices with power densities in excess of 1000 watts. ”
So now that I know… I am wondering about my LED light bar project. I was going to use 1/4 inch copper bar between the enclosure and the LEDs 20mm Copper stars. Is it advisable to solder a large copper object to LEDs? What I am wondering about is how long the LED will be at a sustained temp, it will take a while for a large copper bar to cool down, however compressed air could be used I suppose to aid in flash cooling?
Between copper bar and case obviously can’t be soldered so I will be using graphite thermal pad.
You won’t be able to even get close to the temp profile shown in Cree’s datasheets. Neither the heat-up nor the cool-down will look anything like what Cree recommends.
Also, the advantages of soldering probably won’t come into play in the scenario you described. Like Ouchyfoot said, soldering MCPCBs to pills is setting yourself up for a world of hurt in so many different ways. Trying to do it for more than 3 or 4 LEDs at once will probably increase your chances for failure by a lot. I don’t see any reason to use copper bar at all on a normal lightbar project. Aluminum is the way to go. Typically in a lightbar application you just use lots of LEDs at reasonable drive currents, so there is no need for extreme thermal junction performance, only adequate thermal junction performance. Same for materials - lots of LEDs at a moderate drive current for long periods of time just doesn’t seem like a good candidate for copper to me.
I don’t think that I’m familiar with graphite thermal pads. What’s the deal there?
I too have been thinking through a light bar build. I under your desire to use copper. My calculations have shown that just mounting my LEDS to an aluminum enclosure/heatsink will not be adequate for application. So if I had access to 1/4” copper, I would not hesitate to use it in my build. That said, this is all opinion! I have NO experience here, so I suppse I should keep my mouth shut? But that’s hard to do 
Maybe you could share your calculations.
Ahhh man… I could probably do the calculations again, but it took me a bit of learning to do it. For my light bar using 12 xml2 LEDS at 3amps the total thermal resistence junction to ambient needed to be under .8C/W to maintain a temperature equilibrium of 100C. Trying to get the thermal resistance below 1C/W proved to be much more dificult than I had expected and required a larger form factor than I wanted. I Have the LEDS and the enclosure, but that idea is tabled as for now I’ve been building flashlights.
if you are probing to question my calculations, you are justified in doing so. I followed along to my research the best I could, but I realy don’t know if my procedure was flawed or not? I have not attempted to compare results with other projects. I plan on doing so though.
Yeah, sticking an LED on a slab of copper before the heatsink is a requirement for a light bar, it is a continuous high heat source, and must stay within operating temps for hours on end, much different than a flash light. I will be getting an enclosure machined of aluminum, then attaching a heavy heatsink to the back of the case, much cheaper to do this way. Then inside, use the copper bar between LEDs and case. This is the best way to effectively remove heat away fast and the least expensive.
FYI, almost ALL light bars out there, especially the cheap China ebay/amazon light bars, are run at ~60% power.
I think you are looking at the right things and misinterpreting them. What specific thing do you think that this “slab of copper” is going to achieve?
Operating at 60% drive current is what I call “moderate”; I referenced this above. Lots of other people have already tried (and failed) to build light bars using the same design considerations used for our flashlights (eg 100%+ drive currents). It doesn’t work. It’s all about moderate drive currents I’d say. … and the lower your drive current, the more LEDs you need to get the same total output.
You really don’t know what copper will do immediately off the LEDs compared to aluminum? So because alot of other people have failed at something, that means everyone will fail? Perhaps they don’t understand thermal design? All I can say is keep on eye on my build thread and you will see that my light bar will be driven at 100%, reliably.
Game on! Show us what you got ;P
I understand what you are saying and that is why my project has been tabled for now. I was unable to get a thermal resistance lowing enough without increasing the size of the desired package.
I also agree that the copper bar is not a clear winner. (Though i would still use it if I had it). At the low thermal resistances we need to work within, every junction affects the total thermal resistance more than anything. So we have led to mcpcb to copper to enclosure to heatsink(if separate from enclosure) and then to air. With this many junctions it is nearly impossible to get thermal resistance numbers lower than 5-10C/W within a small form factor. And thus lower amps are required as was stated above. And this would also suggest adding an extra copper bar to the heat path would be counter productive.
The design I have found most promising is to use copper pc heatsinks, attatch the LEDS directly to the heatsink. The copper would be enclosed into the back section of a two cavity enclosure. The divider separating the two cavities would be drilled so that the LEDs could attatch directly to the copper. The front cavity with the LEDs is obviously open, but the rear cavity is enclosed with a fan blowing air across the copper fins within that cavity. Idk if I have explained my idea very well or not, but I have been able to calculate very low thermal resistances this way.
I think I know what you mean about every junction/connection increases thermal resistance, so I will experiment hands on what will work better and definitely report back the results. If Baja Designs can make a 4.5” x 4.5” LED light with 4 XHP-50s driven at 100%, I am a bit of an optimist, but I think I can make it work with 4 xhp-50s and 2 xhp-70s in a 3.5” x 10.5 to 12” size. I ALWAYS over build anything I make, and I always make sure my project can handle the worst case scenario.
HERE is the heatsink I will be using, 75mm width by either 250mm or 300mm, depending on what will be needed (~10 inch or 12 inch bar).
I have a XHP-70 J4-3B and XHP-50 J2-3D coming my way from LEDDNA to test with some optics. I will post the results.
Anyways, we kinda hijacked this thread so I will be moving anything on the topic to my own. Sorry.
Is that 12 x XM-L2 @ 3A each?
What are you referring to?
If you are referring to my build, yes that is what I had planned.