I’m designing another driver and will not need an exposed ground ring to make ground connection with the host… So I was thinking, should I have one anyway to get heat away from the driver board? But then, the host is getting pretty hot pretty fast from the LEDs. Wouldn’t a driver with an exposed ground ring be more likely to suck in heat from the hot host?
I'm curious about this effect as well, especially regarding potting drivers. At what point will the emitter actually pump more heat into the driver and actually make things worse?
When in doubt, add copper.
Not having the path might delay the heat getting to it, but over time it would probably even out/catch up.
It depends entirely on how much heat the driver is generating. What type of driver is it? It should be fairly easy to estimate the power dissipation with basic info about the design
About 18 amps driving 3 MT-G2s. I already have a driver that does this but am making changes to the design to try something new. The host get’s hot fast because of the LEDs. I think the LEDs produce much more heat than the driver, so in this case I think that an exposed ground ring will be counter productive unless I only want to run the light at max in 10 second bursts.
18A is not heat, it’s a measure of current. I know that you know that, but you may not realize how little information that conveys. A more useful thing would be to say “18A w/ x.xV amount of voltage drop over the driver”. Then we could get the wattage being dumped by the driver. 18A in DD is effectively no heat from the driver, while 18A with very strong cells, a low Vf emitter, and a linear driver would generate a good amount of heat.
It’s not just about heat, it’s about temperature. It is possible for the 7135’s (what you used on your previous version) to generate enough heat to be far above “ambient” temp w/in the light. At that point it is good to sink that heat to the flashlight body.
I don’t think that there is a sustainable (regulated) scenario where you don’t need to heatsink 7135’s to something. In your “hot potato” test you held onto the light body for >1minute. I think if you try and hold on to a bare 7135 for a whole minute the test will take less time!
Thanks for your thoughts wight. A lot of things I didn’t think about (lack of knowledge). I am considering heatsinking the 7135s (if I use them), but maybe not to the host. I guess I should have a go at touching the bare 7135s to get an idea of how hot they get.
If you are worried about heat of the driver, pot it. Adds thermal mass and larger surface area to push the heat to the body
My concern is that because the body is so hot, potting will pull in heat from the body and heat the driver rather than push heat from the driver into the body. I guess it’s a question of which is hotter, the driver or the host…
No matter how low the power they dissipate, the heat from the 7135's must go somewhere or else their temperature is just building up infinitely. And that somewhere will have to be the flashlight body. And no matter how good or bad the heat path is from the chips to the body, the heat will only go there if the chips are hotter than the body, and they will be because the heat can not go anywhere else. So insulating the chips from the body may buy you some time (my guess is a few seconds), but with the heat nowhere to go they will then get bl**dy hot and way hotter than the shell.
I see. So they will got hotter. Thanks for the advice.
Yep, can only agree with what everyone has said. Heatsink path from a linear driver to the outside of the light should be as good as possible.
Also consider that you won’t be running the light in a continuous high mode much in real world use, so a poor heatsink path for the drivers means that they will take much longer to cool back down compared with the rest of the light. The more heatsink mass you add to a thermally isolated driver the longer it will keep the chips at a high temperature even after they are no longer working hard. The heat will have nowhere to go!
Finally, the body of the light likely shouldn’t reach above 70-80 degrees, while the 7135 chips if poorly heatsunk and needing to dissipate a fair amount of energy will easily go above 100 in short order. So even when the light is far far too hot to hold onto, the 7135s will still be “cooled” by the body of the light. It’s all about the temperature difference at that point.
If you’re in any doubt about how much heat those 7135s will put out driving 3x mt-g2 @18A off a set of freshly charged cells. Just run the driver outside of the light and see how long it takes before you can’t touch the ground ring anymore.
I call that the “hot chip test” 
The AMC7135 Eagle library I was using isn’t optimal for heat handling. As I don’t run traces under them I made the ground connection much larger, so the exposed underside of the chips have more contact with the copper on the PCB. I’ve seen others here that have done this also.
IMO increasing the size of the contact won’t make a big difference. It’s all about the size of the pour and whether the pour connects to something else (flashlight body, etc).
If they generate heat it should make some sort of a difference. The back side of the AMC7135 is almost entirely exposed, and having such a small area for connecting it to the copper seems a waste, at least if you’re not running traces under them… It’s such a quick and easy fix, and in any case it ain’t gonna make things worse.
Right, there’s no disadvantage.
FWIW the thermal resistance of the soldered joint between the central lug/pin and your PCB’s copper pour is already very low. The main point of having the big chunk of metal on the bottom of the package is so that the IC chip can be bonded onto it. If you peel the center pin/lug off of a 7135 you’ll see what I mean. That’s why the package is setup in that way.