Difference between heat sinking and path to ambient

I see a lot of creative methods for what people call heat sinking and thought it would be nice to collect different ideas on what works(or does not) and why. My understanding is that a good heat sink forms one part of the path to ambient. To do the job well it has to:

(A) Have enough surface are in contact with the heat source.

(B) Enough mass to absorb the heat produced.

(C) Provide a means for the heat to escape

A failur to succeed in any of these could result in a bottleneck, trapping the heat and causing thermal shutdown or damage.

Poor contact or lack of (or too much) thermal compound and whatever mass you have is not being used effectively. Heat backs up to the led very quickly.

Good contact but the mass is too low and the heat backs up to the led.

Good contact, adequate mass( how much is enough?), but either not enough surface area or poor contact with the next conductor. Not as quickly but eventually the heat will back up to the led.

I will be adding more comments and some pics of things that worked and some that didn't. I hope others will contribute their experience and pics to make this a good source.

Proper material is important too.

Metals with high thermal conductivity are mandatory when pushing LEDs at higher currents. Metals like stainless steel, brass and titanium are poor (to very poor) thermal conductors. Ironically most of the "boutique" metals that CPFers pay high $$$ for are thermal insulators (as far as metals go), and trap heat at the source. While Copper, silver, gold are best at heat transfer. Aluminum is a good cost effective compromise.

Heres a list of thermal conductivity values for various metals:

http://www.engineeringtoolbox.com/thermal-conductivity-metals-d_858.html

Once the heat is conducted away from the source, it needs to be radiated to the surrounding air. The thermal emissive properties of black anodize make it a good black body radiator. Hence black anodize Aluminum is often preferred for heatsinking. The better thermal conductors (copper, gold, Silver) are not necessarily the best choices as black body radiators, so they will conduct the heat from the source, but they will not emissively radiate it to the surroundings.

The surface area in contact with the heat source is usually dictated by the size of the LED star and the efficiency of the thermal junction between the star and the heatsink is a function of thermal compound and pressure (which is why I prefer to screw my stars down if possible).

The mass of the heatsink is more relevant to the hysteresis of the heatsink temperature - a higher mass will "buffer" spikes in heat output.

However, at the end of the day, it's the rate at which heat from the heatsink can be transferred to the air which is the usual rate limiting factor and that's almost purely a function of surface area and airflow. More airflow for a given surface area increases the rate of heat transfer, lowering the surface area needed to maintain a given temperature (and vice versa). This is where torches/ flashlights suffer, as they're constrained by how much surface area they can provide but, unless you swing your arms wildly as you walk, experience relatively low airflow.

That's my take on it at least :)

Your right in the ideas behind heatsink design.

1) Thermal Resistance (internally speaking, so from the source to the outer surface) = materials, interfaces, thermal pastes, surface area of interfaces.

2) Thermal Resistance (to the external environment) - surface area of outer surface, type of surface finish.

3) Thermal Capacity (how much mass is involved) - size matters

What is the significance of these in flashlight design? In order of importance for longer term use;

1) Internal Thermal resistance - This determines how quickly the heat can be transported away from the heat source. This is fundamental to allowing high thermal output devices to be used. Having a high internal thermal resistance will simply cause rapid overheating as the heat energy is not taken away from the heat source. While the inside is cooking, the outside of the torch may still feel cold to touch. (so for a Given power consumption, the torch that feels warmer quicker is better)

2) External Thermal resistance - This is commonly overlooked, and is only achievable by having a large surface area (lots of fins), and having a thin dark finish can help with radiation of energy. Having a high external thermal resistance will prevent you being able to Maintain high outputs for extended periods. This will practically mean your torch will get hot very quickly and takes a while to cool off after turning it off.

3) Thermal Mass - This is simply the mass of material that can absorb the thermal energy, but once the whole item has warmed up, its ability to keep an emitter cool will diminish. So thermal mass is only a Buffer to extend the time taken to heat up.

Flashlights are Commonly designed with some thought on 1 and 3, but most fail very miserably on 2, simply because it is overseen, or is not considered important or aesthetic. A lack of incorporation of 2, simply means the heat cannot leave the flashlight very fast, and this creates the problem of limited runtimes on high output. This is a compromise that is somewhat acceptable in flashlights, think of the 500 lumen turbo mode on a CR123 XM-L, limited to a couple minutes before the thermal buffer is expired, and insufficient surface area to maintain the required thermal dissipation.

The only flashlight that comes to mind now is the X6, with a body that is Only heatsink. That torch was possibly designed with the idea of running much longer runs at full output with 6 XM-Ls, and they achieve that by providing Significant surface area to Maintain a continuous thermal dissipation that can keep up with all those emitters.

How important is it to have a dark finish?

On my modded Sipik 58, I stripped off the black Type II anodizing, so the finish is now just bare polished aluminum.

I gather this means the light might not radiate heat as well as if I'd left it anodized. How much of a difference does it make? Will any reduction in heat emission be countered by better conduction of heat to my hand since I usually run the light handheld rather than tailstanding?

Shiny metal is very poor at emitting heat. If you have an infrared thermometer, you have to put a piece of masking tape or similar on the shiny metal, to read the temperature of the object (the tape will come to the surface temp. of the metal by conduction quickly).

Whether the black stuff on the flashlight was high or low emissivity, you only can tell by using an IR thermometer to compare.

(similarly some black roof shingles don't emit infrared well, and some white ones do -- emissivity (emitting heat) isn't the same as reflectivity)

Look at any cartoon raygun and you'll see how a flashlight should be designed to get rid of waste heat.

I'm hoping we see these kinds of cooling hardware start to be modified for flashlight use.

http://www.diytrade.com/china/pd/7523520/5W_Spotlight.html

http://www.diytrade.com/china/pd/7512109/10W_Bulb_cooling_fins.html

(Notice most of these do look like shiny metal -- one would hope they're anodized or painted with a high emissivity coating. A coat of paint is all it takes, as long as the paint has a high emissivity. If they're polished bright and shiny for marketing purposes, they're as inefficient as they could possibly be!

(similarly some black roof shingles don't emit infrared well, and some white ones do -- emissivity (emitting heat) isn't the same as reflectivity)

Exactly the kind of stuff I was hoping to see. Things I didn't know. One thing I have been surprised to learn is how little total mass you need if the thermal path is very good. That is,1/8"of copper behind the leds and good solid contact with the emmissive surface is adequate for bike lights and dive lights where you have either good airflow or an ocean sized ambient. I often use 3/4"-1" copper caps with a second layer of copper added to increase thickness and found this sufficient when connected through the body of the light to more copper outside. I do not "fill up" the caps. They remain hollow. Pills made up of concentric cylinders of pipe I often make with extra layers more as a construction aid than need for mass. I look at the whole assembly and try to eliminate any bottlenecks. For hand held lights a larger mass is probably necessary as a buffer.

As more powerfull leds are produced(sst90 et al) the need for good heat sink design becomes more relevant. Cree is now producing higher voltage leds which run at correspondingly lower currents that should help here.