The issue with cooling is that you either go with massive heatsinking like the GT4 for passive cooling and the GT4 is already at the limits of that.
Or you go with active cooling that is loud and annoying.
Active cooling can be improved but regardless of how it is done it will be really hard to make it truly weather proof over the long term since small fans simply do not play nice with real world grime/weather. It can work well in a lab environment though.
Having given a fair amount of thought to active cooling setups, there is room for improvement for sure but the cost would climb in a real hurry and the lights would loose practical uses with it due to increased size/weight along with it.
The best setup thus far I have come up with is a actively cooled design that gives up on weather proofing and simply makes the fan easy to replace when it fails. The cost for such a designs would be a lot more then existing designs to do it properly though. Like 2-3x the price.
I have one idea for a basic active cooled design that would net moderate gains but it only works with a specific flashlight setup. and would still be bulky.
Ding ding ding, Been saying this quietly for some time as I usually get heavy push back for even suggesting that the thermal path should be compromised.
I did this to an extent with my EDC and love the results.
You know, this is what they do in commercial lighting. Doing it for years with old style bulbs but betcha they are doing the same thing with LED bulbs.
Hhhmmm. deja vu - think we had this conversation before?
LOL, I actually still think that is a good option for the GT4 style light (assuming that it would be manufactured properly, which has been the issue so far, no one I trust to do it right), the GT didnāt output enough heat to make that necessary IMO.
This is a review of the Astrolux MF05 Flashlight and it is compared to the GT90.
You cannot tell the difference in the video beam shots !!
The GT90 is on the left.
.
You donāt really need to measure the LED temperature. The main reason for cooling is just so that you can hold the light without burning yourself. The actual LED can handle much more heat then your hand can. Over 200Ā°C. The main point of cooling is just to prevent you from burning yourself.
You can get a rough idea of the wattage being put out by the LED just by calculating the voltage and amperage. Iām sure the SBT 90.2 is well over 100 Watts.
Heat sink with a large surface area like in the picture can work with a very small point source of heat if you use a good spreader such as an intermediary copperplate. Copper tends to spread the heat out very evenly.
Note, Iām not sure I interpreted what you asked correctly.
The reason for thermal stepdowns and cooling is not about user comfort. Itās about not burning the user and getting sued.
There is nothing baffling about flashlights being round and no active cooling. The explanation is super simple. The round shape is dictated by the reflector. It doesnāt make much sense to machine a square flashlights and put a round reflector in it. Square is harder to machine and has extra weight compared to a round flashlight.
Adding active cooling tends to double or triple the cost of the flashlight and reduces the reliability. It also tends to compromise itās water resistance. This is why you very rarely see it. Nobody wants to pay the price premium for it.
The tongue was because at the very start of the BLF GT project a few years ago, I was sideways involved and suggested a box design with handle (including a crude design drawing) with the led directly on the radiator, like the COB heatsink above. That design was never followed, BLF was in favor of the classic style flashlight shape, which happened. Iām still grumpy about that because I still think I was right.
About measuring the temperature of your emitters: Iām all for measuring everything, but it is not needed everytime. After following temperatures and performance of a number of flashlight builds, over time you develop a feeling for how to mount your led, thickness of material for heatspreading, and size of radiators. I.e. that 13cm radiator combined with a SBT90.2 on a 30mm copper DTP board at 20A, will be fine (btw it will be bolted to a 20x20x0.3cm aluminium plate which adds a bit extra surface area for heat to radiate away). My educated guess is that after 8 minutes the radiator will be fully heated up and in steady state, about 90degC, by which time the 21700 battery already will be two-thirds depleted so that the current will drop significantly. Letās see after the build if I guessed that right
LOL, I still think the box idea is a really good one, just for a different style of light. I have a few ideas bouncing around that would make really good use of it but sadly have not found a manufacture I trust to do it properly and done wrong it would ruin the whole concept in the eyes of manufactures.
Also agreed that it is not necessary to take detailed temperature readings for flashlights most of the time. First off with modern copper MCPCBās the temperature of the shelf is pretty close to the temperature of the LED itself. Meaning you can get a really good idea of the temp with an IR thermometer on the outside of a light and modern LEDās are generally rated for up to 150c+, so it is highly unlikely you will be anywhere near that regardless.
Basically I have never had an LED die due to age, this is not a real concern for flashlights that would be lucky to get a few hundred hours of use with LEDās that are rated for 50k hours.
Correct. The heat slowly flows from the head of the flashlight all the way to the tail. The entire flashlight will start to get warmer and warmer. Then you have the ambient air temperature around the flashlight as well as how much air is circulating around it. Thereās both convection and radiation removing heat. If you happen to hold it in your hand then you also have conduction taking place removing heat through your hand.
Anything over 6,000 or so lumen is going to require active cooling, so there goes any chance of a low price.
Lower powered lights like the PowerTac X10000 Destroyer or X10k cost $700 and thats only 10,000 lumen.
The Microfire Challenger H5 was 5,000 lumen and cost about $475.
I think the light closest to 25,000 lumen and lowest in cost is the imalent R90C at about $400.
I did a thread on a lot of active cooled lights here if your interested. Itās not up to date, though. Imalents newer lights are not on there.
Very few lights use external power. I remember the old Olight X6 Marauder, but it was only 5,000 lumen and $400.
Any light capable of 25,000 lumen for just 1 hour is going to require a massive amount of battery capacity. It would almost have to be externally mounted. The only real application I could think of for the light youāre describing would be some type of search and rescue. This is a specialty application so it would be well beyond and above any type of reasonable cost. Not at all something the average flashlight user would need or could afford. It sounds like something a cityās Fire Department might use and a city can afford to pay the big bucks for a specialty light.
To reach 25k lumens, you are going to need to dissipate around ~250w as a rule of thumb (yes you can get better efficiency then this but it requires more LEDās and thus a larger light and less throw as a rule).
People greatly underestimate the amount of cooling needed for long term use. Check out the massive heat sinks that CPUās use WITH fans and consider that most of those are only rated for ~150-200w. So you would need one of those big CPU coolers with a fan or 3 to have a chance of cooling such a light. This rules out light weight and small size.
Then you have the costs to build all of this, as Djozz said, 2-3x the price is a good guess all said an done.
And after all of that, the power required to actually power this light would drain a 18650 cell in about 2.5 mins, or it would require a 20A+ connection to a car battery (cigarette lighter is only rated for 10A) OR it would require a power brick the size of a xbox 360.
There are always going to be trades offās like anything in life. If you want high output and long runtime, you throw size, complexity, power constraints and usability out the window (think work light or off-road truck light-bar).
The GT4/GT94 is pretty much the best you can do in itās form factor for a passively cooled light. Anything more then this would need active cooling and a much higher price tag.
I donāt know. At this point your basically doing a laboratory based test. You can probably attach a probe or IR meter as close as possible to the junction and assume itās slightly less hot than the junction itself. I still would not trust there is direct correlation between the output in lumens to the junction temperature. Maybe there is, but youād need to test it to confirm.
With qty 10 of XHP70.2's putting out a total of 25000 lumens, it would use about 150 watts.
Of course running under wall power, we are not talking about a flashlight, but rather auxiliary lighting. It's really outside the realm of what we do here.
I dunno what those heatsinks look like, specially interested in the ones you mentioned for less than $50 -- do you have a link or 2?
Yep, I have considered many of these things as well. As djozz pointed out earlier, besides the technical limitations, the biggest hurdle is cost.
To make a light with a Noctua style heatsink you would easily be pushing the price over $1000, most likely WELL WELL over that since it would be a very small quantity light and thus very few lights to spread the R&D costs over.
That said, Djozz box light idea is the best compromise I have heard so far but still presents a lot of design and manufacturing challenges. I have put feelers out for such a light a few times but everytime decided to not pursue it as the manufactures were just not ready for such a light.
It will be a big departure from what they know and having been behind the curtain so to speak, I just simply have not found a manufacture that I think could pull it off properly yet. That is even before trying to convince them to give it a try knowing there will not be much profit in it (which means it would get even less attention in the design phase when it needs far more).
There are a few lights people have put together on youtube using CPU water coolers for big output, they are really neat but very hard to turn into a practical flashlight.
Interesting stuff!
I barely use my GT90, but iām still super happy with it.
1Lumen.com reports a noticeable difference in output between the GT90 and MF05 though, so it might be that in real life the difference is barely noticeable.
Another reason to stick with the GT90
, I still think that the GT should have been a box-design with a fat 15cm x 15cm flat heatsink directly behind the led 
