Testing a Cree XHP50.2 J4 3A led

I am seeing it said that converting voltage and current to wattage is expressed in terms of heat. Like a 3V emitter at around 3A making 10 watts is 10 watts of heat. Somehow that isn’t ringing true, what am I missing? I know that emitters are more efficient at lower currents, less efficient when pushed hard. So the simple voltage x current equation isn’t all there is to it… heat comes in where the inefficiency takes over, as the idea behind an LED is to convert energy to light, not heat. So it would depend on the efficiency how much of the power being applied is being converted to heat. Not all 10 watts, for sure.

This can be seen in a triple or quad XP-G2 as compared to the same set-up with XP-L2 emitters. Pushing the G2’s hard results in a higher percentage of heat as compared to light output while the L2’s will fare better at lower temps while making considerably more lumens. Thus, the Manker quad in 18650 that gets ridiculously hot in seconds… the small emitters are being overdriven and producing much more heat.

Just ruminating on proper thermal design having to take into consideration the end goal of the output target as part of the equation, as well as which emitter at what power level. On the maximum end of things, it gets a lot more complicated huh? :wink:

LEDs might be around 40% efficient at low currents. At maximum power maybe half that. The rest of the consumed power is heat.

So in a 10W set-up, only about 50% is heat. Undoubtedly much more than that in a 120+W set-up…

It depends on the LED but generally 30% is a good number. So yes, in a 10W setup, about 7W will be converted to heat. Although if the LED is overdriven then that number could easily drop down to 15-20% or even less.

Since most flashlights are overdriven, talking in terms of wattage in +/- wattage out is good enough. We are not doing exact calculations, all that matters is that a 20W LED setup will create about twice the heat of a 10W setup. Exactly how much more can be skewed by 10-20% depending on the LED efficiency but that is not worth worrying about for what we do.

What? That’s not correct. Why are you thinking this? You want the heat source to be exposed to as much air as possible. So get the heat to the outer surface.

Your using the word heatsink incorrectly. The pill and mcpcb are not heatsinks. Only the outer parts exposed to air is considered a heatsink. The inner metals are more like thermal masses.

When you say potting, are you asking about thermal grease or thermal epoxy? Are you trying to make a permanent connection or temporary?

Common products are from Arctic Silver. They have stuff to glue things together and Arctic Alumina Thermal Paste which does not glue. I’m not sure where you live, but you can check out the thermal products from MTN Electronics. Thermal Products

I’m trying to help. If you keep calling them heatsinks you’ll just be confusing people.

Anything that absorbs the heat out from the emitter is considered a heatsink. But as soon as the heat reached equilibrum, you need to get rid of them to keep the heat flowing (from the source to somewhere else). In order to get this, we create fins or any other means to increase the surface area so the heat will released to the air by radiation and convection. The higher the temp difference (delta temp) between the hottest point(s) and the cooler parts of the heatsink the greater the heat flow. The heatsinks designed like this is considered a heat exchanger as well. An MCPCB is a heatsink to some degree, and will perform so much better with finned heatsink. With let’s say, 1 watt of power a bare 20mm sinkpad is adequate to cool an LED. You can roughen the back of an MCPCB, adding fins, etc…and it will become a better heatsink too. But, at what power level?

Some automotive headlight still use this approach to keep everything simple. It works but not as efficient as the open air design. The idea is to increase the hot surface area point from 1mm^2 to hundreds or thousand times so even the low conductivity of air will still let adequate heat dissipation to the surroundings.


Older PETZL headlamps only use few connected vias in their water tight enclosed plastic casing too.

I’ve seen worse design in OEM design with NO fins and it works wonder. And it doesn’t even involved MCPCB , just plain old FR4 board with good calculation of copper spread. Check this:

As long as the LED kept within it’s designed operating temp, then it will work just fine. We in BLF usually takes everything towards maximum performance/efficiency while in the real world it’s not the most economical things to do.

- Clemence

If you can keep the LED compartment perfectly sealed, filling the fin cavity with non conductive/corrosive liquid will transfer the heat better while keep everything neatly packed. Some also use fine silicone carbide abrasive sands.

The above makes no sense. Do you understand what fins on a heatsink are for?

It’s simply to increase the surface area which allows the air around it to pick up heat. If you fully enclose the heatsink in a box then the fins become useless. You’d be better off not even using the heatsink.

If this were a high powered flashlight that was used in short bursts I’d replace that heatsink with a big chunk of copper. This will allow for a longer turbo time.

Since this is a light that is going to run constantly, increasing turbo time is not important. I’d just let the heat flow through the gold part and heat up the outer housing. Then the outer housing will be exposed to moving air and pull the heat out. With enough surface area in the outer housing and enough moving air the lights temperature should reach an equilibrium within a safe range.

The above is not ideal, but do not say it makes no sense. If the heatsink is going to be enclosed it is still better for it to be finned. You might not be able to understand what is meant by his term heatsink but the rest of us can clearly see the large aluminum heatsink with extruded fins. :wink: And stormrider is correct that filling the air gap with potting epoxy is still not ideal but better than air.

Stormrider: I pot flashlights with a mix of silicone(rtv) and silicon carbide lapping powder. It works well for flashlights but might not be economical for your application? Outherwise you can buy potting epoxy by the gallon for descent price.

Are you able to pot and seal the front of the enclosure and cut some vents or openings around the heatsink fins to let in some air? I think this would give the best performance but maybe a trade off with aesthetics?

@lightrider: BTW, several years ago I tried filling a Pepsi aluminum can with #320 silicon carbide abrasives to cool 3 hard driven XPGs in the center (absurd art objects). Man…it’s heavy :laughing:
But the heat was indeed spread to the can’s surface fairly well. Got better result with damp SiC sands (better than water alone).

Haha! Ya, I guess it would be like a cement brick. A cool brick at least. :slight_smile:

I’m losing my patience and I don’t have time to give you a lesson in thermodynamics.

Basically, empty air or solid chunk of aluminum makes no difference in your application. Do the cheaper, leave it air. You will have the same performance.

I think you are confusing flashlight design with car mounted light design. They are not the same.

Sorry, no more help from me. Let someone else teach you. Your too resistant to learning.

Good luck on the project.

That’s not too bad. Haha:)
With the small power you are going to be using, you wouldn’t need to expose the entire heatsink. But any airflow will help significantly. I understand that this is what JasonWW is trying to say. But with this amount of power, i think you will be fine with potting if that is what you want to do.

I’m not the one to ask something like that as 1) I do not like the XP-G3 emitter and 2) I don’t recall ever putting any light together that pulled such low amperage.

As far as your drawing goes, if you can’t put a heat sink in the opening that fills the opening then you’re falling short of optimization. Cut off the open air portion and leave the heat sink exposed, don’t pot anything. Just because something is cheaper does not make it better, most of the time it’s quite the opposite. If it’s not worth making right, it’s not worth making at all.

Your goals is to give the maximum heat path to the outside of the light. If you have an empty cavity then your goal is to fill in that cavity with metal or worse case potting material / silicone to give the largest heat path to all outside surfaces you can. Adding heatsinks to the inside of an enclosed space will not help you much.

For your goals, if the driver will be external, then sure, go ahead and fill in the back of the light with silicone to optimize the heat path cheaply and easily. Then close it back up for waterproofing.

That light looks like it should handle at least 15-20W without an issue based on the results with similar sized flashlights.

If you’re using a constant current driver then the voltage of your XHP35 won’t be far from 12,7-12,9V at 1,2A. The LED voltage is current dependent. It’s unlikely to get an XHP35 running at 1,2A with 14V vf, unless it’s broken. At higher current the voltage will automatically increase. LED die temperature also affects the voltage: higher temp lower voltage and vice versa.
14V is your input voltage. Using buck driver even under no load the V out will be slightly lower than the V in. All the components inside the driver eat some voltage to work.

Correct me if I’m wrong here =)

- Clemence

I thought about the same things back when I had to mod my bike headlight. Tried to find any driver to drive XPG R5 at it’s max datasheet current (1,5A). As I learnt, it’s harder to design proper passive heatsink if I were to drive all the LEDs to their max rated current. More current = less efficient LED = more heat = more heatsink cooling needed. I ended up using boost driver and cram as many LED as I could and drive them less than 500mA each. Try not to use cooling fan, it has to be a very good long life IP66 minimum to be reliable. If you are using aluminum handlebar such as Renthal you can use it as a heatsink too.

Wait are we talking motorcycle or bicycle here? For bicycle or e-bike, efficiency should be one of your priority too.

triple 3p XPG3 quad 4p XPG3, XHP35, XHP50, or even XHP70 are the better choice if you can get boost driver. Less current - higher voltage - less heat - less power needed. A single boost driver from luxdrive is a good and very reliable driver. You can easily drive up to 40Vout using 12Vin at cooler running 350mA/LED. That will broaden your choice.
Or you can invest more, using hyperboost from Taskled, up to 80Vout with adjustable current up to 3,2A!! This is rather expensive at USD 40 but will unleash your creativity. It’s also automotive rated (withstand “dirty” automotive voltage spikes).
Taskled Hyperboost

EDIT: Don’t use luxdrive unless you sure the DC power source clean from spike and pulsing. I remember I destroyed 2 luxdrive because my older scooter still used half sine DC output. You can check your front light, if it pulsing on very low RPM then it’s half sine DC. With full rectified DC, Luxdrive is OK. Older bike lead acid battery also fluctuates heavily when you starts the engine, or when hard downshifting (hard engine braking). That’s why linear driver isn’t a good choice for bikes

  • Clemence