wishing for a 5x XM-L, 2x 26650 light

Do they even make non LiMN D li-ions? There's really no point in underdriving XML's. What I want to see is an XML triple, with 3 2.8 amp drivers, behind carclo lens, in a maglite 1C off of one 26650. 8.4 amp draw, shouldn't be too bad.

I don't know what the definition of "underdriving" is, but I meant "anything less than the stated maximum current specified by Cree". Therefore, if 3A is the stated max, then to me, 2.95A is underdriving, as is 0.05A. To clarify what I wrote previously then, I'd be happy with any current 2A or higher (understanding that an XP-G is an alternative at 1.5A and lower).

If I remember correctly, the lumens vs. current chart has a gentle convex curve to it, meaning that the LED is more efficient at lower currents, so running three of them at 1A gets you more output than running one at 3A. Or, running five of them at 1.8A would get you more output than running three of them at 3A.

But correct me if I'm wrong.

The WF-500 head is big enough to accommodate five XM-Ls, but I would also be interested in a smaller light with three XM-Ls, similar to the one srfreddy described.

So, how do we get the attention of Ultrafire, Yezl, Brinyte, Mr. Lite, etc, to ask them to build these things for us?

I just ran across a MicroFire K5 Tactical (now called Challenger II L2000R) which has 19 Q5 bulbs. Coincidentally, I don't have $750 to drop on this bad boy. And all we want is a puny 5 XM-L set-up.

be seen, not entirely true. the more efficient, the less heat.

heat is one form of energy, so is light. more light, less heat, but you are correct, it still requires TONS of heat sinking

4sevens has made a prototype 18 XML searchlight. It uses 32 18650's though, but it includes a charger *phew!* Now all he needs to do is size it down into useable light amounts......

I'm sure that one will fall under the "If money were no object" catagory. LOL

3-XML in a wf-500 head with 2 24-32mm diameter cells would be amazing. I can't see any budget light manufacturer doing that any time soon though.....

Yes, KD will sell you them.

http://www.kaidomain.com/ProductDetails.aspx?ProductId=2708

They are longer than a D and flat topped so you need a shorter spring and magnets to make them work in a stock Mag body.

Im currently using an old school CPU heatsink from the Pentium4 era (a fist sized block of alum) and it can manage to passively cool a sst-50 driven at 18 watts, bringing the temps up to around 20 above ambient for 30 mins of running. Thats a cool 40 degrees C in a 20 degree environment. Those are massive heatsinks, and just are not practical for use, even in a torch. The Legion (on the right) has a very substantial heatsink by normal torch standards, but when placed side by side with the CPU heatsinks, its not fair. The other torch was a discount retailer (BigW) torch that was bought for $15 or something. This is the 4AA version, I also have a long tube version (4D cell) of the same torch with interchangable parts. That torch would have a real heatsinking problem if modded to 10+ watts.

I've also got a rough plot of the measured temperature of the base of the heatsink in my torch over a medium run time. Essentially, looking at the graph, Temp is on the left axis in deg C, and time in seconds on the bottom axis. Temps are seen to rise until I turn the torhc off at 14.5 mins. Extrapolation can show that this torch will continue to rise in temperature if left alone, maybe within an hour reaching 80 degrees C, measured at the base of the heatsink (closest part to the emitter).

This is the Legion, the torch on the right. The thermocouple was placed into the last heatsink groove from the outside. Still a very respectable performance for a torch thats drawing 18 watts from 3x 18650s. Im not sure what emitter power is, but its "apparently" meant to be 90+%ish so even if 16 watts makes it to the led, and half of this is heat.

Dissipating that much heat (30-40W) in a manageable sized light is just not possible. Not with any kind of reasonable design. You'd need lots of very thin (re: fragile) fins and heatpipes to get the heat away from the slug. Look at modern/high end cpu heatsinks and you'll see whats necessary to get rid of that much heat.

EDIT: I just want to clarify by saying, that you can build a light that burns though 40 watts, but thermally its not going to get rid of enough heat to run for extended periods.

Not quite that bad - 30-40W is Pentium 3 territory and CPUs probably care more about their internal temperature than LEDs do. Modern CPUs can easily go to 150W.

Still not easy. My SST-90 light is a 35+ watt LED light and it will overheat if run continually on max for more than 30 minutes. Since the 4 Cs in it last for 20 at full output it isn't an issue for me.

We may have to start thinking about the duty cycle of lights. This is already necessary in small 10440 powered lights. The ITP A3 EOS Upgrade (What a mouthful - A3 from now on) can run on 10440s but the heat on high will rapidly become unmanageable.

You're right that a LEDs can handle higher temperatures. I was just trying to say that in order to make a light that doesn't just keep getting hotter and hotter until you can't hold it anymore, the design has to include a way to get the heat OUT of the flashlight rather than just giving it a larger reservoir to store the heat in it until the batteries die. :)

I've got a XML driven @ 3.1A in a 1x26650 light and within 10 minutes its hot enough that I can't stop thinking about how hot it is. LOL

Basically I agree with you. However you work it, 30+W lights are going to need a LOT of attention paid to getting rid of the heat.

And fans are highly undesirable. I can't see something like this being integrated into a light.

And that was the most likely looking one I came across. Heatpipes are probably going to be needed for handheld use but they are much less effective in non-stationary applications.

I do think fans are viable in larger lights though.

Again a relatively static application where they will work far better than in a hand-held light. At least most of the time it'll be in a relatively constant orientation. Unless you are in the habit of falling off a lot ;)

Modern heat pipes are not reliant on gravity. They work well in any orientation. The fluid moves by capillary action not by gravity.

I looked up the Cree data sheets to get some approximate numbers. I used a current of 800 mA because that's what the 5x XR-E module drop-in for the WF-500 produces:

XR-E Q5 (couldn't find R2 info):

- Vf @ 800 mA = 3.6 V

- flux @ 800 mA = 210 lm

- total power dissipation for 5 LEDs = 14.4 W

XM-L T6:

- Vf @ 800 mA = 2.93 V

- flux @ 800 mA = 315 lm

- total power dissipation for 5 LEDs = 11.7 W

So in the WF-500, doing a straight swap to XM-Ls should give me 50% more output (not considering beam shape). And power dissipation will be almost 20% less. But the driver used in that 5x XR-E module can be modified to push out more current. So at the very least, I could modify the driver to produce 20% more current, so that power dissipation is the same as now. Otoh, I've used my WF-500+ ("+" means the stock xenon light with the 5x XR-E module) for several minutes at a time on high, and the light doesn't get uncomfortably hot. Therefore, I'm willing to drive it a bit harder than that, so let's say 40% more than the stock 800 mA, which is 1.12 A. Ignoring driver heating issues for now, the higher current should produce:

- Vf @ 1120 mA = 3 V

- flux @ 1120 mA = 420 lm

- total power dissipation for 5 LEDs = 16.8 W

16.8 W isn't that much more than 14.4 W, so I'm sure the WF-500 body can handle that for a few minutes.

Now the total output becomes 2100 lm though, not OTF of course. With the stock XR-Es, the total output was 1050.

What if I bumped the current up to 1.3A?

- Vf @ 1300 mA = 3.05 V

- flux @ 1300 mA = 490 lm

- total power dissipation for 5 LEDs = 19.8 W

- Total output = 2450 lm.

For the WF-500, maybe those few hundred lumens aren't worth the extra 3 W power dissipation. But the point of all these calculations was to show that we should be able to get a light like the one I'm wishing for:

- 5x XM-L

- 2x 26650 or 32xxx cells in series

- body & head with good heatsinking (the WF-500 has not much heatsinking)

- 2500 - 3000 lm at the LED; 2000 lm OTF or more

Edit: what about 3x XM-L at 2 A?...

- Vf @ 2 A = 3.18 V

- flux @ 2 A = 685 lm

- total power dissipation for 5 LEDs = 19.1 W

- Total output = 2055 lm.

So yeah, I'd prefer a 5x XM-L module, driving them at lower current.

Computer heatsinks are great at dissipating heat in the order of 100 watts! but they are fan forced. Passive heatsinks that can handle 100 watts are huge and heavy!!!

Even the actively cooled CPU heatsinks are quite substantial and not ergonomic for torchlights. I need to source more metal.. but thle only real solution is just a big heatsink. Have a look at the SR90. Nice simple machined fins, purely functional, and the only way to do it cleanly in a torch. Im also not a fan of active cooling unless I have software means of reducing light output based on heat in the event of failure of the fan, which is quite possible given the highly uncontrolled and dirty environments torches can be put though.

Exactly right there, What too many people and manufacturers really dont seem to understand is the need to achieve a system that is able to DISSIPATE that heat. I've seen too many call a block of metal within a torch a heatsink, or have drop-ins with fins...

The idea here is that heat has to get OUT of the torch, and into the external environment.

There are effectively 2 concepts here, Depending on what you would like to achieve.

1) Thermal Mass - short term

- A big chunk of metal is fine, it will sink the heat away from the source, but will only act as a resevoir. Lets think of your source of heat as a running water tap. When you turn on your LED, the water (heat) will flow out. If there is a big thermal mass, its like having a big container, which can hold the water. A bigger container will take longer to fill up, but it doesnt get rid of the water.

- Thermal mass is useful for short term heatsinking (in the order of several minutes). More thermal mass means we can run the tap faster than it can drain out the bottom, but only for a couple minutes.

2) Thermal conductivity (to external environment) - Long term

is effectively how well you can get the water out of the system. Lets imagine having a hole in the bottom of the container. A small hole will drain slower, and is like having a small torch with small surface area. A torch with a big heatsink, lots of surface area is like having a large hole in the bottom of this container.

- Thermal conductivity is the ultimate flow rate, it determines how big an LED we can use continuously, and also how short the cool down period is between over-driven systems.

Some Examples.

a) Small torch, small thermal mass. Small hole in a small container. It wont take much to fill it up, and empties slowly, so we are really limited to a slow flowing tap, a small LED.

b) Small torch, big chunk of metal. - A small hole in a big container. It will take more water to fill up, but drain at the bottom is still slow. Becasue there is more water, it will actually take even longer to dry out (cool down). Think minimag with a big copper slug, We can use this torch for a little longer, or we can use a bigger LED, but only for a couple minutes, and it will still take some time to cool down. SOO here we can use a bigger LED, and we have gained short term output capacity, but no long term gains.

c) Big heatsinks, Small thermal mass. - Big hole in the bottom of a small container. Fundamentally this small container is still prone to filling up really quickly, but becasue of the faster drain, we can have a larger long term flow of water from the tap (bigger LED). Soo we have gained an increase in output for the long term, but we cant really handle much more in the short term.

d) Big heatsink, BIg thermal mass - Large reservoir combined with a big drain hole. We can have short bursts of really high flow rate, but the big hole still allows decent drainage rates. This is a system that allows increases in both Long term running and enough mass for short term bursts.

Excellent analogy. You explained it far better than I could. :) I think the only thing that isn't 100% is that you don't really have a tap. A tap will flow 24 hours a day.

What you really have a BIG bucket and much smaller bucket. It would be nice for a light be able to run forever, but how many people run a light until the battery is depleted and then replace the battery and keep going?

I'm coming from a PC building/overclocking background where you get your temps as low as you can and expect them to reach a certain level and then hold even under 24 hours of 100% load. I've got to convince myself that this isn't the same thing. Ideally any light would be able to dissapate all of the heat it generates, but we aren't talking bout a light for Joe Sixpack to keep in his junk drawer for when the power goes out. If I were building lights for the public, I'd make damn sure those would be able to run forever without thermal runaway. For an enthusiast the rules are a bit different, BUT improving the lights ability to dissapate heat increases its useability. The only other thing we can do is wait for the efficiency of the LEDs to improve, but thats boring. :)

Just something to think about: Can you comfortably hold a lit 60 watt incandescent bulb?

BTW, I think I want 3 XM-Ls direct driven off of a LiFePO4 32900. :p