How hard can a XM-L2 be driven ? (BARRETT40 Super Thrower Project.)

I have seen some lights & mods where an XM-L has been driven to 6 amps plus, but usually always limited by the cooling ability/heatsinking of the light.
As the project Super Thrower is under construction, (BARRETT40) i have been pondering over a design of a hybrid Liquid cooling for the emitters using custom built copper heat-exchanger blocks, and a modified CPU cooling pump/radiator system from Performance Computer cooling parts. Being a custom Gamer PC builder, i have lots of L/C parts kicking around, and with the machine shop i have designed a compact & unique way of custom building Liquid cooled copper noctigon star plates for the four Defiant Super Thrower heads that will be used in this project.
Basically this will keep the emitter base stars at a constant cool temperature regardless of load, as the thermal eficiency of liquid coolant and a 120mm radiator with variable speed fan is many times greater than any passive cooling-fin design of traditional lights, meaning cooling the emitters will be constant regardless of their heat output through the noctigon copper base.
My goal for the BARRETT40 is driving the four XM-L2’s as hard as possible to achieve the most lux/lumens they can on a constant cooled base.
This project is without cost limits.
if i need to use two, or three Skyray King bodies to supply the amps to the emitters, then so be it.
Twelve Panasonic 18650’s in parallel can provide alot of amps, and the L/C pump & fan will be driven by a seperate battery pack.

This project is going after a record throw distance.

All the parts are already in the machine shop on a bench and being worked on. So far i have tested one of the modified DST heads using a dedomed XM-L2 driven with a SST90 driver to 5 amps, and it out throws my Olight SR90. multiply that by four times, and push the XM-L’s to an amperage that they beg for mercy.

any suggestions or thoughts ?

I have some thoughts....

the reason that the xml2 blows (I managed to do that at about 9A) is that the bond wires burn through in the middle of the silicone dome (based on the maximum current of the xpg2, I expect the xml2-die to max out at about 12 or 13 A, but the bond wires are the bottleneck). If the dome stays on the led, the middle of the silicone is a difficult place to reach for cooling. However, if the led is dedomed (I guess that is going to happen anyway if the goal is throw), those wires are exposed to the air and that gives some more possibilities for cooling, although I do not have a suggestion. One thought I had was to carefully solder a blob on each bond wire to make it handle more amps.

Good point. maybe some air circulation too may help keep those thin bond wires from going “incadescent” and burning off.

Cost not an issue, record throw distance, please tell us more :slight_smile:

as i mentioned… i already modded one of the DST heads to beat down on my Olight SR90. J)

How about trying to take heat away from those bond wires? Put some non conductive adhesive on it, then a flattened solid copper wire that can wick the heat away. The fragility of those wires and limited space would make that a daunting task though.

so thats the whole point of this project, to beat a DST?
You should be able to do that without unlimited money, right?

true, if only CREE had used thicker Bond wires on the XM-L2 emitters.

well its already beating a stock DST, ( the modded test with one of the four DST heads.)

A project to savour :bigsmile: love out the box thinking and with definite aim 0:) at the end is a bonus.

Quick thought, if you are already designing a custom cooler plate/pad why not make one with a pedestal and mount the emitter directly to it. the noctigon would just add another interface between the led and the coolant.

I have been trying to consider all the various active cooling solutions myself, so I will watching this project with relish.

That reminds me, the gold/nickel base of the noctigon will reduce heat transfer slightly, as will soldering or reflowing a star to a heatsink, something to keep in mind, you may consider grinding it down to copper for this application

You may also want to do your own reflowing onto bare noctigons, making sure to press on the LED while reflowing to have the thinnest solder layer possible

Thoughts? Are you going for an XM-L2 record? Because the record is with smaller die leds, such as the XP-G2 used by the Deft-X, which claims at 1,000Kcd, four would be 4,000Kcd. I don't think a large die like the XM-L2 will ever get to that point, but you can probably easily hit an XM-L2 record if you are using aspheric lenses.

So thicken them. Electroplate them until they reach the desired thickness. Simple!

I don't have any thoughts on the cooling. I've been wanting to do a build with an acetone-based heat pipe, but don't have any practical knowledge to share.

I do think you will have to go at least 2S on the cells to achieve the current you desire.

I though you were going to do 7 DST heads? Or was that somebody else? Oh, and screw the SRK battery holders, just use an RC lipo pack. I’d go with a multi-cell pack and buck driver(s).

Like ImA4wheelr said. Parallel cells with not get you where you want to go. Unless you use a good bunch of cells LG D1 cells to power each emitter you will not see the emitter amps you desire. Even with those cells you will not see 6A+++ for that long (assuming you use lots of them)

Basically, you need to go with a buck driver.

Some useful threads:

High current drivers for single XM-L2

At what current do you kill LEDs - which driver was used?

The HX-1175B (Some know it as the Lustefire triple etc) seems to be one of the drivers that can do very high amps without issues. I might be doing a build using 3 of these at more than the 6,4A I down adjusted one to drive an XM-L2. Time will tell If I can pull it off and when emitters will die..

Like djozz said. Your main worry is probably the bond wires. Im skeptical that you will able to run it at higher than 9A even with your cooling. But I hope that you prove me wrong. J) Even 8A may not be reliable in the long rung.. who knows..

Another potential issue are current spikes from the driver circuit or whatever it is. The MF TR-3T6 driver seems to be killing emitters in the 6,5-7A range. Hopefully the HX-1175B will not have any signs of such behavior that might limit the peak amps where you kill emitters.

Lots of variables when going for the extreme. Be prepared to kill some emitters and find the limits of your own setup.. Maybe start with some cheaper XM-L2 T6 1A when finding the limits.. ;)

Either way, with low emitter temperatures you should see nicer figures than anyone else.

I ended up using a couple of the DSTs for other projects. I thought about a RC pac, but i have a fiew SRK parts and a ton of 18650s alredy.

Dale had some success repairing bond wires - maybe the same technique could be used to bridge the connections, eliminating (or at least reducing) the weak link.

I did a cheap quick test of liquid cooling and showed that it does help. I also was inspired by computer cooling methods.