Thank you Justin,
You mean like this?
And this new combo led/board goes glued on the existing star (after the old led removal)
using the thermal epoxy?
Yes, reflow the XP-G onto a datiLED MCPCB (or a Sandwich Shoppe MCPCB). IIRC, for a datiLED MCPCB, you then need a shim of about 0.06" thickness to raise up the LED die to the same height as for the Luxeon III emitter that you are replacing. If you have a metal punch, you should be able to punch out some shims of copper or aluminum sheet. Thermal epoxy the shims to the MCPCB to get to the desired height. You might be able to punch out a single shim of 0.06" thickness, but it is definitely easier to punch out a shim from thinner sheet such as 0.03".
The existing star in the Gladius has a circular hole in the center, within which the LED sits on the Gladius heat sink. That hole is probably about 10mm in diameter. That is one reason why keeping the star is helpful. You can easily center the replacement XP-G by eye.
This is the view of the stock Lux III emitter sitting in the circular opening of the star:
Note the LED + and LED- "legs" that connect the Lux III to the star. Those are the connections that you will need to emulate using strips of copper, bent appropriately.
Once you de-solder the legs, pop off the LED, and clean up the thermal epoxy, you should have a nice heat sink surface to glue down your XP-G.
As I mentioned before, the advantage of retaining the Luxeon star is that everything is already centered for you. Once you have your XP-G2 on a shim of the right height, all you need to do is glue it in the hole in the star and that makes it very easy to re-center the new LED. You can do it by eye, or you can drop the reflector over the LED to self-center it. Once centered, let the epoxy cure.
The next photo shows the bottom face of the reflector. The "ribs" that you see surrounding the reflector opening fit in the narrow gap between the LuxIII case and the edge of the hole in the star. That will give you your self-centering action. The ribs are not continuous. The gaps provide clearance for the LED+ and LED- legs. Also notice that one of the three reflector feet is clipped off (the one at about 12 o'clock). That's because that foot would normally drop into the recess right where the red and black driver wires cross above in the first photo. The other two feet sit in the recesses at about 1 o'clock and 5 o'clock.
Once you have the XP-G/XP-G2 on an 8mm board and shimmed up, it is physically just like the stock LuxIII, except it doesn't have the LED legs to connect to the star. What I did was cut some strips from copper sheet and bend the strips to emulate the LED legs of the LuxIII:
Reassemble and you're done. The copper is thin enough to clear the edge of the reflector. Regular wire is too thick to clear the reflector edge.
The Gladius uses a two-board driver:
Here is a shot of the top board of the two-board driver:
Looks like the Gladius uses a TPS64200 step down controller! Interestingly, the datasheet says that the recommended max supply voltage is 6.5V, with an absolute max of 7V. 6.5V is right at the open circuit voltage for two 123A cells in series, which doesn't seem to offer a lot of safety margin. However, several folks on CPF, including myself, have run the Gladius on 2x16340. You would think that with an absolute max of 7V that the driver would instantly fry itself. So perhaps the TPS64200 is conservatively rated. I've also successfully run my Gladius on 2xLFP123A.
The sense resistor also looks like it is 0.10 ohms, giving a nominal drive current of 952mA. I measured something like 985mA drive for one Gladius, so the calculated value is pretty consistent. Lux III Vf seems to be typically about 3.8V, while XP-G/XP-G2/XM-L/XM-L2 all seem to be in the 3.0V range. So you get a big jump in light output along with a 20% decrease in power draw.
The other components look like a 5uH inductor, ST Electronics STPS2L25U Schottky diode, Kermet 10^7 pF 10V tantalum cap, and some unknown brand of MOSFET.
The orangish ribbon cable seen above connects the driver board sandwich to three Hall effect sensors (only two shown below) that are positioned at the bottom of the light (at the tailcap end):
This might be a useful tip. If you do lose that annoying ball bearing, an expedient substitute is to cut off a short stub from a 3mm metric machine screw or a #6 machine screw. All you need is a chunk of metal to fit into the ball bearing recess to keep the heat sink from twisting around (risking shearing off the driver-to-LED hookup wires and possibly the ribbon cable) when you screw down the head onto the body.
