Rewiring XHP-70 for 3V use

I’m sure lots of us have wished very bright emitters like the XHP-70 could be used easily with single cells. I decided to try to rewire the XHP-70 so that all four of the dies were in parallel. I bought an XHP-70 P2 1C on a 20mm MCPCB in 6V from MTN and did some surgery.

In 6V orientation, there are two 2s strings in parallel. The two dies on the positive side of the MCPCB have their positive contacts connected to the positive contact of the MCPCB. The two dies on the negative side of the MCPCB have their negative contacts connected to the negative contact of the MCPCB. I will call the dies on the positive side of type 1, and the dies on the negative side of type 2. The negative contact of each type 1 die is connected to the positive contact of the type 2 die closest to it. So, to make all the dies connected in parallel, I need to sever the connection between the type 1 and type 2 dies, connect the negative contact of the type 1 dies to the MCPCB negative, and connect the positive contacts of the type 2 dies to the MCPCB positive.

With these dies, the small bond wires leading to the top of the die are the positive contact. The base is then the negative contact. I sliced the dome off as best I could with a razor blade. I then used an exacto knife to cut away parts of the silicone to the sides of the dies. I did some exploratory contact testing with a DMM; this is how I confirmed the die orientation I explained in the last paragraph. There is a thin contact running along the edge of the type 2 dies that is isolated from the type 2 negative. This contact is the type 1 negative and the type 2 bond wires are connected to this thin contact.

I very carefully cut away the silicone around the type 2 dies and very very carefully the silicone right around the type 2 bond wires. I looked through a 20x loupe while making these cuts. Then I severed the type 2 bond wires from this thin contact. Connecting the type 1 negatives to the MCPCB negative was pretty simple; I used solder to bridge the thin contact (type 1 negative) to the metal next to it, which is the MCPCB negative. After cutting away the silicone, the surface of the package around the dies is a silver colored metal. If the surface is scratched away, it looks like a copper colored metal. I scratched away the surface wherever I wanted to solder, but even then the solder did not flow/adhere like it would on pure copper. It adhered well enough, though.

Connecting the type 2 positive to the MCPCB positive was the hard part. After considering different wires to use, I decided on some 0.01” diameter magnet wire I had. I scraped the enamel from the appropriate parts at the ends and tinned them. I very carefully removed the silicone from around the bond wires so that I could place the magnet wire bridge close enough. I very carefully placed the magnet wire so the bond wires would lay down on the magnet wire, then epoxied it down. After the epoxy had set, I soldered the bond wires to the magnet wire and the other end of the magnet wire to the MCPCB positive.

I destroyed one XHP-70 before my method was established. For me it was very important to use the 20x loupe while I was making the very sensitive cuts. Hopefully the pictures below will make it clear what I did.



I decided on the Supfire L6 to use the modified emitter in. The C8-sized head is a good compromise between good throw and portability, and 26650 cells can provide more performance than 18650 cells. I read a review of an XML L6, and the reflector had a large opening that I hoped would be large enough for the XHP-70. The L6 is not readily available, however, and the only one I could find was from fasttech and it used an XPG. The reflector hole was too small for the XHP-70, so I sort of crudely opened it up with a file. I fashioned a reflector spacer out of the plastic insulator thing that came fitted around the driver spring in this light.


Currently, the beam has a pronounced donut hole. I think the beam would be very useable outside, however. Maybe careful adjustment of the reflector focus could improve this.

I’m using a FET+1 from MTN running Bistro FW. With a fresh Efest purple 4200mAh, I measured 11.6A through my ammeter, which I measured to have ~35mOhms resistance. So there should be a bit more current when in normal use. I measured 60kcd in the brightest part of the beam right outside the donut hole. Measured at 6m with a Tondaj LX-1010B.

Thats pretty slick. Thanks for sharing.

How far up “into” the reflector is the LED? I’m thinking that that donut hole in the beam may be because the LED is too far “back” and if you can somehow get the LED more “forward” into the reflector, it’d reduce or eliminate that hole in the beam.

Yep. :beer:

Thanks. This mod is less necessary/special now that decent capacity 26350 cells are available. I did the emitter mod before the recent news about the 26350 cells; I probably would not have done the mod had I known good 26350 cells were right around the corner. I would hope that with the 26350 cells, one could get similar performance and runtime as the mod above, without having to do the LED surgery.

Stock, the XPG was located pretty deeply in the reflector. From a bit of messing around, I found putting the emitter even deeper in the reflector improved the donut hole. Currently, while the donut hole is pronounced, it is pretty small, and I’m pretty sure I’m getting maximum lux in the area around the donut hole.

Awesome! Will you start a business with this because I would buy a few! How much lumen drop is there? If there is any…

Some of you folks ingenuity and soldering skill truly amaze me :beer:

Ha, no I don’t think I’ll start a business. I’m not sure about the reliability of the mod yet. I think the bond wire connections are pretty solid. I’m actually more worried about burning spots in the silicone. I was very careful to clean the silicone right over the dies and keep it as clear as possible, and there are no signs of burns in the silicone there. But to do the mod I had to make cuts in the silicone very close to the die edge, and there is one spot where a burn has started, either because there was a trace of flux left behind or some other reason the area might absorb an excess of light. So I will have to carefully cut out this small burned spot and hopefully prevent further burning.

I don’t think there is any significant lumen drop associated with the rewiring. The dome slicing would cause some lumen drop, but the die luminance is pretty good considering I’m getting 60kcd while driving each die at only ~3A.

That’s crazy! Do you have a lumen approximate?

I think you just made the brightest single 3V LED flashlight in the world.

Thanks! I don’t have a measured estimate, but given that it’s 4 XPL dies each driven at ~3A, I would estimate in the 4500 lumen range.

I like the idea of the XHP-70; essentially the output of a quad XPL, but in a smaller package which enables more flexibility in host and reflector choices. With the introduction of decent 26350 cells, it will be easy to get this performance in relatively small lights.

Very impressive :slight_smile:

Pretty neat handy work, to bad Cree didn’t do this I the first place.

Dude!! You’re a badass!!

Doing that work to something that small is just ridiculous….

Hat’s off to you. :beer:

Thanks for the comments. A short update on this light: I cut out the small portion of burnt silicone at the edge of one of the dies and it seems to be good now with no other parts burning. I did a tailspring bypass and gained 4kcd so it now does 64kcd. I measured the current in its in-use state by charging the Efest 4200mAh cell, running the light for 1min, then charging again with my hobby charger to measure the charge put back in. 228mAh was put back in, which means the 1min average for the current was 13.7A. It’s a shame I haven’t gotten to really use it outside :FACEPALM: , but man it is a bright little light.

Nice work!

That was slick my dear!

Cheers ^:)

I only found this mod thread now - pretty awesome job! I don't think the 26350 obsoletes this mod, because the 26350 cells I have/tested are not low resistance, and don't match the capacity and low resistance of the new LiitoKala 26650 and it's re-labeled equals.

Plus nothing special is needed in the driver (zener or LDO).

Thanks! I remember when I first came to this forum and learned about the world of flashlight modding your threads were very educational.

I agree there are definitely some advantages to to a single cell setup.