It will give purple, and it will be a bit disturbing because it is a combination that is pretty unnatural for your eyes (perhaps a colourful sunset could provide something vaguely like that) That combo is used often in display lighting.
Maybe I’ll keep the next one simple until I get a feel for the colours. Maybe a quad Amber.
Over the last couple days, I modded a ~20-year-old optical comparator. That's pretty young for a comparator, but old enough they didn't know what a power LED was.
The original bulb (IIRC some sort of inert-fill bulb with ~25 hour life) was deemed obsolete at some point, so the manufacturer came up with a retrofit kit using a halogen bulb; this was installed perhaps a decade ago. Turns out this kit uses an EKL, which is a 21V 150W bulb with a 40 hour rated life, but the power supply is a variable transformer (variable for soft-start) which maxes out at 14V under load. With a 30% undervolt, the bulb lasts forever, but it's not particularly bright or white.
I've been meaning to upgrade this for a while, starting when the big question was whether to use an SSR-50 or SSR-90, but I held off because the driver was a hassle -- it'd be easy to use some buck converter to drive a 3V LED at constant current on anything from, say, 9V to 20V, but I really wanted to keep the dimming functionality, even though it was no longer necessary as a soft-start. And resistor-driving an SSR-90 from 20V (14VAC rectified) means burning off a hundred watts or so in your resistor, so that's a no-go.
But this time, I decided to upgrade to Cree's finest -- an XHP35 HI (E2 3C, from RMM). As a 12V LED, it's pretty easy to run from 14V, and the 2.4mmx2.4mm emitting area closely matches the size of the EKL's filament.
I used a simple drive circuit with a bridge rectifier, big electrolytic capacitor, and current-limiting resistor. (Resistor drive because as I said, I wanted bad regulation.) There's one small oddity -- I put the current-limiting resistor before the cap rather than after, to improve power factor. Utterly unnecessary, but it makes me feel good. A 12V GPU fan is in parallel with the LED, so it sees no more than the LED's Vf.
I calculated the 2 ohm value for a nominal 1.05A current, but I didn't bother allowing for the transformer's impedance -- with the reduced load, the voltage is higher, and I knew I'd wind up with more current. Turned out to be 1.5A, which is fine with me, as DTP Cu star + 1" square Al bar + active cooling = ticket to overdrive city.
The previous shots are all at very low output, with the shop lighting on. Here's one at full blast, with the same lighting. (Taking pictures with my phone, can't be bothered to set same shutter speed, but this should convey an impression of it.)
As for final output... Using my phone to measure lux at the comparator screen, I got 70lux with the original yellow underdriven halogen, and 358 lux with the beautiful, overdriven, 5000k XHP35. Now I expect some difference in response to different spectra, so maybe it's only 4x as bright instead of 5x, but still -- a ridiculous improvement.
Very nice upgrade, it’d be interesting to see how the comparator works now as compared to how it worked before. 
Yeah, I meant to take before-and-after shots, but I got carried away putting it together and forgot the befores. I may go back and do it in a day or two -- the way I made the LED tower, it's really not too hard to change back.
I added GITD and Narsil firmware to my DQG triple today!
I just moded a piece of shit Ultrafire C8 i changed the LED from a LB and i changed the cables to 22AWG and i put an metal textured reflector on it. It had a plastic one and i tweaked the diver because i have no spare 20mm drivers and plus its a hollow pill so it cant do big power!
Never tried it in a hollow pill before but, you might be able to make a copper disk like I did for a couple of my lights for the MCPCB to sit directly on and that all sits on the driver shelf to help with cooling.
Also add under the driver shelf some copper mass , you can buy a 10cm x 10cm 1mm copper sheet for 3$ for ex from banggood .
I’m in the middle of a mod today and everything is going south fast. The undefinable thing that wasn’t in the plans. I’m taking a quick break before I head back to see if I can recover. I’ll have to go back to the start. Grinding may be involved. One optic in the trash.
There is a sliver of hope, so hopefully the hammer won’t have to get involved.
Yesterday I put an XHP-35 in the MaxTOCH Shooter 2X. This meant making a 2x cell extender, which I did by using a tube in 6061 and threading each end, then made a collar threaded on the inside to act as a sleeve to join the extender to the stock battery tube. This collar and the new extension are both knurled. I then cut an aluminum plug to fit into the driver shelf, with the same thickness on the shelf but about 8mm deep into the driver bay, this is recessed with it’s own bay for a 17mm driver and houses an LD-2 modified to work with 4S cells at 12V output and 2.5A current.
I wound up with 2.89A at the tail and 2346 lumens out the front, at 281Kcd for 1060M throw.
All in all, it worked out pretty well. I’ll get some 3000 or even 3400mAh cells and it should be good for some 2 hours run time.
No hammer involved. 
(I do know the feeling though Ouchy, when I soldered the negative lead into the via on the LD-2, a very thin wisp of solder made contact with a neighboring via which has a component on the other side… the light wouldn’t work. I couldn’t find out why to save my life, finally took it completely apart and was trying to identify everything before tossing it all in the bin and I saw that little tiny bit of solder bridged over. Removed it, put the light back together, voila! So yeah, I came really close to breaking out the hammer on this one too.)
The inital trial run, before knurling the tube…
At 97 yds to the red oil drum…
At 610 yds to the little barn…
This is the 5000K Neutral White version of XHP-35, a couple of bins below top power level and only in the 70CRI class. It’s an HI version, purchased over a year ago from Mouser.
Separate small problem encountered…and fixed.
My first reflow on a quad tpad. I followed the diagram for parallel and could only get half the board going at a time. Both halves worked, but not at the same time. Very frustrating.
There is only one + marking on the entire board. It took me some time with a DMM to figure out that the emitters orientation needs to be reversed on the right side of the board. A couple small + - signs printed on the board would have saved me some time, or at least on their mounting diagram.
There are little black dots on the corner on the positive side.
On the left, minus to minus in the middle, on the right, plus to plus. I missed one of the bridges first time and had one emitter not working. Then another time I had an emitter backwards. They are a bit tricky, and they appear to use a double layer of trace so all the traces aren’t visible to track for orientation purposes.
Edit: Sorry, the black dots identifying positive are on the Cutter T-PAD Triple XM-L/XHP-50 boards. Got that transposed, sorry.
Cutter T-PAD Quad, orientation for parallel wiring circuit
Red circles are positive,
Black circles are negative,
Blue circles are solder bridges.
Positive and Negative lead placement marked within the circles.
It’s easy once you figure it out, but the first time is a bit confusing. Leaves you scratching your head. I think the official parallel diagram should be a little more detailed.
Wow, nice mod with the XHP 35 HI, Dale. That emitter seems like the best way to get a large really intense beam.
Does the xhp 35 use an all new die? My understanding is that the xhp70 is 4 2mm xml-sized dies and the xhp50 is 4 xpg- sized dies.
Pretty sure the XHP-35 is 4 XR-E2 dies crammed together tight and wired in series. You can see that there’s 4 dies, but there isn’t a gap between them like the other two in the series (Extreme High Performance, XHP)
Still, the overall die surface is large and as such it really can’t be expected to be a serious thrower without going really large on the optics or reflector. It does throw, no mistake about it, but not like a single smaller die would.
All the same, 2300 lumens reaching out like this is pretty impressive. I like it, but for the length of 4 18650’s in series… I normally go for a smaller light.
Ah I see. Well 281kcd seems pretty good to me. That’s practically pretty close to the 360kcd of the stock dedomed xml.
If the dies in these multi-die emitters were wired in parallel that would be awesome. It seems, in principle, there isn’t anything preventing it. I guess some of the bond wired would have to be bigger.
We have to keep in mind that most of the emitter manufacturers are building emitters for overhead lighting, like warehouses, malls, parking lots, even roadways. An array of these emitters are put in a larger light bay and run at 120V or 240V, depending on where the market is. We are just adapting them to our needs.
Was playing around with an XML that had stopped working, learning a bit about the electrode architecture. Ripped the dome off and of course tore off the bond wires. Decided to reconnect the bond wire. It works again! Just a bit of fun; I don’t expect this fix to take a lot of current.
