Frankenquads using 20mm 2x2 3535 MCPCBs

I posted about a quad build I did in the what did you mod today thread a few weeks back, but the recent discussion of CCT mixing in a SST-20 thread made me think it worth a new thread.

I’ve been interested in doing something similar to JasonWW’s Supfire L5 XHP70.2 Mod since I first read about it. I acquired a host back in the spring, but I didn’t get any further because I decided I didn’t want to mess with 2x26350s. I initially thought a boost driver would do the trick, but the combined expense of driver + emitter combined with driver power limits + limited choice of e-switch boost drivers and the prospect of new boost drivers from Mtn, Lexel, etc made it hard to pull the trigger.

While I was waiting, I started to wonder if there were higher-power 3v emitters. The SST-90 was an option, though an expensive one. Someone here had modded a (failed?) XHP50 or 70 to be a 3v emitter, which sounded great, if only it was available off-the-shelf. Then I learned about the Luxeon MZ, and that it was available in a 3v version, and high-cri to boot. Problem solved, right? Nope! I hesitated again, I think to wait for Led4Power to start offering higher-binned versions than what was already available in Q1.

And then I stumbled across these odd-ball MCPCB’s on AliExpress: 20mm, DTP, with four footprints for 3535 emitters arranged in a 2x2 array. The first listing I found was only available in 2s2p and 4s configurations, but I later stumbled across some 3v/4p variants. I knew the beam would be problematic with four widely spaced dies, but the MCPCBs were $1/each, and one already populated with XP-G2s was less than $5. I ordered some.

When they arrived, I put the pre-populated one into a dissapointing host. The beam was predictably bad, but promising. I tried stippling the smooth plastic reflector with a mist of clear spray laquer. It definitely helped, but the beam still had an unacceptable dark spot in the center. I figured I needed a better reflector, but didn’t want to waste money on the crummy host, so I set it aside.

A week or two later, still itching to do a compact, inexpensive, high-output light-build I started thinking about the emitter array again. I remembered that I had a OP C8 reflector, as well as a Thorfire C8s acquired for cheap that wasn’t getting much use. I reamed out the reflector, opened up a centering ring to fit the array, and built the light with a FET driver. The results, sadly, were little better than my earlier experiment, but, having come this far, I decided it was worth trying DC Fix. It did the trick! The light formed an even (very floody) beam after ~3-4 inches.

While I had the light apart at some point, I tried holding the stock reflector in place, and found that it too produced a good beam paired with DC fix, so I reamed it out and used it instead.

I was pretty happy with the light, but in the meantime, I’d begun to appreciate high-CRI emitters more, and I’d managed to accumulate a number of LH351’s that were calling out for a quad build. So, I made a solder paste stencil, and reflowed four of them onto one of the MCPCBs. I prefered the tint of the 4000K emitters I had, but I also didn’t want to run through my stock of them too quickly, so I decided to pair two of them with two of the 5000K variety.

The results were great. Looking into the lens, I could distinguish the two different CCTs, but in actual use they blended into a nice neutral high-CRI flood beam, and a rather bright one at that. With spring bypasses and 18AWG driver leads, I’m able to get ~18A on a fully charged 30Q.

Since then, I’ve done another quick build with some cheap 4000K 80+ CRI XP-G3s and a host with a slightly smaller reflector. The DCFix takes care of the XP-G3s ugly/messy beam.

A final build note, the MCPCBs don’t have “solder steelers”, or whatever they are called to help deal with any excess solder, so its more important not to overdo the solder paste than it is with other MCPCBs.


I tried a number of experiments with some unusual quad-3535 MCPCBs. You might be interested in the results:

The MCPCBs have inherent problems, but they can be overcome with DCFix if a floody beam is acceptable. The upsides are that they are inexpensive, enable high-output 3v lights without requiring a specialized host, or major mods beyond opening up the reflector opening, and provide a ready platform for color-mixing.


Update: Some beamshots:

Left: ~4’ from surface. Right: ~8” from surface.
Surfaces are different shades of off-white, so colors aren’t comparable between shots, but they should give an idea of color uniformity at each distance.

Nice, thanks for sharing the results. No beamshots?

I too have looked for high output 3V options. I was the one who rewired the XHP70. I thought about these MCPCBs, but couldn’t figure a way to make the beam good. I’m usually too concerned with beam intensity in my builds. For me a triple reflector/TIR is a better option, but as you mentioned this requires special hosts or modification of a spacer.

I knew there would be a question about beamshots. I am not happy with any of the real-world comparisons I’ve done, but I did just take some shots against blank surfaces at a couple of distances that I’m going to post and share just as soon as they make it from my phone, to my computer, and then up to my server.

I’m with you on the intensity shortcomings. My C8 is floodier and less intense than my D4s with 219c despite making obviously more lumens

I assume a bigger reflector would help with that, but then I’d be into the territory of less common and more expensive hosts, at which point the economics vs doing a high-powered single emitter or a multi-reflector or optic build. I also tried to find a narrow spot optic that would work, but the largest I found would still need boring out.

I may try another build with this MCPCB in a VG-10 style host, but beyond that, my high-output plans are high-CRI upgrades to a Q8, trying LH351D and/or SST-20s in my D4s, and another Sofirn C8F build.

Interesting. Subscribed!

Nice quad mcpcb led set up. I wish there was a XHP 70.2 mcpcb like that. :beer:

@CNCman, are you sure that would be safe? The heat density would be absolutely insane with 4xXHP70.2s close like that.

You’d need some indium solder, and liquid metal to actually prevent thermal runaway happening almost instantly.

Yeah, I posted those a while back, dunno if you saw that or did as I did and happened upon them at random. :smiley:

I was going to use them for lanterns, split the current 4-ways for more efficiency vs 4× the current through a single emitter.

3000K LH351Bs should work quite nicely!

Stick an aspheric lens in front of it and make it a nice flooder. Vs a reflector or even TIR, you’d have a perfectly even blanket of light.

Yeah, I saw one or two posts about these on AE but it wasn’t until I came across them on AE myself that they clicked as a 3v high-output option.

It already makes as nice flooder as I need with just DCFix. I’d really like find an inexpensive way to go the other direction.

I haven’t found an optic that fits. I picked up some that are supposed to fit XHP70 on AliExpress, but the bore isn’t nearly big enough. Of the more completely specified optics I’ve found, none have a large enough bore. I thought of using something designed for COBs, but the interface between the multiple domes of the emitters and the flat surface of the optic seems like it would waste a lot of light. Slicing might help, but that’s yet another complication, something I’m trying to minimize as much as possible.

If anyone has good ideas on optics to try, I’d love to hear them.

Maybe I can get some other diffusion film in a range of diffusion angles and see if there is one that gets me to decent beam uniformity without spreading the beam as much. Trying different films with the OP reflector might be an option as well. I need to look at XHP70.2 beamshots in C8-sized reflectors, though, to get a baseline on how narrow I can even hope to get the beam with a reflector.

Still say aspherics would be the best for that…

Was waiting for a build using the close quad mcpcb setup.

I second Lightbringer's aspherics :-D rulez proposal.

There are zoomies which even in flood suck with regards to out the front light extraction, but this can be solved with a dual precollimator + collimator lens setup if needed. That would create a wide, uniform pathway of light with high otf efficiency. ಠ‿↼

Cheers ( ͡ᵔ ͜ʖ ͡ᵔ)

Beamshots don’t look bad.

Using the larger XPL2s would reduce the size of the donut hole some.

Do anybody has a good driver + host + optics recommendation for this quad of Far Red?

In 4S configuration (10V), suitable 1S cell drivers can be either KD's H2-C (∅22mm) or KD's H1-A (∅20mm). Both drivers use 75mV sense voltage, thus for the H1-A you'll need to swap the stock R025 with an R050 to limit maximum output at 1.5A, as stock it comes ready to drive 2S emitters (XHP50/XHP70s).

For host I'd choose some zoomie. The BLF Cometa is on sale and takes a ∅22mm driver:

Cheers ^:)

Would they fit UTorch S1 Mini?

Thank you for the quick and complete answer. Unfortunately, I did read, that the CPF (not BLF) Cometa has a few design flaws, which can be dangerous (I am fairly newbie in modding, and I would skip troubleshooting that host, if possible) - and that host is rather big for now with its 260 gr net weight (without battery). It would be better, if it would be possible to make this build in a host with sizes roughly between Convoy S2+ and C8 :-)

/sliding a bit into offtopic starts/ I tried to think about a single emitter variant with S2+ host, but all I could find is a single mode 1.2A buck driver (probably efficient enough, just I'd prefer at least one lower mode as well - and it would drive a single led slightly over specified max. power). /offtopic ends/

Do you have any ideas for another host, and (maybe) 2S2P configuration?

Between Convoy S2+ and C8 there's the Sofirn SF30A zoomie, takes ∅22mm driver. The flood out the front output performance sucks a tiny bit because the emitter platform sits ≈5mm lower than it could, but no big deal. I myself have one with an :-) XHP35 HI. You'd need to widen the threaded emitter board clamping ring hole, it's made of plastic so no problem. Or you could try some SK98, I also have a high powered one with H1-A but good luck with the pill, I had to fix a ≈∅11.1mm hole in mine LoL.

I do not believe 2S2P configuration is the better one in this case, red emitters have low Vf and far red ones may go even lower. This means maybe 1.8 to 2.6V emitter forward voltage range. In 2S2P this is 3.6 to 5.2V, this is bad for boost drivers and for buck ones unless they're fed with 2S cells. On the other hand, 4S1P is 7.2 to 10.4V and good output voltage range for boost drivers fed with single cell, whereas 1S4P board is suitable for buck driver fed with a single cell also. The LD-29 (available at FastTech) is a very good buck driver but I do not know what is its maximum open circuit voltage when fed with 2S cells, it may be low so 1S4P would be the only safe bet.

With regards to hosts I cannot recommend anything better, of course I do not know if there's some reflector based torch out there with a huge enough emitter opening and a pleasing beam profile for frankenquads. Of course, you said nothing about donut :-D holes in the beam.

Cheers ^:)

Originally posted on Wed, 09/26/2018 - 03:25. Edited for a few fixes.

Very useful thoughts again! I think, I'll go for that SF30A host :-)

Do you think, that this driver could also work for 1S4P?

Well, you will need some driver size adapter ring to put that ∅20.8mm unit inside the SF30A. Should work well with a 1S4P board, but cannot say for sure.

The H2-C fits well, even with a tad of play (driver boot is closer to ∅23mm), and comes stock with 1.5A maximum driving current with progressive stepdown to 1A after 60 seconds, this is neat for a 4S1P XP-E quad I think.

Cheers ^:)