New VirEnce MCPCB for E17/E21/119/144/233U

They’re so pretty. Can’t wait to order some!

Final touch by Texas_Ace :wink:

Awesome results, really excited to finally have a good option for 144A and E21A!

Great results!
These boards are also nice for an Oslon version, because the thermal pad of Oslon Black’s (that is connected to led-minus) can then be electrically insulated from the body of the flashlight.

Created many centering/positioning jigs, spent two VR16SP4 boards, some E21As (fortunately the R70), and one full day without success. Only later, I remember I didn’t stir the thermal paste after left in the fridge for weeks. Now the reflow works flawlessly. No gimmicks, just simple drops of solder paste and light tapping using flexible bamboo stick……and voila! The said to be difficult to solder E21A moved themselves to the correct position. The ultra thin solder mask do works. Perfect quadtrix board!

Many manufacturers gave up on E21A because it has very little to no self aligning behaviour during reflow. Thus requires very accurate PNP and solder dosing machine. But with very thin solder mask the self align effect restored back, even at very little solder paste applied. Thinner solder bond line, easier manufacturing, at the expense of fragile solder mask.

I use Sn96,5Ag3Cu05 (max. 260C @30seconds), with Sn63Pb37 reflow should be far easier.

I still have few hours before my flight, will post later.

- Clemence

OP updated with the latest test results of E21A

- Clemence

I’m still wondering how in parallel config those E21As performed significantly far more efficient. I’ve never seen anything like this before. Tested some XPGs 219Cs in single and quad parallel, but the difference between the parallel vs single config were never as high as E21A.
In parallel, the current divided into 4 lower current paths hence the boost in efficiency. But look at the chart below. It’s not lumen, just measured lux from my lux meter. At the same power level (near 3A), the output already doubled. My guess is E21A is super inefficient above 2A, while in low current its the other way around. To the eye 4p was indeed much brighter than the 1s config (at the same power).
The lower voltage also a good thing, most likely due to the lower resistance not only due to the hotter 4 cramped LEDs. The voltage reduced from as low as 100mA where heat is unlikely to affect the output.
Any idea?

- Clemence

I will have to read it all a bit closer (at work now) but that looks like a ton of good work! (it is not often that development work for new products is just published like this :slight_smile: ) And the boards work very well, that should make them, regardless of the cost, at least worthwhile for niche aplications where performance is critical (flashlights! :smiley: ), and who knows how cheap you can get them eventually. And these boards probably makes Nichia happy as well :slight_smile:

Hi All,

I’m still waiting for the other testers to confirm about VR16SP4. This far no single complaints about the VR16S1. Looks like the VR16S1 prototype leftover is ready for sale. This board is ready to use for 144Ax, 233x, 119x. The next upgrade would be smaller pads gap to better accommodate E17A and E21A (after some scrap-mod). And perhaps better guide markings for any LED other than 144Ax. I haven’t measured the output with 233A/B Nichia UV LED, but judging from the 144A results, I’m more than sure VR16S1 is the best non DTP UV board available.

I see no strong reasons to use E17 and E21A in a single setup on VR16S1 other than they have very wide high CRI selection available. But hey, you can solder two E21As or three E17As on VR16S1 to get oval beam (2p or 3p config). If you interested in doing so then please remember to space the LEDs as far as you can to make the phosphors cooler. Unlike common flashlight practices, photon cross talk is a big problem here. Anything escaped from the side will heat the neighboring area. This problem is what actually limits E series from high power single LED use. I will explain this phenomenon later in the OP.

The VR16SP4 should work excellent with E17A. Paired with quadtrix E21A it’s limited to only 20 watt before the phosphors start to burn.

VERY limited quantity working prototypes available for sale. PM me if you’re interested

- Clemence

It is interesting that 4p config starts to drop at just 1.2amps per led. The heat is still really limiting these leds. So in 4p it only gains the ability of 4 more watts but the efficiency is significantly better and results in much higher outputs regardless of the heat sharing.

Tested in 2s2p (6V) last night. Peaked at around 22 watt @3,8A. The lumen output was very close to 144AM up to 3,8A, only ~ 8% less at peak current. Not bad for such a domeless LED. I expect even higher efficiency and more total power in 4s config, would probably beat 144AM/AR. This is a strange breed of Nichia

- Clemence

Strange and interesting indeed :slight_smile:

Now we just need the rest of the tint spectrum in E21A R9080. I’ll be picking up some of these boards and the 3 E21A tints currently available once everything hits your shop.

OP’s chart updated with 2s2p test result for VR16SP4

I’m sure I’m not the only one curious about how this board performed compared to copper DTP. Am I correct?
Check the OP :wink:

- Clemence

I was, at least.
And they hold up very well! There’s no real need for a DTP board with these. If this technology is getting a bit cheaper these type of boards can be a solution (for flashlight use) for leds without neutral thermal pads.

Well…some people in HQ “forced” me to do it. Now, I’m thankful to them. These are not cheap at the moment, but it will be.
I’m thinking single sided driver boards can benefits from this technology.

- Clemence

Mini review of the VR16SP4 and E21A LEDs.

The quality of the board is very good, nice thick copper tracks and well setup to be very versatile. Also has a nice thin reflective white solder mask, which allows a thin solder layer and reflects most light which hits it. Assembly with E21A is fairly difficult though, and is not for the faint of heart.

My first attempt. I put too much paste on the top left which resulted in a small solder ball popping up in the middle of the 4 during reflow. Thankfully it came right off. Second picture is after a cleaning with IPA and a cotton swab to remove any flux residue. All emitters functional! Setup is 2S2P for 6V type.

After running at 2.8A for about an hour on the worst heatsink ever. Can see some slight browning of the phosphor. Still running great though. I put them on a much better heatsink at 2.1A today for about an hour. I don’t have a photo but the brown seems to have actually gone away somewhat.

Seriously, worst heatsink ever, random piece of leftover phosphor bronze sheet. That colored spot is the back of where the MCPCB sat. To be fair I only intended to run it for a few minutes but got sidetracked. The new heatsink setup is MUCH better. Thankfully the victims of my crime were just R70 emitters Clemence gave me to practice with.

My second attempt went much better, almost perfect alignment and no real cleanup needed. Still gave it a bit of a wipe with IPA just to be safe. All emitters light again. As is I really like these boards and plan to buy more of them. The changes for the production version should be beneficial for assembly and emitter longevity.

Once the production versions come into stock I will begin to experiment with tint mixing using these boards. I have a light I plan to build with 50mm quad optics using 4 quad E21A boards, so 16 emitters in total, and as many tints as I can get by the time I build it.

I will do a review for VR16S1 tomorrow. A little note for both boards though. These boards rely on thick copper with as much surface area as possible being on top of the nano ceramic base layer. This gives them the great thermal performance they have. However, this thermal transfer also means soldering leads is very difficult. I would recommend a nice powerful iron, especially if you want to attempt to solder it in a host because the board really sucks the heat away from the pads. Either that or pre-heat the whole thing in an oven to near reflow temps.

Thanks for the review Kyle. Actually, in this case, the copper spread has little effect to cool the LED. The low thermal resistance of the dielectric dominates most of the thermal transfer via through thickness direction. Lateral heat spreading only plays significant role up to 1,75 mm radius from the heat source point. This means the same board performance can be achieved using board size as small as only 3,25 mm larger than the LED used.

Thick copper foil will, unlike traditional thoughts, increase through thickness thermal resistance and lower the board performance. Sounds like nonsense but the factory lab test proved this. To give you better perspective, normal high performance MCPCB use 0,4 - 11 W/MK, 100 - 200um (0,1 - 0,2mm) thick dielectric. This board use 7,5 W/MK, 15um (0,015mm) dielectric. At this scale level, additional 1 Oz. copper (35um or 0,035mm) is huge. Long time ago, Barkuti pointed us (here and in CPF) about nano ceramic made in China boards he found on AE. Most people especially in CPF, mocked him for using the nano ceramic words (while in fact it’s the wording used by the manufacturer no him). He insisted on the potential performance of the boards, and I remember many people shoo-ed him. Now, the same technology approach is what I’m using. :stuck_out_tongue:

The decision to use 2 Oz. Copper foil was to increase current carrying capacity while keeping the heat generated as low as possible. This is useless for VR16SP4 since the LED max current is only 3A - limited by the phosphor temperature. VR16S1 gains benefit from 2 Oz. Copper because the compatible LEDs that can be used on the board have much higher limit.
Thick copper foil also makes reflow harder. In most case some of the paste will fill the gap between LED’s pads. This is especially problematic with lead-free solder with worse wetting capability compared to leaded solder. Less wetting equal less molten solder adhesion and more chance to solder bridging.

The VR16SP4 rev.2 will have 0,4mm gaps between LEDS, 1 Oz. copper, and SMD (Solder Mask Defined) pads. Current prototype is using hybrid SMD and NSMD. Don’t worry about the increased gaps would creates ugly beam under tight optics. With current gapless design it’s ugly already, much like XHP50 beam profile. VR16SP4 is never designed for tight optics anyways.

- Clemence

I’ve just caught up with your testing here Clemence. Thanks for taking the time to do it. Its all good useful info. :+1: