New Luminus SST-20 (3535 size)

We may have to go with this new Luminus LED for the BLF Lantern project, as apparently the Samsung 351 series in 2700K & 5000K are hard to order for the manufacturer of the lantern, but have not seen or tested this SST-20 emitter yet.

Personally I think it would be a good option. Going with CRI above 9080 really gets quickly into diminishing returns in terms of light quality. Going above 9050 quickly reduces efficacy. SST-20 is already quite inefficient, but I don’t think it’s too inefficient for this lantern…and for a CRI 95 emitter it’s good. It’s quite a bit smaller than LH351D which negatively affects its performance at high currents, but this lantern won’t have high currents. It’s already quite rosy and tint mixing will make it even rosier, we can expect CRI to suffer because of that. But is that going to make the light more or less pleasant?

As this emitter is only available in the 2700-4000K CCT range, this should be less of a problem. Mixing tint with a 2700K and 4000K emitter should give desirable result (maybe a bit on the rosy side between 3000K and 3500K).

I think, Nichia E21A could be still considered for the lantern (but it would need different PCB), as it has CCTs down to 2200K and 2000K with R9080 CRI (these are designated as R9050, but maukka measured R9 above 95 for the 2000K variant). Nichia E21A could be a good candidate for tri-channel mixing (e.g. 2000K, 3000K, 4500K), but I know, that it is only a wish of mine :slight_smile:

I fear that the E21A would give handling problems when reflowed by flashlight manufacturers, it is a delicate operation and I’m pretty sure they have no pick-n-place machines at Sofirn.

I’m all for the SST-20-W led, at these drive currents it is not that much less efficient than the LH351D. I think though that a few people regret that the colour temperature will not go up to 5000K anymore. Btw, I had understood from Barry (from a different discussion) that Sofirn is able to source the high CRI SST-20 leds.

Yes, the bare phosphor of the E21A seems to be pretty sensitive, so I would not try it myself as well.

Good news :slight_smile: I had a talk with Sofirn a few months earlier about how high CRI can be good.
So I hope, if they can source those LEDs, they will :slight_smile:

LH351D diffuse really well as it is. Even though SST-20 may work it seems like a compromise in every regard. Efficiency, CCT, and beam shape. Makes you wonder what is truly going on with Lumintop and Sofirn with some of these projects…

Yeah, I was really wanting the lantern to go a little higher than 4000K for sure. However, since it’s a lantern and not a flashlight, I think 4000K would still be acceptable.

I do hope they can source 5000K LEDs too.

If SST-20 is considered, 5000K is a very big step back (available only in ~65 CRI instead of ~95 CRI). If only SST-20 is used, this should be kept in mind.

Has anyone made more progress with throw numbers before/after slicing? I’m curious whether it’s worth the risk, or if these things are “good enough” with the tiny little dome on. I’d love to have some options that came even remotely close to the Oslon White Flat for throw with warmer CCT and better CRI.

@coherent, you’re asking about the sst-20 in general or in this particular light? I just happen to have pre & post slice throw numbers from an single led / reflectored light I built with an sst-20 two days ago if you’re just looking for general info.

Dome on the throw was 17,500kcd, after the slice and refocusing the little light was doing 33,200kcd. Tint and lumens both dropped ever so slightly.

Cereal_killer, I was curious about the SST20 in general, so your numbers are perfect. Nearly double the Cd and 40% more throw, sounds like I need to buy a few extras to try slicing. Do you have any advice for getting a good, clean result?

Wow, that’s quite a difference for an LED that’s already pretty throwy to begin with. How did you do the slice?

where to buy good SST20 4000k or 5000k? I can only find ones above bbl

Sorry just saw this, I did the washer method. Slice with a fresh new blade that you scrub with alcohol before using.
To be honest, had I spent the time focusing the emitter pre-slice that I did focusing it post-slice the difference would be less dramatic. I probably could of got the domed emitter focused a good bit better, just an amature best guess here but I’d say, if I was as meticulous with it as I was the sliced one I probably could of gotten it so the difference was <25% increase instead of nearly 50%.

I only spent 10min focusing it the first time around, only a few tries sanding the disk. I had set a goal of 40kcd for the light and I could see I needed something dramatic so I did the slice before spending much time on it.

Oh I forgot, I came back to this thread to ask what my options were for 5000k SST-20’s currently for sale in the highest output bin (the base 70cri version will be fine for this one)?

The highest bin available to consumers now is the L3 Flux Bin SST-20 5000k:
http://kaidomain.com/Luminus-SST-20-L3-DA-Neutral-White-5000K-LED-Emitter

I have a 3500K SST-20 from BlueSwordM/Mouser in my Emisar D1S right now. Using a GA (being conservative) it makes about 680 lumens and 595m throw vs. 1020 lumens and 665m with the stock 5D XP-L HI. Tint is acceptable and everything on the trail looks much, much nicer thanks to the dramatically better CRI (especially R9).

Today I tried a couple of slice dedomes. The first one went fine, but the second I shaved a little close and clipped a bond wire, which melted the moment I turned it on. With the “good” dedome, output is 30% lower (470 lumens) and I gained a grand total of 5m throw. The hotspot is much smaller and there is less corona, and subjectively the tint is a lot rosier. It’s possible I would gain more throw by playing around with the focus, but I don’t have the material to do it here right now.

The effectiveness of slicing is sensitive to the thickness of silicone remaining over the die. If it’s not very thin light can reflect off the silicone/air interface and bounce down to the side of the die where it’s more likely to be lost. This reduces the output and reduces the throw gain. If the silicone is very thin this reflecting light will bounce back to the die where it increases the die luminance.

With some of these LEDs the bond wires are tall which makes it harder to get a close cut.

A theory: the reason that by slicing a warm/high CRI led you loose so much output (20+% compared to 10% for cool white)is that there is much less blue light to be recycled to turn into more visible green, and at the same time phosfors are present that recycle green into less visible red.
I have not fully thought this over, but I have seen with the spectrometer that when a high CRI led is illuminated with 530nm green light, apart from scattered green you can measure some red coming off the led.