Why isn't this used more? It looks like it would fit on a MT-G2 Noctigon, so we know a direct copper bond is possible. Minimum forward voltage is about the same as an XM-L2, so we have lots of driver and host possibilities. The dome is about the same size as an MT-G2, which makes finding suitable reflectors a bit tougher, but it should be possible to use Carclo optics for the MT-G2. I only see two issues. It's not very efficient and it's expensive.
The two issues you mentioned are indeed the ones to contend with. And the idea that an XM-L2 U2 1A can produce nearly 2,000 lumens with the right tweaking makes the SST-90 (producing maybe 2,100 lumens max?) not so hot anymore. Die is larger with it and so you’ll see less throw, and you need half a Chevy engine block for the heat it produces, plus the ridiculous current needed…nope, it’s just not a logical choice anymore.
That's a good point Leaftye. On copper it sounds like one could go for 15 amps if they so choose. Is anyone aware of any SST-90 output on copper graphs or data?
Because the MT-G2 is twice as efficient and probably costs less, unless one has bought a X6 and swapped it out already in which case its free. does the SST throw better for a given reflector size? No. It’s inferior in just about every way to the MT-G2.
MT-G2 requires higher voltage, which means it's a non starter for single cell high current applications.
It turns out the SST-90 datasheet is very old. It only shows the N3 bin. There's a P bin that Eagletac already uses that can put out up to 3240 lumens at 9A. There may also be a Q bin that does 3915 lumens at 9A. Imagine if you could drive a SRK triple at 9A with Q bin SST-90's.
nope, that you can not do.
I don't think anyone has done measurements on copper (the led is too expensive for me to do it)
The thermal resistance is very low, that is good: 0.64 degC/W compared to 2.5 degC/W for the XM-L2 (the size of the chip makes this possible). But the big surface area of the thermal solder pad makes a fairly good heat path already on a board with dielectric layer, so who knows what the gains are on copper?
Who is going to try 20A ?
Why do you say that when the pads line up well enough not to short?
Everything is so big on this chip that you can file a chunk of copper off to create a thermal pad pillar and just solder wires to the elctrical connections of the led.
That's possible. I did check out the datasheet though, and it's close enough that there's no need to. It looks like there's no need for a Noctigon since the Comex board already has a direct copper bond. Thanks to pinkpanda3310 for checking.
[quote=pinkpanda3310] As promised, I checked out the star and it seems direct to copper. I didn't grind it with the dremel as I wanted to reuse the star again.
I guess the 0.1 could be a margin of error due to my fingers??? In this next pic is the MT-G2 top and sst-90 bottom
[/quote]
Now I need to find a newer bin, see if LED Supply will sell the MT-G2 optics, and see how far a FET SRK driver can be pushed....and cough up a big chunk of cash. Still though, it'd be a hell of a bargain if a SRK/M6 can push out 10,000 lumens for a little over $100 in parts.
there's not many descriptions on seller's pages that you can blindly trust. Especially if you drive the led way out of specifications there's only one way: buy one and measure ('meet het en je weet het' :-) )
That's the goal, and many of us are already doing that. Remember back when the XM-L/XM-L2 could only be driven to 3A? Fortunately some people are willing to push artificial limitations.
I got an email back from Luminus. They don't list the newer bins on the datasheet because they can't support them, which seems to mean that they lack the production capacity to make them in sufficient quantities in a timely manner. I did mention the Q bin though and it wasn't denied that it really exists. That Q bin has nearly 50% more efficient and maximum output. That puts it at very nearly the same level as the MT-G2 you crash tested. The MT-G2 is listed to have a maximum current of 3A, yet you pushed it to 12A before output started to drop. The SST-90 has a 9A maximum on its datasheet. Imagine if you could push it just as hard with the same kind of results...to 36A and ~8000 lumens. It'd heat up any host very quickly, but what a glorious few seconds that would be.
I should have looked at that other thread more closely. The P bin is available, so that gets closer to MT-G2 efficiency and output, but at XM-L2 voltages. Q bin looks like it'd perform just like the MT-G2, but at the voltage of a XM-L2.
http://kaidomain.com/product/Details.S022075 (6500K)
That's sold out, but this one is still available.
http://kaidomain.com/product/details.S022180 (6500K)
It's expensive, but the N bin bare emitter (4500K) costs half that price and copper mcpcb's cost less than $2 each. Hopefully bare Q bin emitters are available and at the same price.
i’ve heard many people say “it possible, to put sst90 on mtg star,……” but i have not seen anyone actually doing it.
looks like it should, in theory, is not definate will, in reality. seen it too many times.
It is doable. Old-Lumens has done it. I’ve seen the SST-90 being sold on MT-G stars. I have 2 Noctigons on order that I will be flowing my SST-90’s to (for now!)
show us when you’re done. i do not mind to be proven wrong when i am.
This post is mostly for myself.
Suitable part numbers may be:
SST-90-WCLS-F11-QN3
SST-90-WCLS-F11-QN4
SST-90-WCLS-F11-QP3
SST-90-WCLS-F11-QP4
SST-90-WCLS-F11-QQ3
SST-90-WCLS-F11-QQ4
SST-90-WCLS-F11-QR3
SST-90-WCLS-F11-QR4
SST-90-WCLS-F11-Q3N3
SST-90-WCLS-F11-Q3N4
SST-90-WCLS-F11-Q3P3
SST-90-WCLS-F11-Q3P4
SST-90-WCLS-F11-Q3Q3
SST-90-WCLS-F11-Q3Q4
SST-90-WCLS-F11-Q3R3
SST-90-WCLS-F11-Q3R4
SST-90-WCLS-F11-Q4N3
SST-90-WCLS-F11-Q4N4
SST-90-WCLS-F11-Q4P3
SST-90-WCLS-F11-Q4P4
SST-90-WCLS-F11-Q4Q3
SST-90-WCLS-F11-Q4Q4
SST-90-WCLS-F11-Q4R3
SST-90-WCLS-F11-Q4R4
WCLS is color white, 4500K & 4000K, standard CRI (typ. 70).
Q/Q3/Q4 are possible flux bins. 3240-3915 lumens.
N3/N4/P3/P4 are chromaticity bins at 4500K.
Q3/Q4/R3/R4 are chromaticity bins at 4500K.
…now you’re gonna make me get one and tie it up with an FET driver! Curse you Perry the Platypus!
Edit: See? Kaidomain had the P bin Q4 tint so I have one coming. I’ll wedge it into my C8 with it’s huge copper pill and FET driver and Tofty 10A+ switch and we shall see…
That's going to be a monster Dale. That should give you the brightest single cell, single emitter light out there. By well over 1000 lumens if you can drive it hard enough. Hopefully we can find some Q bin emitters to increase that potential by another 500 lumens.