I love a good tint. The Nichia 219 gets very accurate color rendition, and most of my favorite lights use it. But I wanted something even better… so I built a wide-spectrum light. The theory is that, while high-CRI shows accurate color, wide-spectrum should show vivid color. Here’s the host I used for this experiment:
I had already altered it slightly, adding DC-Fix (sand) to the lenses to make the beam smoother and floodier. If you mod the emitters though, I recommend doing that first and adding DC-Fix later so you can see whether the emitters are centered in the reflector. It’s hard to get the DC-Fix off since the lenses are recessed. So, these are stuck on unless I find some serious motivation to change it:
BTW, if you want to add diffuser film to your BST, a UK 2-pence coin is almost exactly the right size to use as a stencil. Just peel the protective paper off the DC-Fix, drop the coin on the sticky side of the film, and cut around the edges with a razor blade. Peel it off the coin, and it should drop right in to the lens. If you don’t have a UK 2-pence coin, a 2 Euro coin is almost as good. Same size, but the edge isn’t smooth.
It doesn’t have to be exact — leaving half a blade-width of extra material around the edges will actually make it fit tighter, since the 2-pence coin is part of a millimeter smaller than the inside of the bezel. A small gap between the film and the bezel also shouldn’t matter, due to the shape of the bezel.
Of course, if you go far enough to remove the bezel from the head, you can just use the lens itself as a cutting guide. But I hear the bezel is rather difficult to remove.
With four emitters, I can mix four different tints. For maximum effect, I want the widest tint range I can get. I didn’t get quite as wide a variety as I had hoped, but these were easy and convenient to get on copper:
(XM-L2 U2 1A, XM-L2 T6 3C, XM-L2 T4 5B1, XM-L2 T4 7B3)
Under UV light the emitters glow different colors. I couldn’t really capture the full effect on a camera, but here’s a rough idea how it looks:
The first step is to take the heads off. This was actually pretty easy; I just had to grab tightly with my fingers and twist. The first one was the hardest, since the other heads were in the way. No tools required though.
The second head had some weird gunk in its threads. It wasn’t particularly easy to clean out, but I eventually did so. I’m not sure what it was; possibly dried-out thermal paste, maybe.
With all four heads off, I noticed I could see a blue glint inside. I think it was from the anti-reflective lens coating.
After removing all four heads, it looks like this. I think the emitters are XM-L2 T6 2A or something along those lines. Definitely XM-L2, but I’m not sure about the tint or bin. It looks slightly warmer than my XM-L U2 1A, cooler than a 3C, and similar to a XP-G2 2B.
Right, so now to replace the stars. Pretty easy to do, actually.
Looking at my first target, I noticed that the wires inside are actually pretty thin. I’m not planning to replace the wires right now (not going for maximum lumens), but I might do it later. Here they are, compared to some 22AWG wires:
Unsoldering the old star is a simple matter of gently pulling up on the wire with tweezers while quickly touching an iron to the wire and blob of solder.
The old star had a mess of thermal goop on it. I cleaned off each star and each spot for a star in the host. A tissue removed most of it, though I had to use tweezers to get into the corners. Isopropyl alcohol cleaned off the rest to make a clean surface for the new star.
After cleaning, I could see that the host had a rough edge which could raise up the star and interfere with thermal contact. Click to see a zoomed version with much more detail.
So, I filed off the rough edge with one of my diamond-coated files. It didn’t take long, though I had to be careful not to damage any wires or other electronics underneath, and get rid of the aluminum dust afterward so it wouldn’t short the driver.
With that edge filed, it’s time to put in the first new emitter. I decided to start with the warmest one, then work my way back toward cooler tints.
Just as a quick sanity check, I tried fitting the new emitter. There’s plenty of room for it, and though it’s the same size as the old star I wonder if I might have been able to fit a bigger star.
To get the star ready, it needs its contact pads pre-tinned. Here’s my view during that process. It looks a lot bigger in person, especially since I’m looking through some stereo magnifying glasses:
A little messy on one side, but I got the first star pre-tinned. At first I had trouble getting the solder to stick to the pad (the iron itself kept sucking up solder), but it appeared to leave behind a clear residue that I assume was flux. Then I touched the iron to the contact pad and there was a “pop” sound and the solder all jumped onto the pad. Yay flux! Anyway, the other pad went more easily. I hope I didn’t make the blob too thin on one side.
Next I put some thermal paste on the star, put the star in place, squished it around for a while to spread and thin the thermal paste, and soldered it. And squished it around some more for good measure. The pad which didn’t have much solder was still pretty thin, but it seems to have good enough contact.
Just to be sure, I turned it on to make sure it still works. My, that’s a warm tint…
Oh, and don’t forget to put the centering ring back on. I almost forgot, all four times. This pic also showcases my noob soldering skills; neither connection is terribly clean.
That’s it for the first star. The others are mostly the same, but with slight variations. The second star, for example, took a big blob of solder along with the wire, so I had to clean it off.
Solder wick to the rescue! It was pretty easy to just wrap wick around the blob, hold it with tweezers, and heat it up.
The new emitter this time is a 5B1 tint. I put the two warmest tints diagonal from each other, in hopes it will make the colors more consistent up close.
… and after going through the same steps as before (including filing off the rough edges), star #2 is done. Pretty quick and easy.
This mod is now half done:
At this point, the doorbell rang and I had to play hostess for a while, so I put the heads back on for a bit of testing. Even half-done, it gave out quite a nice quality of light, with more vivid color rendition than I’ve seen from any single Cree tint. My camera refused to focus due to the BST’s slow PWM, but here’s what it looked like with two emitters swapped:
Several hours later, I was able to continue. Star three is up next, with a 3C tint.
Here’s a close-up of the old star with the thermal paste still on. It should provide a rough idea how much thermal contact the star actually gets, and where.
And, as usual, this hole had a rough edge which needed to be filed to keep it from causing a gap between star and host. I think this pic shows the roughness pretty well.
… and a couple views of the edge after I smoothed it out with a hand file:
Here it is all soldered in. The black lead was rather short though, and was a bit of a pain to get in contact with its solder blob.
The last emitter is also the coolest — 1A tint.
On this particular star, the wire leads were barely stripped. I had a lot of trouble getting them to make contact with the new star, so I ended up stripping them a bit more, very carefully, with all the parts already in place. This is how they looked before:
All four star plates had rough edges. Here’s the last one:
Some of the noctigons had rough edges too, which I filed off before putting them into the host:
After doing this a few times, I think I got better at tinning the pads. This last one came out pretty clean:
Woot, all soldering is done! The stars sure don’t like to stay centered though; this host design lets them slide around quite a bit. I had to individually re-center each one before putting its head on.
I attempted to show the emitters under UV light again, but it just doesn’t look the same on camera. It would help if the stars were closer together or I had a wider beam on my UV light.
At this point, I was ready for some lube. The reflector heads were pretty dry and I don’t have extra O-rings of this size, so I fixed them. Hopefully it’ll keep the O-rings in good condition for a long time.
With everything assembled, I got a shot of how the four tints look next to each other. The handle is at the bottom of this pic, so the most extreme tints are opposite the handle. I tried to make it so that each set of two adjacent heads would have as close to a balanced tint as possible.
Next, I tried to compare the beam against my new Convoy S7-219. It’s not a fair comparison though, since the S7’s highest mode is still lower than the BST’s lowest mode. Here’s a short exposure, with S7-219 on the left and BST-wide on the right:
… and a longer exposure. The BST-wide looks warmer in the picture, but in person I have trouble determining if it’s warmer, cooler, or the same. I think it’s actually both warmer and cooler. But it’s hard to compare since it could possibly just look better due to being brighter. It’s like how music sounds better with the volume turned up.
I also tried to get pics of a colorful subject under various lighting, but about all I could determine for sure is that my camera is far too forgiving. Even with the white balance stuck on one setting, it makes most of my lights look nearly the same. On camera, icky tints look far better in photos than they actually look in person, and really nice tints don’t look quite as nice as they should.
Here’s the BST-wide:
… and the Convoy S7-219B:
And for reference, the same thing under normal incandescent overhead lights (same white balance as other shots). Yuck:
My eyes might be weird, but this is how everything looks to me under incandescent light. I can’t stand how yellow everything is.
In any case, I really like the result. Walking around the house last night with my BST-wide, I was amazed at how colorful everything was. I kept thinking, “The colors! The colors! Soooo many colors!” At the very least, it does make colors look rather vivid. I’m not totally sure whether it’s better than my Nichia NVSL219BT 4500K 92CRI light, but my subjective perception tells me that the colors “pop” more with the wide-spectrum light than they do with a high-CRI light.
This BST-wide has some downsides, though. First, it casts multiple rainbow shadows. It also looks weird when it’s really close to its target, because the beams haven’t gone far enough to fully merge yet. It still has pretty slow PWM and limited output modes (which I’ll fix later), and it’s big and heavy. I also think it might have been a good idea to spread the tints out even further… like 0/2/6/8 instead of 1/3/5/7.
The upsides are that it has the most vivid color rendition of any light I’ve ever tried, the diffused beam is wide and silky smooth, it has a tripod mount for use as a work light, it’s easily bright enough for anything I need, and it’s relatively ergonomic for its size so it’s easy to use. And for fun I can cover up one or more heads to see how different tints look… All four tints individually look pretty terrible compared to their combined output.
My next steps for this light will probably not happen for a while, because I need to learn and practice more first. The plan is to find a way to give it new firmware (even if that means a whole new driver), give it better modes and faster PWM, and give it an adjustable true-strobe mode for stop-motion fun. But even without those, it’s already a very useful flooder for any tasks where I want to see details and vivid colors.
—
Update: new beam shots, showing how the tints blend at various distances.
This is how it looks when it’s resting directly on a flat surface:
It’s a rainbow beam!
If I’m standing and holding the light in my hand, the tints have already fully merged by the time they reach the ground. There are no weird colors anywhere in the beam. Even just a short distance from the ground, the beam is reasonably consistent, though it seems to take more distance to blend at at around 45 degrees… These show the light resting on top of a tissue box:
