I have a lot of projects I want to do, so I’m trying to learn how. I started with a really simple one — converting a Convoy S7 with XM-L T6 3C @ 2.1A to use a Nichia 219B (~4500K 92CRI). And of course I documented the process, in case it might help anyone else get started with modding.
The 3C tint was good, but Nichia 219 is better. And with a stainless steel host, 2.1A really was more heat than it could handle. So, I decided to reduce it to 1.4A and give it an even better tint.
Here’s the host and its new star: (click for larger version of any pic in this post)
To start with, I have a few tools… This is what I’ve got for soldering: A cheap Weller 40W iron from Home Depot, a cheap “helping hands” soldering station from Harbor Freight, some stereo magnifying glasses from Amazon, tweezers and Arctic Alumina from RMM, and a ZL H51w. Oh, and a flexible desk lamp. I generally put something on the table to protect it while I’m working, too, such as a flat sheet of metal or a ceramic tile.
FWIW, the desk I’ve been using is part of a treadmill desk, so I suppose you could say I like to solder while I’m on the treadmill. o_O
I’m a little new to soldering though, so first I decided to watch some howtos on youtube and try messing around on an old PCI card. I sacrificed a DEC Tulip for the cause.
To start with, I removed some little parts. Most of them either broke in the process or damaged the circuit board (or both):
After that, I got a little better…
And then I managed to remove some of the same parts without destroying anything, though I still scorched the board. I also removed and re-soldered some bigger parts, which seem to have turned out okay.
On the smallest parts, resistors I think, I mostly managed to get one half off without damage but the other half had its connector break off.
And here were some of the parts which took part of the board’s circuit trace with them. Oops:
So, once I felt ready, I went on to the real project. First step: removing the pill. This is really easy with the right tool.
With that done, here is what the pill looked like:
Next step: Remove two 7135 chips to get it down to 1.4 amps. Fortunately, there are two on the back of the pill which are easy to access. I kinda made a mess in the process, but I got them off. The first couple things I tried didn’t work; like, I couldn’t get solder wick in contact with the solder to pull it off. I ended up holding the sides of the 7135 chip with tweezers, lifting up slightly, and holding a wide soldering bit across the three small pins until the chip lifted off. Here was how it looked afterward:
I’m not sure if those 7135 chips are still functional or not, but I put them into my parts bin in case they’re useful later.
Next, I wanted to know if I had totally ruined the driver. So I put it all back together and checked — does it still work?
Yup, it still works.
Next step, take the old star off. Unsoldering the leads was easy enough…
Taking the star out was easy too, since it used thermal paste instead of thermal adhesive. Quite a bit of thermal paste, actually.
Next I tried to put the new star in… but the Noctigon was slightly too big.
The next hour or so was spent making the noctigon smaller. I don’t have a vice or any power tools, so I put leather into some pliers and gripped the star with the padded pliers and held it against a table. To remove material, I used a diamond-coated file I had laying around. It took a while.
I also discovered that the noctigon is thicker than the stock star. This didn’t cause any issues, but I took a pic of the relative thickness anyway:
Right, so after getting the noctigon resized, I put on some Arctic Alumina. I’m used to dealing with CPUs, so I started with a pretty thin layer and put it in to check for contact.
As it turns out, the star made almost no direct contact with the pill. I checked to make sure it wasn’t due to the noctigon still being too big, and the original star actually had just as little contact. I don’t have a lathe or any other tools to flatten the surfaces, so I decided to glob on the thermal paste instead. I know, I know, this isn’t a great idea. Direct contact beats thermal paste any day of the week, and thick thermal paste is especially bad. But I didn’t have a way to fix it, so I filled the gap with something better than air.
You can probably guess how much thermal paste was used based on the mess near the wires:
So, next step: tin the connection pads on the star. Pretty easy, even though I had never done it before.
… and connect the power wires. Again, fairly easy. I didn’t do the cleanest job of it though.
I also got a nifty little XP-G to XM-L reflector adapter, to help center the emitter and protect the star. It ended up being a bit thick though…
The emitter was nicely centered with the size adapter in, but it made the beam ringy. I ended up attaching the old XM-L centering ring to the reflector instead, using it only as an insulator. The beam ended up nicer, but the emitter is slightly off center:
The light still works nicely after the changes I made. Here are some ringy beam shots with the size-adapter centering ring installed:
… and the beam shots with the adapter removed. I like this much better:
After getting everything back together, I took some tailcap measurements. It wasn’t quite what I expected.
- Low: 0.073A (5.25%, ~46 hours on a 3400mAh cell)
- Medium: 0.69A (49.6%, ~5 hours)
- High: 1.39A (100%, 2.5 hours)
I thought the medium mode would be 30% or 40, not 50 Perhaps this will be a good excuse to take the thing apart again later to reflash the firmware. But not today. Today I’m happy just to have a beautiful stainless steel light with the latest high-CRI neutral emitter inside.
Next I intend to upgrade an old UltraFire HD2011 (better emitter, better driver) and increase the throw on a UF-838 (de-domed XP-G2, more heat sinking, better driver). I’m also converting a Blackshadow Terminator to be wide-spectrum (simple emitter swaps), and intend to find a way to flash its firmware. I may mess around with nanjg/attiny13 drivers first though, on something a little easier.
