mod: UF HD2011 -> Nichia 219B + NLITE

I decided to improve my UltraFire HD2011. The goals were better tint, higher CRI, better modes and mode spacing, no visible PWM, and a better interface.

Unlike my previous two mods, this will be a much shorter post. I simply wanted to get the mod done and didn’t take any pictures.

Before the mod:

A couple holiday seasons ago, I tried out Wallbuys for the first time. They were new and doing some good holiday sales. I picked up a few lights, one of which was an UltraFire HD2011. It’s a lot like a cheaper version of a Convoy S2, but this was about a month or two before the Convoy lights were available. I think this was only $8, which was pretty cheap for a half-decent 1x18650 cigar/tube light.

As it was sold, the HD2011 had a really cool white (blueish) XM-L, probably around 7500K tint, with slow PWM and poor mode spacing. I measured it with my only-sort-of-calibrated light box today before starting the mod, and I got the following: (BTW, the modes go in this order)

  • High (100%): 612 lm
  • Med (~59%): 358 lm
  • Low (~18%): 109 lm
  • Strobe: ??? lm
  • SOS: ??? lm

The default driver had mode memory, so it required going through strobe and SOS to get from low to high. However, at least it didn’t have *next* mode memory. I could leave it on low and have it always start there if desired. The beam was nicely floody though, thanks to a really shallow OP reflector. And if desired, the bezel could be removed to run it in mule mode with a full 180 degree beam.

Mods / After:

The mods I did were:

  • Replace the driver with a Qlite, 4x7135 380mA (1.52A total), with standard NLITE firmware.
  • Solder driver option stars 2 and 4, to give it 4-modes with no mode memory.
  • Replace the 0* tint XM-L with a Nichia 219B 4500K 92CRI emitter.

After finishing, I measured the following levels:

  • Moon (~1%): 3.34 lm … Not really a proper “moon” level, but still useful.
  • Low (5%): 17.9 lm
  • Med (30%): 134 lm
  • High (100%): 403 lm

I did both sets of measurements with a protected Panasonic 3100mAh NCR18650A, so it probably wasn’t the absolute highest output it can produce. I only had the battery about 90% full, too.

Notes:

The HD2011 host:

In this particular host, there cannot be any 7135 chips on the battery side of the driver, because they would prevent the driver retaining ring from touching the negative contact area.

Another thing about this host is that its MCPCB shelf isn’t flat. It dips down in the center. So, I ended up using a lot of Arctic Alumina and I’m glad I didn’t attempt to drive the emitter any harder. Also, the stock light has the MCPCB glued into place with thermal adhesive, so it was a bit of a pain getting that and all the glue off.

After finishing this mod, the front-most part of the head doesn’t want to screw down all the way. There’s just enough of a gap for my fingernail. I’m not sure why, but I suspect it may simply be the extra thickness of the Noctigon compared to the original MCPCB. So, I should probably replace the missing O-ring in the host…

The original emitter isolation/centering sheet (butterfly style) proved useful for the new, smaller emitter. It’s just rotated 45 degrees off from how it would normally be, which keeps the reflector from shorting anything and helps keep the emitter centered (only the corners of the emitter touch the centering ring).

This host does not tail-stand well. The switch cap sticks out too far and makes it wobbly.

Soldering:

Soldering the option stars to the outer ring is hard! I’m probably missing something obvious, but I made rather a mess and had to clean it up with solder wick then start over. It still is pretty messy, but at least I got it to work. Anyone got any tips on how to do this?

Beam:

The beam has a significantly smaller hotspot now, but that’s to be expected when putting in an emitter half the size of the original. This also means it probably has higher lux than the original, though I didn’t measure that first so I’m not sure. (2/3rds the lumens, 1/2 the emitter size, so maybe 4/3rds the original lux?) I’m considering adding DC-Fix to make the hotspot wide enough for biking.

Tint:

This light now has the coolest tint I’ve ever seen on a Nichia 219. I bought two identical emitters from RMM, but in this host it looks a lot less warm than in my Convoy S7-219B. And quite a bit cooler than a L3 L10-219. I don’t have a way to measure tint, but visually it looks like maybe 4400K for the L3 L10-219, 4800K for the S7-219B, and 5100K for the HD2011-219B. It’s still a nice, pleasant white, but it’s colder than I expected. Also brighter, considering my S7-219B only gets about 270 lm on high and this one gets 400 lm.

The cooler tint might be partly due to being driven at 1.52A (the S7-219B is driven at 1.4A, and the L3 L10-219 is much much lower), and might be due to using only a reflector instead of having DC-Fix smoothing out the tint variance across the beam, and it’s possible this batch might be 5000K tint emitters instead of 4500K. Or it could just be the normal variation within a single tint bin.

NLITE firmware / Interface:

The NLITE “moon” mode isn’t really a true moon mode. It’s about 3 lumens, with this setup. I consider moon to be under a lumen. However, 3 lumens is still a very useful level.

On-time memory with no mode memory is a little weird at first. It’s nice having it start in the lowest level each time, but if you want to bump up from “low” to “medium”, it means half-pressing the switch three times (moon, low, med). However, this becomes pretty intuitive after a while, and I like it a lot better than on-time memory with memory.

The interface is simple. No matter where the light is set now, tap (half-press) the switch a consistent number of times to reach a new mode:

  • For mode 1 (3lm), tap once.
  • For mode 2 (18lm), tap twice.
  • For mode 3 (134lm), tap 3 times.
  • For mode 4 (403lm), tap 4 times.

And if the light was off when you started, make the first tap a full click instead.

Final thoughts:

It’s still day time, so… I haven’t had time to really use this yet and see how I like it. More thoughts later, perhaps. It’s definitely a lot nicer than the original light was, at least.


Update: Pictures, etc

Although I didn’t take pictures during the mod, I did take some pictures afterward… and made a couple more changes too.

Here is our lovely host, the cheap tube light:

On the other side, it proudly advertises that it has a Cree XM-L T6 inside. Heh. We’ll fix that. :slight_smile:


(actually, I’m not sure how to remove the printed label without damaging the ano… but it’s what’s inside that counts, right?)

Installing a new emitter star was really easy. It dropped right in. Here it is after fixing everything in place:

One weird thing was the process of trying to get the centering ring / spacer installed. It turns out that the spacer was most useful when I rotated it 45 degrees or so. This not only made it easier to make sure things were centered (since the edges were closer to the corners of the emitter), but it also nestled up against the solder beads to prevent it from turning while I tightened the reflector.

And about that reflector, this is what it looks like with the front end assembled. The diagonal centering ring looks a little odd, but it works well enough.

On the other side of the pill, the driver retaining ring really got in the way. Even after removing all the 7135 chips on this side, the driver still barely fit. I was worried the ring would short something, but fortunately it all works fine. I did have to use solder wick to clean up any extra solder near the outer edge of the board though, to make things fit properly. I think the solder on the two stars I bridged might actually be the only thing keeping the retaining ring from shorting the two other stars.

Additional tweaks

Now, everything was working, but it still had a couple issues. First, it doesn’t tailstand well since the button sticks out. It looks more like The Leaning Torch of Pisa:

I tried to remove a bit of thickness from the nub inside the rubber switch cover… but it didn’t help. The button is simply rounded; the switch isn’t keeping it pressed out. So, now I have a couple millimeters of dead space before the switch cover touches the switch itself, and then a couple more millimeters between there and when the switch clicks. I’ll probably look for a different switch boot in case I can find a flat or thinner one.

Update: Later in the day, CRX gave me a solution. I managed to find a suitable O-ring, stolen from an ancient old incandescent light I had laying around. It’s labelled “Super MityLite ™ LMX” and it came with a pocket knife. I received two sets for free a long time ago; I forget why. The specs on the light are hideous, but then… it is a model from all the way back in 1980:

  • Power source: 2xAAA alkaline
  • Lumens: 7
  • Watts: 1.11
  • Runtime per charge: 2 hours
  • Bulb lifetime: 20 hours

And, of course, the donor light doesn’t even work.

So, back to this century, the switch boot no longer interferes with tail-standing after doing this to it:

Also, I glued back on the cut-off bit of “nipple” from the inside of the button, to reduce the dead zone before it contacts the actual switch.

This boot mod did affect the battery tolerance, unfortunately. My extra-long protected Panasonic cells no longer work. They technically fit, but I think something must be shorting or not connecting right because it’s too long. However, some slightly shorter protected cells work, and all my unprotected cells work too.

Beam

The next thing which needed some improvement was the beam. Honestly, it was a nice beam pattern, but it was throwier than I wanted and it seemed like the hotspot and spill were slightly different tints. Not great on a high-CRI light. So, I added some DC-Fix.

I tried to get the lens out, but it’s sitting on a shelf in the head and is held in by an extremely tight retaining ring. I tried and tried to unscrew the ring, but couldn’t get it to budge… and I scratched up the anodizing in the process. So, I resorted to just sticking something onto the exposed part of the lens.

The first thing I tried as a cutting template was the bezel of my T10T, since it seemed about right. But it ended up just a little too small, and the resulting beam had some really annoying artifacts around the edges where the light escaped untouched at the edge of the lens.

Then I found something better. For this light, the best cutting template was a Canadian dime. It was almost perfect:

No more beam artifacts. And a close-up shot shows the fit was indeed very, very close:

Here is a view with the light turned on:

Now the beam is much smoother and floodier, so it’s nicer up close and can be used for biking. It also fixed the tint difference between spot and spill, and the tint looks slightly warmer overall (maybe 50K or 100K warmer) — closer to my other Nichia 219B light, but still colder.

Output

I re-measured the lumens after adding DC-Fix:

  • Moon (~1): 3.26 lm (2.4 loss)
  • Low (5): 17.1 lm (4.4 loss)
  • Med (30): 124 lm (7 loss)
  • High (100): 379 lm (6 loss)

This was not a terribly well-controlled, careful measurement. I think it’s accurate enough to show that the total lumen output decreased by less than what a human can perceive though. Any perceived difference in brightness is due to the shape of the beam, not the total amount of light coming out.

In any case, I’m pretty happy with the results. Now I just need to find a flatter/thinner button, replace a missing O-ring, and maybe drill holes so I can add a clip.

I fixed the tail-standing issue, added missing O-rings, and lubed up all the threads and O-rings. Now all it needs is a clip.

To bridge the stars use a big blob of solder and drag the top of the blob over to the ring. Once this is done you can drag a bit more over, just use the tip of the iron and just the tip of the blob. After a few passes it will be a good solid bridge and you can let it heat up and flatten down.

Another way is to use a thin wire. Just connect it similar to method above. Try an old driver to practice on.

Have you tried solder paste? it makes soldering so much easier. I had problems with soldering until i got some of that - what a revelation!

Nice improvement! I think this high cri 219b led is about the best choice for that type of flashlight.

I never have much trouble with bridging stars, only a bit, it might have something to do with the temperature of the iron, too hot makes bridging more difficult.

All the 92cri sw45 219b's I have are significant cooler tinted (around 5000K) than the old 219A. This observation is shared by many, including Cutter. I dedomed a few now, the tint goes down to ~4000K, much better to me.

I always just use a pencil mark to select stars. So easy to do, so much easier to undo should you want to change options later.

I have a 2011, great introduction to Li-ion lights. Nice to see it get new life.

very nice mod

+1 to the solder paste - it helps

Pre- tinned wire between 2 blobs ..impossible to screw it up

I always forget about the pencil trick ...why not

I use conductive paint for the Stars…

After trying this new mod at night, I think I’ve ranked my best-tint lights according to how well they render color, from best to worst:

  • BST-wide (and soon, probably a SRK-wide… edit: my SRK-wide ended up very slightly yellow, great overall but not as nice as the BST-wide)
  • L3 L10-219 or Convoy S7-219B (different color temperatures but both fantastic)
  • HD2011-219B
  • Thrunite T10T NW (edit: after looking more closely, I no longer put this on par with the HD2011-219B; I simply picked test subjects which happened to be compatible with the tint)
  • Zebralight H52Fw / CNQG Brass Beauty XP-G2 3B / JETBeam RRT01 early-2013 model (last XM-L model; XM-L2 ones seem to use a much colder tint)

Everything else I have does mediocre (or worse) color rendering, but all of the above are good. It was a big surprise to see the T10T NW fared so well… there’s definitely a gap between it and the two Nichia 219 lights above it, but it seems to be on par with the Nichia 219B light I put together today. It’s a step above any other individual Cree tint I’ve tried. And I’m surprised that the two 219Bs I’m using didn’t end up on the same tier, since they should be the exact same emitter.

Anyway, the Nichia 219 seems to deliver the best tint I’ve found in a single emitter. But multiple combined emitters spread through the ansi white spectrum seems to work even better, and I’m really liking wide-spectrum lights lately.

Oh, and I’m not sure how much tint lottery there is on the T10/T10S/T10T series. The product page on amazon specifies “NW” as “3500K to 5000K” and I think mine is right at 5000K. The “CW” models are listed as “5000K to 8000K”. So, YMMV.

Updated the OP with pictures and more info.

You would have more success using this type of mod to get the light tailstanding, you might need to sand the retaining rings down a bit and maybe remove the spring from the driver board so you just have a solder blob in it’s place to fit longer cells as this mod shortens the battery tube.

Canadian money IS good for something. Nice mod.

Thanks! That worked pretty well. It took a few tries to find a suitable O-ring or gasket, but eventually I found a good one and it can finally tail-stand. :slight_smile:

This was also a good excuse to add the missing O-rings and lube up everything which needed lubing.

Can you please post some beamshots? I also want to mod some cheap flashlights and am curious about the improvement in color rendering between a Nichia and a basic xml. Also how good is the heatsinking, can you run the flashlight at max for extended periods?

There are a bunch of Nichia vs Cree beamshots around. It can be hard sometimes to really show high CRI color rendering differences with pictures since it adds the camera sensor & your LCD screen in the mix.

I was asking because there are very few relevant pictures and movies of the low cri/high cri duo. Also we are in a flashlight forum and all.

There's a reason why there's so few comparison pictures around, Helios was mild about it: it is impossible to do justice to what you see in real with pictures and computer screens. I myself really tried, and although some of the difference is captured, it does not even come close. In pictures you do not see any difference between a 70 cri 4C tint and a 92cri sw45 tint, in real it is instantly clear.

It is very difficult to get good pictures of CRI differences. I have a pretty inexpensive camera and, although I did try a while back, the results really don’t look much like they do in person. I mean, high-CRI lights look good in pictures… but so do low-CRI lights. In person, it’s a very different story.

It’s also somewhat difficult to capture useful beam shots to show the brightness variation throughout a beam, especially on diffused lights. Doing that properly requires a high-dynamic-range photo setup, and ideally a monitor with more than 8 bits per color channel. I try to use a multi-exposure approach instead, showing one underexposed image, one overexposed, and one in-between. It works, I guess, but it still doesn’t really convey the true appearance very well.

In any case, the beam fades smoothly in brightness from the center all the way to the edge, and it’s about 170 degrees wide. The slope of the brightness curve varies throughout that range, but it’s very smooth with no sharp or sudden changes in the drop-off rate.

Without diffuser film, the beam is fairly traditional — spot, corona, spill, and outer drop-off, all in four distinct segments.

I haven’t tried to leave the light on high for very long, but I’m kind of a wuss about using the maximum modes on any light. I treat the highest mode as a turbo on every light, and I try to keep the temperature quite a bit lower than strictly necessary. In any case, I doubt the light would have any heat issues at only 1.5A on maximum.

So is the high CRI visible in these new 219Bs? Does it still make a difference to your eyes?

Thank you very much for the response. In your opinion, is a Nichia 219 setup on a budget light worth the cost/labour, or is it all hype? It just so happens that I do have a 10 bit panel, being a photo amateur.