Modding the DQG 18650 Tiny III into a triple

hmm I’ve seen weird things with e-switches on dual channel drivers, but I’m afraid I can’t be much help on the Nanjg

GADzoooks!

I would bypass the switch, if that doesn’t work, then another driver (or swap it in from the old Triple cubed) . In my experience with trailer lights, it is ALWAYS the ground if you hooked up the wires in the right order. I don’t know if that applies here though… The switch may have some issues with that specific driver, sometimes the tolerances do not stack in our favor….

I think I figured out the problem.

Keeps fingers crossed hoping I got it right

For the bottom of the driver I was using a copper post made of 3 pieces of sheet copper layered on top of each other. I don’t think I made the post quite high enough so with flat top cells part of the cathode might’ve been touching the tops of the lower layer of 7135 chips or possibly the resistor. Not sure if that’s the issue, but I noticed the light worked better with button top cells.

I stuck the remains of the foam ring that was on the underside of the original driver around the copper post and now it seems to work fine. Hopefully it will continue to do so.

After insulating everything possible with arctic alumina or Kapton tape I fixed the problem with the unintended moonlight. I also discovered and fixed a loose ground connection from the driver to the body.

Trouble-shooting this was an exhaustive process. The leadwires and bondpads on the star are really showing wear. There’s a real risk that with too many more desolders I might have to install fresh driver wires. I might even need to reflow to and grind down a fresh star.

I discovered another problem. The switch PCB seems to be press-fit from the inside of the light. Somehow it got pushed down making the switch take an excessive amount of force to operate. I pushed it back up and used arctic alumina to fix it in place. Unfortunately, I got a little too enthusiastic with the probe pushing it in and think I damaged one of the vias through the board to the switch. Result is the switch in the unscratched host no longer works! GRRRR.

I’m not sure any of the microswitches I have on hand will fit. I’m going to have to see if I can pry out the PCB and install a different switch, or maybe bypass the PCB. Oh well… task for another day.

Took the host with the broken switch and tried to remove the switch.

Prying it out from the inside didn’t work. I ended up completely destroying and removing the switch mechanism… everything up to the metal cover cap. I did come across a thin rubber membrane inside the switch mechanism. I think this does mean the stock switch is waterproof.

Not so the replacement switch I’m going to try to install though…

Finally got it all working in the new unscratched host.

4.5 amp DrJones Mokkdrv driver
3x XPG2 neutral 4000k 5A2 tint.

After destroying the original switch I had to install a replacement switch. I glued a small micro momentary switch I had on hand in the pocket where the original switch went with arctic alumina. Was a pain to get working after I accidentally ripped one of the contacts off, but it works now. All modes function fine. The new switch has a softer touch and is more responsive than the old switch. The old one didn’t always function even when I felt it click. New one always functions with a click.

This light still has a downside though: Now that I have everything working I’m thinking 4.5 amps is too much! At max power it gets hot quite fast. Too hot to touch in 1 minute or so. Thinking I might strip off 2 or 3 of the 7135 chips.

I am in awe of your persistence sir….

Great stories again, the failures and subsequent repairs sound all too familiar exept that I am not half as persistant as you are!

Thanks. Took a lot more time and effort than I thought it would.

Near the end I was thinking I’d have to make a fresh star. The pad for the positive wire was no longer even visible. Its inner edge had already rubbed off long before and I’d covered that portion of the pad with arctic alumina to prevent getting a short to ground. Unfortunately, the outer portion of the pad was now no longer visible either. I couldn’t get the solder to stick without a visible pad.

For the first time I tried using solder paste in the area where the pad was. That actually worked. It cleaned off the gunk around the pad and created a solid bond. My bondwires are looking raggedy though. Hopefully I won’t wear through them removing the driver to take off 7135 chips.

On the upside, the outside of the light looks absolutely pristine and it works great (except for being a little too high current).

The thin glass lens I was using in front of the Carclo optic cracked. I was worried that would happen as it was very thin and had a lot of pressure on it with no o-ring. I was also concerned it might shatter if the light fell bezel first. I ended up removing it and replacing it with a much thicker coated acrylic lens.

Ordinarily this would have prevented the bezel from screwing on because the new lens was so thick. But I modified the lens by filing down a ring on the outer edge. Result is the lens is quite thick, but the portion under the bezel ring is quite thin.

This actually makes the optic look more balanced in the light because the Tiny III’s stock TIR lens was built in a similar manner: with a flat ring going under the bezel slightly back from the front surface of the lens.

Problem solved.

The issue I was having with the light not turning off was caused by a bad ground connection. The driver was soldered directly to a leftover trace from the stock driver. However, the new driver was flexing or moving slightly each time the battery was installed and this was putting stress on the trace. I added more arctic alumina to help anchor the new driver in place, and changed the ground connection from direct solder to a small jumper wire to allow for flexing. Light has worked flawlessly since.

I’m currently running it with DrJones Mokkadrv in a Nanjg 105c with 10x7135 chips. Gets quite hot to touch at the head after a bit, but not quite burning hot.

I think I should order some of the new 2015 Mokkadrv. I’ll bump up output back to 12x7135 and use the configurable turbo timer in that version.

I was planning to convert many of my triples to CREE high-intensity XPL, but I think this one isn’t a good candidate for it. At 3.8 amps, I don’t think it draws enough current for the XPL to make much of a difference over XPG2.

Or I could change the driver to a custom FET from RMM with special modes. Perhaps 100% with configurable turbo timer, followed by ramp down to 25 or 30% (50% is too high), and then have other modes below 30%. That way I could have ridiculous lumens (probably 3000+) for something like 10 seconds before the turbo timer drops output down to 1200 lumens or so.

Actually, as much as I like Mokkadrv, I have to admit the idea of installing a FET driver that gives ridiculous lumens for 10 or 15 seconds sounds pretty appealing.

Guess I’ll have to check with RMM to see if he can program the turbo timer to ramp down to less than 50. With a FET driver and a Samsung 25R even 50 is too much for this light.

You sound like you are on to something, that would be a jaw dropper to pull out the Tiny III and pop ~3000 lumen off, then casually drop it back to 800 or so to finish looking around… Turbo from off would make that thing a good option for disengaging from anyone not wearing a welders helmet at night… Let us know what you decide on and how it works out. Fantastic thread for me so far…

Besides thinner wires & an extra short turbo timer, you can also use a high capacity battery like 3200-3500mha for a longer running time AND a more contained heat profile.

I plan to use all of these ways to build a hopefully well managed, but very bright FET driven EE A6 XP-L triple in the future :wink:

To use a really strong high drain battery like the Samsung 25R, is almost asking for overheat issues in very small triple 18650 lights.

Also i hear that a triple XP-L is a cooler option than the XP-G2, because you don’t run the XP-L’s as close to the limit, and you get more output also :slight_smile:

Interesting. I assumed triple XPL would be hotter as it draws more amps.

However, I tried this light with a high capacity ICR cell and FET driver. At least with XPG2 inside, it didn’t help. It still got way too hot way too fast. A short turbo timer is still needed.

Of course there are other options with a FET driver. Such as having the maximum output cap out at less than 100%.

Yeah i also thought that at first, but i asked in the A6 GB thread. And it think it was Dale who explained it to me. And when you think about it it makes sense, the more overdriven the emitter is the less light & more heat you get, and it is a smaller less efficient package.

And i don’t think a triple XP-L necessary need to pull much more than a triple XP-G2, if you don’t go crazy with the low resistance mods.

What high capacity cell did you use? If the 25R can pull 9A i would assume for example the pana B cell would only pull maybe 6A, that’s still a lot of course, but it could hopefully be manageable with a 30-60sec turbo timer depending on how hot you like your skin that is :smiley:

Firelight2
If you want to get head from the body just try to heat thread part up to 150°C and then unscrew them.

who can make this modding for me?))

Hello Firelight2,

This mod looks really awesome! I want to make a triple, small, 18650 crazybright light and the DQG seems the perfect formfactor.

Did you try the mod with 3x XPL HI and FET driver?

I’m tempted to try it, but any links to parts or previous knowledge would most certainly save me some broken or wrong parts :smiley:

Okay, just bought a grey DQG tiny 4 for €21 on gearbest flash sale…

Guess there is no way back now… :open_mouth: