S2+ KD 219C triple build and questions

Flashlight Modding and DIY Parts Featured Mods
1

Hey guys,

I had a matte black Convoy S2+ with Nichia 365nm UV LED, but decided to take it apart because that UV LED needs a boost driver because of the high Vf.

And then Kaidomain got their 219C triples in stock and i had to buy it for that price, because they’re a true bargain at $7.87 including optics.

I decided to use a Nanjg 105C with double 7135 x 8 = 5.6 Amperes regulated.

Driver potted with epoxy:

I gave it a copper sheet wall for better heat management.
I also did a spring bypass.
Drilled a hole through the 1st star (and removed that trace) and solderd it on the + on the other side.
From there a short piece of wire that sticks out, to solder the LED wires onto.
It fits in here:

Pill with raised shelf, couple of copper 1mm shims soldered together with the pill.
You can see i have no lathe…
But it’s the minimal height you need to be able to tighten the lot together before you run out of threads in the head.

Nearly all parts that go in the head.
More copper sheet to fill up a little between triple and head, also improving thermal path forward a little.
It centres the triple, which is necessary because the MCPCB is not screwed or soldered onto the pill shelf.
To (try) make things easier i used some super glue to fix the optics on the LED board.
The legs that stick through the board are filed down so it is flat like the pill shelf.

Decided to solder a copper rim around the driver unit, and added a retainer ring.
That should hold it in place and make a good thermal path from driver to pill and head.

I have opened up the head considerably, so that the bezel opening is almost 20mm so it doesn’t shroud the 20mm optics
I put in a 21mm glass lens.
So there’s only a narrow 0.5mm ledge / rim left that holds the glass lens in.
No way an O-ring is gonna work, it will squish out.

…so i stuck a disc of adhesive foam on the front of the glass lens before i assembled the head.
Then i cut the centre out and removed it form the lens, which was a messy job…

Another messy job was the thermal grease on the threads of the pill, to improve thermal conductivity among the parts in the head.
I have no pics of that, i needed my full concentration (not to smear it on the visible parts etc…).

There, that’s better. :slight_smile:

You may notice the matte finish of the optic.
Turns out one of the LEDs has a bit more solder under it than the others, is what i discovered after having glued the optics to the LED board… :weary:
So i had to compensate for the height difference by sanding the front of the optics where it was higher…
I kind of liked the idea of a matte finish anyway, so that’s what i did.

Before i tell you about the result, a pic of the tail switch assembly:

Double springs, fat switch. :+1:

Because the driver turned out a little large, i used a thin metal ring between switch and boot to gain some lost space.

So, everything would seem okidoki, right?
Well, i’m not sure…
Yes, all seems to work fine and it gets really hot on a fresh VTC5 too, obviously.
But i expected more than what it does.
It’s not that much brighter than a XM-L2 on 2.8 Amperes.
Did i ruin the optics?
The tint of the light is quite yellowish, and the reds, although better than Cree, aren’t impressive either.
I’m not sure if these 219Cs are ‘9050’ actually…
And they look more like 3500K in stead of 4000K.
So what is the quality of the KD triple emitters then?

So i end this topic (of my first triple build) with questions.
Anybody else try something with those KD 219C triples yet?

2

I build one. Tint is as should be.
Mike

3

How about the Lumens output?

4

I didn’t measure, but it is insane bright and current draw is over 10A on turbo with FET+1 driver.
Mike

5

You have done a neat job on the build. Maybe the thermal grease on the pill threads is not allowing a good earth path to the driver?

6

Over 10 Amperes??
Link to your topic: S2+ host build question

Mine is limited to 5.6 Amperes by 16x 7135
Way too much already for that tiny S2+ i.m.o.
Do not leave it unattended on high!

7

Thanks.
Thought about that, but that should be good because of the driver retainer ring i added.
There the threads are clean(ed).

Unfortunately i still haven’t got a decent setup to measure the current…

8

Isn’t it that issue? Triple build, XP-L HI scraped by optic?

9

My XP-G2 and XP-G3 triples in S2’s have a lot of spill, so my S2+ XML looks brighter, but the triples put out more light. I didn’t bypass any springs and used FET+1 drivers. Including my meters cables, I measured 3.7A, so probably 4A without the meter. They get warm very quick on high. I went for the S2’s because the copper spacer was bigger.

I find flow-able silicone sealant is great for sealing tight gaps. I didn’t widen the bezels, but thinking about it now.


10

No, fortunately that’s not the case here. :slight_smile:

11

I did improve the electrical path, because i use a constant current driver, and i wanted to be able to have full output power until the battery runs too low.
But comparing it by ‘ceiling bouncing’ with a C8 with XM-L2 with half the current, i was disappointed in the output.
I think something is not right…
So i wonder about the KD triple now…
I think i’m gonna have to build another one and see if it will do better, because i bought 2 triples from KD.
If it’s the same, i can reflow 219C from Clemence and see if it also does the same.

My 4x XP-G3 S41S on VTC5 is MUCH brighter than this…
But that one possibly does more than 10 Amperes… :smiley:

Well, apart from the copious amounts of thermal grease, i want to be able to take it apart again too.
So i didn’t try glueing or sealing the lens in the bezel.

12

I would say its a combination of the amps across the 3 LEDs and tint. I was just looking at (2) X6 and SS triple, Fet builds side by side. Both XP-L HI but one is 3A other 5A2. The 5A2 looks nowhere near as bright as the 3A.

Is that first sheet of copper around driver soldered to the driver? Hard to tell in the pics and making me think weak ground. Not questioning your work but it hard to tell some things in the pics. Was the SW mounted to the board when you doubled the springs? I learned the hard way to take off the switch whenever soldering springs or bypassing.

13

At a given current the spot from a single will be smaller and brighter than a triple even if the triple totals more lumens just because it’s spread out(like the same amount of jam on a bigger piece of toast). Also, if you can’t check current how do you know all the chips are firing? I still get misses on occasion. I’ve become cynical about my own work so I do current checks before assembly(especially with potting).

14

Hmm…I guess it’s possible i just expected a lot more lumens…
But compared with my recently built C8 with MT-G2 on 4.9 Amperes (14x 7135 zener mod), this triple is much less bright while still more than half the power of the MT-G2 light

Yeah, i don’t always have the camera up and ready, so some things are not shown too well.
Yes,the copper sheet is soldered to the driver.
More solder between the 7135s and the sheet for thermal (and electrical) conduction.
The (later added) ring around it is clamped between pill and that alu retainer ring, all is screwed down quite tight.
I can’t imagine there is notenough good contact between body / pill and driver.

Haha, yeah, i learned the same the hard way too… :smiley:
Those switches are made of butter… :smiley:

15

You’re absolutely right.
I should get a setup to measure the current.
I have the stuff to do so in stock (power supply and regulator unit with display) but haven’t gotten round to actually setting it up yet… :person_facepalming:

16

I love your fabricating!

17

Did you have luck finding a 700mA 3.6V boost driver for the Nichia LED?

18

I have several AA / 14500 boost drivers from SK68 clones and what not, and with 2x AA i think it will be just about right.
Now to decide what will be the host…
Maybe a black Romisen RG-A6 (2x AAA)

Today i bought the Uni-T clamp meter so soon (well, soon… maybe 2 months shipping time…) i can measure tailcap current of this S2+ triple. :+1:

19

Thanks for documentation, it helps to see it.

20

:+1: