Triple P60 single 18650 adding AMC7135 chips, how many?

It's probably a triple XP-G or triple Nichia. Or, less likely, a triple XP-L. If it is a triple XM-L, let us know what PCB you mounted the XM-L's on and we'll tell you how far you can go. Otherwise we'll all assume that you've got whatever mounted on a triple Noctigon.

If you've built your triple well and achieved a good thermal connection to the P60 host (the body of the flashlight), I'd look at what RMM is doing with his S3 EDC over at mtnelectronics. He's using 7135 based drivers like what you have.

I’ve built triples with 7135 based drivers up to 6.0 amps but I know others have pushed harder. I find 4.5 amps is a happy medium in terms of output, run time and heat as long as you wrap the pill. That’s for both XP and Nichia emitters.

At that drive current I seldom need to use the high setting anyway.

Great response thanks guys, it’s a triple cree XP-G2 from KD so the pcb isn’t copper and the battery is a Panasonic NCR18650B.

So when i’m stacking additional chips on top does it actually matter which one I stack on top of? for example the driver already has 8 chips so if I were to add an additional 4 would it matter where?

If you want to go much higher than 3 amps than you’ll need a higher discharge battery.

For longer runtimes id suggest the panasonic PF’s or BD’s or if you want maximum brightness at turn on and lower voltage drop than the vtc5/20r/25r/HE2/efest 35’s will all do fine with higher discharges.

As for how high, you won’t be able to push it much harder without a direct thermal path copper MCPCB. I’m not exactly sure how hard you’ll be able to push it with an aluminium MCPCB. Match’s results suggest that for an XPG2 on aluminium, anything higher than 2.4 amps is just wasted as heat. Although I’m sure those numbers would have little correlation in regards to a triple board. Maybe someone else can chime in on that part :slight_smile:

Did you push it that high using normal single 18650 batteries or was it IMRs etc…? My triple nichia seems to be stuck at 2.7amps even with 12x 7135s

My triple XP-G2 S6 originally did 13A from a single 25R, had to detune it (by switching to 22AWG) to bring to down to 7.5A with a turbo ramp down to 5.5A. It pulls 6A from an nrcB. I’ve never seen any other light be able to pull over .5A rom an NRCB (however my battery tester can pull 8A from them but again I’ve never seen any other light get more than ~4.5A)

Like we all said in your other thread I wouldn’t trust your DMM readings man.

Yeah I know but I just wanted to know if the batts used make a big difference, the dmm might be off but off by 2amps?

Comparing the brightness between a 4.5a xml at 1500 lumens and an xml rated nitecore 800 lumens, and they both look the same…

Right so after a quick look at batteries it would seem I can get Panasonic cells rated at 10A discharge which are compatible with my current charger (thrunite MCC4) and anything higher and I will have to buy yet another charger! ah a typical hobby $$$

Just to clarify then, if I did have a higher discharge rated battery and lets for argument sake say I used a copper pcb, if I was to add an additional 6 chips does it matter which of the 8 currently there I stack them on top of?

No, it doesn’t matter where you stack them. That’s just for the sake of information of course. Doing a lot of stacking to feed a non-direct-contact PCB is futile.

Yeah, or more. Sometimes significantly more when dealing with DD mods. If you doubt it and don’t want to do the normal work so many other people here have done to their DMM, follow comfychair’s suggestion. That should make what’s going on pretty obvious.

Non direct contact PCB?

The metal in a typical MCPCB does not directly solder to the thermal pad on the back of the LED. This creates a high thermal resistance, which in turn causes a larger temperature rise while the LED is operating. Noctigons, Sinkpads, and a couple of less common boards allow you to solder those things directly together, reducing thermal resistance.

Take a look at this diagram in order to get an idea (current gen Sinkpads and Noctigons are slightly different):

Also, to see a real world example take a look over here to see djozz peeling back that (bad) dielectric layer on a non-direct-contact MCPCB. Both djozz and Match have posted useful graphs showing the advantages of directly soldering the LED to the metal. Note that they both use exceptionally large heatsinks.

Excellent info thank you.

I definitely need to look into this, I have an XM-L2 p60 module which I was using at 3A with a 7135 driver and it de-domed itself due to the heat!

Did you do the foil wrap? If it gets that hot, you should feel it. Unless you're tailstanding your lights on high...

Oh it got hot alright ha I was just playing about dabbling as it were

…I’d like to pass on that I learned the hard way.

It’s already been mentioned that you can place the extra 7135s anywhere you want (they’re all parallel).
As long as there’s room in the pill I first stack on the LED side. That way if after the light is assembled I want to raise or lower the current I can add or remove chips from the battery side without removing the driver from the pill.

If this is your first go at stacking chips and you have an unwanted driver lying around I suggest you strip off a 7135 or two and practice stacking, it’s a bit of an acquired art. If you don’t have an old driver you should probably order a few of whatever you’re going to use (Nanjg or KD). Hopefully it won’t happen but it can be pretty frustrating when the only driver you have doesn’t work after hours of tinkering and you have to wait for the slow boat for a replacement. (Been there, only once!).

I test my driver multiple times. First before modding to make sure it works properly and is actually the driver I ordered. I’ve received regular KD V2s when I ordered the moonlight mode version and vice versa as well as mix ups with Nanjgs.

Once the driver is confirmed OK and what I think it is I test after each 7135. I make sure it’s not only still working but the amperage has increased as expected. Either is not the case it’s time to check my work.

When I first started stacking I did two drivers with one or two chips, so excited with my success I put 8 on my next driver (for my first triple). Put it all together and nothing! I wasted hours trying to figure out if I had bridged something, overheated something or what. Never did figure it out and guess what, I didn’t have a spare driver.

Never again, always have a spare part or two and check every step of the way.

Good luck!

Many DMM’s won’t accurately measure current above 1A even if they are rated at 10A. Often this is due to under sized probe wires that add resistance. The easiest fix that’s good up to 5A or so is to just use thicker probe wires. I did this and current readings jumped ~20% in the 2A range. Beyond that it might be necessary to mod the meter to get reliable readings. I don’t believe clamp meters suffer from this but don’t have one so couldn’t say for duress.

It’s really not worth the effort to stack chips beyond 3.5A unless you go to the trouble to maximize heat transfer from the led to the host. Direct thermal path pcb’s are the first most important step, excellent surface contact to the pill is next, and finally good contact between the pill and host completes the thermal path. At high drive currents a smallish host will and should heat up quickly. This indicates a good thermal path.

Those KD 3-up boards are only 1mm thick, you're investing a lot of time and effort into something that's going to be extremely limited on output before you've even started. Put those LEDs on a real board!!

Just buy this and reflow the emitters.

Thanks everyone for the advice i’ve ordered some Noctigon boards and look forward to mounting them on my copper pills for some excessive because I can torch action

The TIR that comes with the KD triples is pretty good, but the pegs are larger than the Carclo ones the Noctigons are made for (2.5mm vs. 1.8mm). Fairly simple fix, just drill the holes a bit bigger.