Solarforce K3 Modded & Compared - Pics w/Vid *Pic Heavy*

The zener diode is a controlled leak in the power supply to the MCU. Being installed 'backwards' between the MCU's power pin #8 and ground, any voltage above its rated voltage is 'leaked' away to gnd, so a 4.3v zener will work like a pressure regulator or a bleed-off valve and maintain voltage on that line at... 4.3v. If the voltage applied is less than the rated 4.3v, the zener will block current flow like a normal diode.

The ~200 ohm resistor is there to limit the current available to be leaked away through the zener. Without the resistor the zener would either burn up or waste too much battery power.

Appreciate the explanation. Loving the modded K3! Hot little demon that it is…

In trying to take amperage measurements at the emitter, I took a close look at my stacked chips. The tail numbers I’d gotten previously seemed low at 4.05-4.12A. Seems like one of the chips I added wasn’t making a good connection so I made an attempt to add solder to it to see if I could regain that chip. And it seems to have worked! I’m now getting a solid 4.40A reading, even off not-quite-so-fresh cells. No flickering numbers, easily repeatable 4.40A. The numbers add up to 4.56A at 380mA chips, but I understand the ratings aren’t usually correct and 380’s tend to run a bit lower while 350’s tend to run a bit higher. So this makes perfect sense now and all seems to be working as it should.

Also added some thermal paste around the copper slug in the pill to help it make contact with the sides better. And as mentioned I added a copper plate on top of the chips.

I’ll get the cells charged up fresh and run the heat test again, see if there’s any visible difference in the amount of time it takes to reach Hot! :slight_smile:

Oh, and you're aware that's a faux-copper MCPCB with a dielectric layer, right?

So I might just have to implement my plan and take it a step further, make a pillar atop the heat sink to bring that dead phoenix back up from the ashes. And here I was thinking they did something right! That’s it! I’m not buying another Solarforce light no matter what! This week.

Thanks for letting me know.

<—-goesoffgrumblingandcursingunderhisbreath,swearingvengeanceonthepennypinchingtightwadsthatunderdesignflashlightsforaliving

And where’s the best place to buy Noctigon pads for MT-G2?

Had I known that the “copper pcb” was of that variety, I’d never have glued it down! Now how am I gonna get that bad boy outta there? Do you know what bin the MT-G2 is? Like, does Solarforce use the 5000K PO group emitter? Is that one readily available? Cause I’ll flame that one out of there if I have to! Hate liars, and that bogus copper board is a blatant lie! OTOH, it does make a ton of lumens, so a change would only make it better, right? :slight_smile: This is the side of modding that gets exasperating…

Thank you very much for a very detailed and in-depth experience of the K3. Sure looks like a must-have :)

Strip the pill down to just the pill & star, blow hot air from the bottom, lift the LED off. Then pry out the MCPCB.

IOS is the only source for Noctigon, it's their own house brand.

of being full of hot air, but even I don’t have what it takes to do that! lol And besides, there’s that chunk of copper under there too. Glued in solid with more thermal adhesive. Might as well put it on the burner and reflow it off. IOS here I come! Anything else they specialize in that I should drop in my shopping cart while I’m there?

You’re welcome Lothar. It is indeed a very nice little light. In stock form with a couple of 18350 IMR cells it’d be a pretty neat light to carry. Not all that difficult to make it into a beastie if you so desire.

Going to go ahead with my original idea and create a solid copper column to fit into the host and go through the aluminum shelf in the pill, the Noctigon will sit directly on top of this copper column and heat sink straight into the host body. I will probably reflow the Noctigon directly to this heat-sink in the same step I reflow the emitter and solder contact wires. We’ll see how that goes. I will do the driver end as I’ve done the HD2010 and relieve the solid copper for the driver to sit into. Maximum heat sink and maximum thermal capabilities in one solid chunk of copper approx. 7/8” diameter by approx. 1” in length. I might even take a few pics when the time comes. :wink:

Copper bar also ordered, and then some…for the next build. lol Somehow this budget hobby of mine turns out to be not-so-budget.

http://www.bernzomatic.com/item.html?id=16

My wife has a small torch for carmelizing crème brulee, that would do this as well. I need to keep that in mind for heat shrinking too. Thanks Comfy! :slight_smile: (Whew! Finally! Something I don’t have to buy!)

No no, you don't use the flame, you use the soldering adapter but with no tip screwed in. There's a ceramic catalyst burner thingamajig in the adapter and nothing but hot air comes out the end.

I hope they’re not expensive, cause I need one of those too! Did I mention next Friday is my birthday? LOL “Don’t worry Dear, I’ve got presents on the way! :slight_smile:

$26 huh? My neighbor has an oxy/acetylene torch…and a .50 BMG! :slight_smile:

I decided to let my camera tell me the pecking order of my lights. So I shut the blinds, set up the tripod and got the G1X ready for some consecutive shots. Camera on Aperture priority, locked at f/5, ISO 3200, and then shutter speed according to the available light. Each light was pointed at the ceiling at an even distance from the canvas as the camera. The canvas is a 16x20 from when my son was 6 mo. old, he’s about to turn six later this month.

The lights…1) L2P w/EDC+ Triple XPG2 720 lumens OTF 2) HD2010 XM-L2 T6 @ 4.18A De-domed 3) L2P M3 XM-L U2 @ 4.28A De-domed 4) 3D MagLight Elektrolumens Triple XM-L2 @ 8.48A 5) Solarforce S2200 Stock Form 6) L2P K3 MT-G2 @ 4.40A

The camera says they fall in the above order, with the K3 being brighter than the S2200. And I didn’t even bring the batteries up to fresh charge! The heavily modded HD2010 and M3 fall in the same range as the XM-L2 is driven at 4.18A while the XM-L is driven at 4.28A, they’re both de-domed and have Erik 5 mode drivers. The HD2010 is sporting a nice chunk of copper to help it keep it’s cool.

Hey, which tailcap is on that no-text L2P second from the right? Do you know if it's sold separate?

I do believe they sell it separately, if not, I have 2 from putting the SS ones on.

Your choice of forward or reverse clicky. This is forward http://www.solarforceflashlight-sales.com/product_detail.php?t=FA&s=18&id=125

Man that k3 sure is nice! I havent really ever looked at solarforce lights just because they dont seem to do it for me. That k3 however is WOW.

VERY nice light. Must be fun, I would walk the dogs with is and light up the park for a little while. My dog loves the park at night and runs tree to tree peeing every last drop :stuck_out_tongue: Whilst I have fun with my LOOMENS!

I’m enjoying it immensely! :slight_smile: Just ran a secondary heat test after adding the copper plate to the chips. Started at 86.8º. It was a couple of degrees cooler at 5 minutes, I ran it 10 minutes and it got up to 146.2º…only a 13º rise in the second 5 minutes. The heat is going all the way down to the tailcap, the tailcap read 101.4º after 10 minutes on high and the top battery was 121.6º at the positive end! The cells went from 4.21V down to 3.94V while still warm. I’m planning to let em rest a little while and see what they are.

Definitely puts out some serious heat! :slight_smile: I’m thinking you could over-easy an egg with the out-the-front heat :wink:

Medium is something like ~700 lumens, very nice tint and very useable output for a variety of things. While a 10 minute run on Hi on a hot summer night might allow it to make your hand tingle, the same run on a cold winter night would probably feel very nice!

Ahh right, since the LED drops a majority of the voltage, the 7135s only see a fraction of the 8.4v. And the 6v limit I assume is on the signal side of the 7135, which is regulated via the zener/resitor (by limiting the power sent to the MCU).

How come this solution hasn't come up before for driving LEDs in series?