What did you mod today?

Great job Docc :+1:

I’m lucky enough to have a lathe…. motivated enough to use it too :innocent: The last few days I’ve been putting in some time here and there. Head is bored out, pill is made, bezel is made, body is bored and threaded at one end. Taking my time getting the fit and finish nice. These are the only pics I have so far.

With the help of my brother who has more soldering experience than me, I finally did the driver swaps in my Convoys. We struggled a lot with a too-weak soldering iron, but in the end I can proudly say none of the lights in my house have the dreaded blinky 3/5 firmware now :smiley:

What driver are you running it on? I’ve got a host (amutorch AX3) inbound that I think will be a good test platform for the lantern branch of anduril.

I swapped leds in a grey D4 twice today and both leds are not what I hoped for.

Start: grey D4 with XP-G2 S4 2B. I had a D4 before with these leds and they were the old type that had the best throw when dedomed. This grey one has new type XP-G2’s.

On fresly charged 30Q, at 10 seconds:
3030 lumen
16.1 kcd
When removing the ledboard, the heatsink paste left on the shelf had a distinct pattern, likely it accumulated at distinct postions when the light heated up during use, which does not sound ideal to me. It will be replaced with AS5.

First led swap: 4000K Luxeon V2, from Arrow. I expexted a comparable tint as my D4 with Luxeon V’s, but it was not so. The hotspot was a bit yellow-ish and the spill a bit pinkish. Only a snob is bothered by it but I hoped for the close to perfect tint that the Luxeon V has in my D4. Perhaps the smaller dome makes it behave different in the TIR, not just making the smaller hotspot. I must check this led in a reflector light as well these days.

On a freshly charged 30Q at 10 seconds:
3714 lumen
18.8 kcd
The output at least is pretty good!

So further to another led swap: 4000K 80CRI LH351C’s. I found those at Arrow for 80 cents a piece with free shipping. The 90CRI variants that maukka tested were too much above the BBL, perhaps these are better, just like the 4000K 80CRI Cree’s often have good tint. But no, the tint of the hotspot was also above the BBL so not perfect either. At least the beam is slightly better than with the Luxeon V2: not yellow in the hotspot and this time some bkue in the spill. Btw, nothing of this can be seen on photographs so I won’t bother with beam shots.

On a freshly charged 30Q at 10 seconds:
2817 lumen
Did not measure throw.
So the output is less than both the XP-G2 S4 2B and the Luxeon V2, but with this so-so tint the 80CRI is not really worth that IMO.

Result today: a not so impressive D4 with messed up ledwires after 2 swaps, and some new knowledge about 2 leds I was curious about. Up to the next promising led to check out :slight_smile:

I did 2x L1V2 4k and 2x L1V2 5K, it probably takes the yellow out some but the pink remains I believe. Need to double check. I liked it better than the 80CRI 219C 5K that was in it but, like you, I think the original Luxeon V 4K is really nice and even.

Speaking of the D4, here is mine with its new Neodymium magnet :

I put the body and tailcap in the lathe at work and did a 15mm hole through it.
I then put a 3mm thick magnet and glued it along with the spring board with some loctite

Very elegant solution on the D4!

I’am building a fet+1 driver now and it will go in a Skilhunt DS20 with Luxeon V2 led and Andúril firmware.

Mine’s a lot smaller than most of the mods in this thread, but I swapped the emitter in my Nitecore TUBE with a 95 CRI 5mm Yuji LED. I’m very happy with the results. The current limiting resistor also had to be swapped. I put in a 33 ohm resistor. This should give me about 36mA to the LED.

Not something I physically created today but I did the layout on this 15mm Utorch S1 3 channel driver. 1+4x7135+FET built for Anduril. Probably the most insane routing job I’ve ever had to do. Space was at such a premium I had to put a resistor under the Tiny85 (which has bent pins to fit on a T13 pad so it’ll be slightly raised off the board)

Here’s the copper layer

I was going to suggest to drill a small hole between the pads to let R4 sink into the board, but then I saw that even in that small gap is a trace running.

Wow. :open_mouth: Love your work. :+1:

I "think" I have an S1 and would be very interested in this as well, thanx CK!!

You probably know this, but I think the most important parts are the C1 cap and the R5 resistor (4.7 before the diode). That 4.7 was the key to cleaning up the signals, DEL discovered. Not sure if the 47 ohm and 100K are that super important. Of course the diode itself is optional, but adds reverse polarity protection.

I'm not a fan of the triple channels - I prefer the simpler FET+1. I had a lot of troubles getting the FET+bank+1 ramping looking smooth and it's still not that smooth. Not sure if TK worked out the kinks in Anduril, but if she did, would love to see how. I believe, not sure, she overlaps the channel transitions which may have smoothed it out. The NarsilM FET+1 tables ramp great, but the triple has a noticeable transition. Of course to create the ramping tables, I used TK's tool, a Python script, to build them. The other major problem with triples is they use up my limited supply of eagleclaw 7135's.

Not a problem because I can still use this driver as a FET+1 with NarsilM v1.3.

I left C2 off already, now I was thinking about excluding R3 (100K FET Pull-down) cause worst case you could add it across the FET’s S & G legs if it not being there ended up being detrimental somehow. That’d let me move R4 to R3’s position getting it out from under the MCU.

I'd go by DEL's 17 mm design here: https://oshpark.com/shared_projects/qsbLVgh0

He eliminates the least needed components, C2 and the 47 ohm in line to the FET gate, but kept in the 100K. I know he had a specific order of priority and definitely that 4.7 ohm guy before the diode was very important. Before DEL's design in of the 4.7, it was hit and miss whether a 25/45/85 would work or not. Richard at MtnE had a different approach that worked but DEL's design is better, as proved on the scope.

Ohh, fyi. The classic 8x7135 ATtiny13A driver design was out of spec for the 13A, but seemed to always work. The 25/45/85's turned out to be more finicky/sensitive, so that's why the 4.7 ohm is needed -- it should have been in the classic 13A 105C/105D boards all along.

I think what I’ll do is net the two R4 pads together, that way if anything is wrong (doubtful but still…) I can cut the trace and solder the resistor in place later.

Edit:

Welp I’m an idiot…

My first thought was 0402’s but nixed that right away. better think again

Stuck a Lexel TA driver and a Luxeon V2 (3000k) into my Wowtac A2. Got to break in my Zeny soldering station.

Follow on …

Sweet. Love the centre. Did you make the adaptor?

Pinkpanda, nice stuff, will continue to what with excitment (and envy.)

Enjoyed your earlier 21700 light as well, hope you finally got some cells to power that beauty :wink:


Reflowed a couple of SST-20 4000K 95+CRI into a CF8… now its warm and tight, the way I like it (but probably not as bright on Max.)

Yeh, I made the adaptor because I didn’t want to use the body to hold the head for turning as it is a soft metal. The head is ti.

Thanks mate. Yeh, I got a couple cells eventually for the 21700 light. Looking back now that was a pretty simple build in comparison to this one.