Stock driver is first, Lume1 is second, both shot at 1/4000 shutter speed.
My FW3A with lume driver looks like this. In real life it doesn’t look like pwm flicker, but more like candle mode. I tried to record a video of that, but even on manual camera settings it doesn’t really show up.
@Kapsyd One of my older drivers used to flicker in the same manner you’re describing when driven at moonlight lows. To the naked eye it appears to be a “candlelight flutter”. The issue had to do with the switching regulator being driven at such a low current that the voltage changes on the feedback pin weren’t sufficient enough to provide steady output. The sense resistor wasn’t generating enough mV of voltage drop to properly provide a compensation ramp within the regulator, and the result was completely random and sporadic bursts of light that could be considered circuit noise making it to the output.
The resolution would have been to use a larger value sense resistor, but that causes serious thermal issues at high output and lowers the efficiency of the circuit a great deal. As with all engineering, it’s a bit of a balancing act.
FIASCO!!! So, I installed the Lume1 board. If your FW3C is a first generation, the board is flat on one edge, so the Lume1 is not a drop in board. Okay, fine. Filed it unit it fit. Also, the solder pads on the triple emitter board are in a different place.
Then, I looked at my stash of wire…mostly 28AWG. Close enough! I used my medium tip Weller soldering station. And of course this didn’t work. Solder everywhere. No problem; I’ll just unsolder the AUX board and get a finer tip. NOPE. Lost two of the VERY small resistors.
Buy the tools first, then build the thing.
On the bright side, my FW3C is not running the new board (minus the fancy AUX LEDs).
Hello all,
Thanks to all for all your feedback! While I'm glad that I've heard a good number of folks who have successfully got their driver working well, I also hear some challenges and problems faced by some other members. I'd like to apologize that the experience hasn't been as smooth as it could have been, and I really appreciate the feedback.
In all my prototypes, I've tried to make the performance of the driver as good as possible while keeping within some specific electrical requirements I set for myself, as well as additional constraints such as layout and cost. I wasn't expecting the product to be for sale like it has been, so I think I could have made the design a little more user-friendly for people who did not have the equipment for micro-soldering, as well as to have a more detailed notice about a fair amount of DIY required. For example, I hand solder all the components on the driver myself, so soldering on the wires for the Aux board is easy in comparison for me, but I understand not everyone has the same setup as I have! I do appreciate the patience that everyone has.
The key to soldering this in my opinion is to use very fine soldering wire, needle nose tweezers, and some sort of magnification if possible. If it helps, here are some of my suggestions for basic tools which will help in any sort of soldering:
- Precision tweezers - any of them works even the really cheap ones: https://www.amazon.com/Hakko-3-SA-Non-Magnetic-Microelectronics-Applications/dp/B00FZPEWI6/
- Very fine solder wire - this is very important - personally I use a 1lb reel of Kester but any brand will work, you want 0.3mm or thinner: https://www.amazon.com/Kaisi-Solder-Sn63Pb37-fluxed-1-76oz/dp/B07S7H1WP6/
- For mistakes - solderwick, I use a large reel of Chemtronics since I go through a lot, but a narrow, fine-braid one works very well for flashlights and small projects: https://www.amazon.com/MG-Chemicals-Desoldering-Length-Yellow/dp/B008O9W78C/
- TIP - e.g. if one of those 0402 resistors pops out, use the solderwick to clean all the pads first so they are flat with no more solder domes on top. Then apply a small amount of solder on one pad, tack the new resistor on with the tweezers and solder tip, then solder the other pad down.
- Flux - It says no-clean but I always clean the boards after. Kester 186 works well too: https://www.amazon.com/Kester83-1000-0951-Soldering-Flux-Low-Solids-No-Clean/dp/B004X4KOWS/
On to the boards: I was finally able to get a few production Lume1 boards directly from Neal soldered up on my test bench, and I've been able to test the basic functions.
I'm glad to say that so far, everything actually looks good on my bench for the two drivers I've tested.
Component inspection looks good at first glance (I.e. it looks like all ICs are genuine), but I've also requested for a detailed BOM from Lumintop because they did replace some components despite the specific BOM I sent over (e.g. the sense resistor is the same value as the one I specified, but not the exact one I chose). So I'll give an update on that. However, the main components and large passives such as the big caps which are important, seem to check out OK at the moment.
Next, I was also able to test all main functionality and I did not see anything amiss. Specifically, some users seemed to have issues with the lowest modes, while others seemed to have no issues.
As contactcr correctly points out, at the lowest setting for the Lume1, there is in fact a small 'pwm' due to pulse frequency modulation of the buck converter. However, the performance I measured on my oscilloscope was in fact better than expected, with a current modulation frequency of about 2kHz only at the lowest possible ramp value (measured with a 100MHz Keysight current probe), but with very consistent levels (i.e. brightness is consistent so there should not be any flickering like a candle). I'll post more scope shots later on. Keep in mind that this is not actually PWM, and it should appear visibly better than an actual PWM at the same frequency. This modulation frequency very quickly increases before leveling out and should not be noticeable to the eye without moving the flashlight around quickly, and it certainly should not appear to be anywhere remotely like candle mode.
In addition, Kapsyd mentioned something about flux on the surface of the boards, and calipsoii talked about sensitivity of some part of the circuit - the general idea is correct. At the lowest current levels, the feedback the circuit is receiving is very small, on the order of tens of uV. This is a very sensitive analog signal and care was specifically taken in the design to ensure best physical routing around these sensitive lines. However, it is possible that any additional dirt / flux / contaminant on the surface of the PCB can contribute to leakage current and messing with this signal line at the low current levels, which can cause 'flickering' issues. The easiest way to see this in action is to touch the feedback components with a greasy finger when the driver is running at lowest mode, and you may see the output flickering! Even probing the circuit with a scope needs to be done with care since the scope probe adds significant capacitance to parts of the feedback circuitry.
That said, from the samples I got from Neal, the boards look to be very cleanly soldered, and I haven't been able to replicate this issue in my production driver boards, or even the pre-production ones from Neal.
Regardless, the components most sensitive to this are the op-amp (the smallest 6-pin component on the board, near the edge of the PCB surrounded by a bunch of resistors and caps), as well as those resistors and capacitors. The specific op-amp chosen has very high precision, very low input current and very low input off-set, so it's critical that this component is used and not swapped out. A small amount of contaminant in a humid environment can also introduce leakage currents.
For cleaning, I like to use a stiff small brush and isopropyl alcohol to clean boards. Something like this with the bristles cut down work really well, or a small stiff paintbrush: https://www.homedepot.com/p/Lincoln-Electric-Flux-Brush-KH589/202939867. I hope that could help alleviate the problems.
Meanwhile, I'll continue to validate the driver boards I have and I'll post more updates over the next few days (will be out during the weekend but back after).
Thanks to all again!
I’m very impressed by the driver and aux board; well done! And, yeah, we’re modding lights. It would be weird if it all went smoothly.
I also have a first gen FW3A and as soon as I opened the bezel I say the negative solder point on the MCPCB is not where the aux board wants it. I have not looked below to see if the driver sockethas a flat and requires sanding of the Lume1 driver but that is doable. The question is, is there a compatible MCPCB I can get, of the proper size and thickness and with the solder points in the compatible place?
MTN and Kaidomain should. Beware of polarity though, it might not be what is printed on the aux board masking. LED4Power is a no, Noctigon is a no.
Personally I would try scratching a new pad since I think only one is out of place. It looks possible but very narrow and probably not recommended after all
Sorry for the out of focus pic this is from when I installed Lexel board:
I got mine installed and working. The aux board was… not easy, but it only took two tries 
I’m pretty satisfied by the aux (red) level in the dark, which is probably all I’ll use. I do wish the lowest level were a bit lower, but plenty of time to play with that in software I s’pose. Thanks loneoceans!
How do I know I will get the negative if I scratch the masking where the aux board has the openning?
I only have my picture to go by but it actually looks like the space might be a little too tight (I edited my post to not recommend it). You have to stay on the “raised” area and it gets really narrow around the optic leg hole and postive part. I had more room to work cause Lexel’s board put the connection further out I think.
In any case it’s good to know this so here goes (my drawing might be slightly off cause im going off this image).
Look very closely at board and stay on the black shaded side of the orange line. If you look at the board you can see where the raised traces are for pos and neg. The sunken part (orange) is thermal DTP area and if you cross into that you will short to ground. You only want to scrape the white part until you see some copper.
Thank you contactcr! I will have a look tomorrow to see if I can se the traces and if I have room to scratch.
In te meantime, does anyone know if the thickness of the MCPCB is te same in all of those FW3x lights that had that kind of board? I think that in more recent lights, the thickness varies, between my LM10, FW3T and FW3C. But maybe it is just my impresson, as I did not measure. If anyone has the dimentions of the MCPCB contatct is showing above, it would help me to see if I have or can order one, before disassembling the light.
Titaniums and SS had thick MCPCB where others have normal 1.6mm i think
BTW send Victor Zhu a msg that you wanna purchase FW3 MCPCB. He send some my way when i asked.
^ the practical answer
Ordered 3 drivers, so hopefully I’ll be able to get at least one running. Thanks everyone for all the tips and advice. If anyone wanted to do a video of the driver install, that would also be awesome.
lume1 live stream this weekend! come create bonds while I create bridges.
smash the like and subscribe.
(in all serious I considered doing a video but I feel like I would have to edit it down a lot)
Dangit… now I think I want to swap the LEDs in my LM10 with Lume1 driver. I’m not looking forward to soldering all those tiny wires again. 
Recently I pulled out my bare aluminum FW3A. It has 4000K SST20 in it of a superior tint bin to the 4000K in my LM10. No green at all. Noticeably better than the tint bin in my LM10. Also that bare aluminum FW3A isn’t practical. It gets burning hot too fast. Now I want to swap the LEDs in both these lights.
Grrrrr…… the effort required to try to get a “perfect” light.
Did you try swapping just the lenses first? If it’s that noticeable it’s probably not just the lens but worth a shot.
Both lights are using 10507 optics with Sapphire lenses that I swapped in afterwards.
Actually the LM10 doesn’t look that bad. Tint still looks way less green than that in my Emisar D4 with SST-20.
But when put side-by-side, the tint of the SST-20 in my bare FW3A is noticeably better than that in all my other SST-20 lights. Even at low-power, it’s no greener than the tint in my high-CRI 4000K Nichia 219C light.
PM me values you need. I was too cheap to pay for a “book” so it might take a full day to find it. LOL
