What did you mod today?

I made a pair of S3 quad E17As 21700 flashlights - one with 2700K and the other 5700K. I got the S3 hosts from Kaidomain, the 6 amp drivers from the Convoy Store, and the Nichia E17As from Virence.

This time I used 26AWG Teflon wires (vs 24AWG silicone) which were much easier to solder to the VR16SP4m boards and to curl up into the driver cavities. The effective gauge (Wirebarn) of 2 26AWG wires is 23AWG, so good enough capacity given the 5cm lengths. I super glued the reflector spacers on to the boards to prevent any damage to the delicate E17As. The 2.5mm thick shelves do a great job dispersing heat which can be felt almost immediately when on high. There were some burrs on the shelf holes, but other than that, the hosts with included pocket clips and lanyards are of high quality.

The 21700 size is starting to push it for carrying in my pocket, but still acceptable. I had to frost 1 side of the lenses to remove the donut holes. The resulting beams have noticeable hotspots and the tints and outputs are impressive. So is the extra capacity of the 5,000mAh batteries. I’m torn as to which flashlight I like better, but I’m leaning towards the 5700K despite my tendency for warmer temps.




Great build!

Would you mind posting a beamshot?

Here you go. I’ve alway struggled with white wall beamshots since they never look as good as what I actually see, but I understand why people want to see them. I’m sure my next iPhone will be much better, but that may be awhile.

Looks great. I don’t see and donut in the beam. Can you confirm it’s not visible also in longer distances?

agree,
to get rid of the donut, he had to frost the lens, though the photo looks like it was taken before frosting

Very nice modding work, and great photos! :+1:

What an awesome build ! Nice work! Love the 21700 host version.

I just picked up my first s3a NW and what I nice host. I did an emitter swap in my first E03 today and it was tough due to the small size but came out good. I was proud of myself and going to post it here until I just saw yours, coincidentally in a similar host LOL. I’ll admire yours and just mention mine . Nice job

I recieved my first e03 this week. Supposed to come in with NW, but accidentally shipped to me in CW. Figured I’d mod it with an extra sst20 high cri. I was surprised when I opened it in and saw a 4040 emitter in the 14mm (?) mcpcb. I should’ve know but I didn’t, so I threw in an extra sst40 5000k I had around. It’s now brighter then all of my newer 14500 lights. Looks great.

Side note- weirdly I think the ui order changed to L-M-H. I’m pretty sure it was M-L-H when I got it. I don’t know why that would’ve happened, but it’s definitely L-M-H. Also, there is a significant difference between the three modes, and originally M and H were so similar I couldn’t distinguish between the two. Anyway, love the light more now.

Original pic- compared to my NW S3a.

Now, compared to the same NW S3a (this pic was taken today, the “original” comparison was taken days ago, at night - hence the different looking cct. The emitter is the same in all of these pics in the S3a on the right)

E03 on turbo vs S3a on turbo (hard to tell, but e03 is brighter now. Again, this pic taken today during daytime)

Thanks for looking, and appreciate any feedback!

After early struggles with huge caps, etc., DEL finally solved the problem properly, first appearing in the Q8 driver, and his own OSHPark drivers --> you really need that 4.7 ohm resistor on Batt+ coming into the 85. We have it before the diode. Everything is happy after that on the scope. All new 85 drivers since 2017 have that resistor. I see it duplicated in all the Chinese Anduril, and custom firmware drivers, even many non-85 drivers.

Odd. Sounds like it’s needed almost (if not exactly) like a terminating resistor on a transmission line. Current spikes might cause reflections at the end of the line, screwing with the voltage the little beastie sees.

There definitely aren’t any donut holes at any distance. The frosted lens takes care of it remarkably.

Here’s where I tested just to see if I got lucky with this reflector, but the donut hole still appeared.

Here are the frosted lenses and a pic to check to make sure the reflector is centered.

Thanks Artiet59. And nice job on your emitter swap!

I have the 4.7ohm resistor with capacitors layed out the same as the D4 driver and still saw that behavior. But I’m not sure what the capacitor values are and the physical positioning of the components on my driver is different so maybe it is not suppressing the spikes as well.

That's strange. The resistor has been the single thing to make the big difference. Looking at the Lumintop EDC18 driver (same design as the FW3A) last night. I noticed 3 caps of 3 different sizes to ground - this is a common for reducing spikes. The one cap, we use across the MCU V+ and V- should be as close as possible to the pins - I know DEL emphasized that, posted here:

https://budgetlightforum.com/t/-/39069/1103

This refers to C2, the 0.1 uF cap, schematic:

A lot is explained in DEL's driver thread's OP here: https://budgetlightforum.com/t/-/44006

C8F 21700 + D4v2 driver + dedomed dogfarts :slight_smile:

Depending on battery max current draw is between 8A and 12A.

Reflector of 21700 version seems to be worse than 18650 version, there are some faint rings (visible only on white wall):

It does not look a lot bigger than 18650 brother but it feels more chunky because of diameter and tube texture:

Difference in tint is obvious:

Work in progress pics:

Button leds are connected to red channel:

D4v2 driver was 1mm too wide, so almost drop in replacement :wink: And the main feautre - flashing pads!

Here is the test with old 30Q and temperature limit set to 50C. Turbo lasts over 4 minutes. It stabilizes around 2A:

Album with all pictures

I managed to modify configuration so lighted button works, dim in first 4 levels, can be configured dim or bright in lockout mode. hwdef-Emisar_D4v2_c8f.h and cfg-emisar-d4v2-c8f.h

Thanks. My attiny85 on qlite mod had mostly been working fine, but there were a couple instances of strange behavior where I had to disconnect the battery for awhile to let it “reset”.

Here is how I made the circuit to emulate the D4 driver.

That C1 is in the same position as the C2 in your diagram; it’s connected right to pin 8. But I found that it is a 10uF instead of the 0.1uF that is recommended. So that could have contributed to the bad behavior; it would not filter the spikes as well. I switched it out for a 0.1uF and hopefully it fixed it. I cannot reliably reproduce the bad behavior so it’s hard to tell for sure if it’s fixed. If I still see problems I will increase the R5 resistor to 10ohms.

C1 and C2 is in parallel, so i removed in my drivers 10uf as useless

I’m no expert, but from my reading, putting different capacitors in parallel can be useful for decoupling. They act at different frequency ranges.

Im using like this for E-switch

Its kind off snuber .Best way is to connect oscilloscope to see real picture.
And if you have no fet in circuit so i think you dont need that snubber at all

I don’t have an oscilloscope so unfortunately I can’t see exactly what’s going on. I was seeing lots of bad behavior when I was testing on the bench with longer wires. Then it got much better when I put it in the light. So it made me think it was a voltage spike issue. But yes, it is only 8x7135 driver so I’m not sure what’s going on.

There are some awesome mods on here, very inspiring stuff.

Im back with some more LED swaps…

1. Installed a CULPM1 in my Copper Noctigon KR1 - amazing emitter for this light, doesn’t get hot because of heavy copper.

2. Installed a xhp50.2 in my Aluminum Noctigon KR1 - great all purpose light, lumens must be around 3,000.

3. Installed a Osram CSLPM1 Green emitter in my Astrolux C8 - WOW! thats a lot of green, must throw 800 meters too, with a lot of spill.

will upload pics asap.