[WIP] 17mm DD+single-7135 driver / single sided / Dual-PWM

You should post a thread on your F6 build. There have been many discussions, but I can’t remember seeing any detailed pics of the insides of one.

Ohh, sorry, didn't take detailed pics of the insides - it's a tight pill cavity, not enough vertical space to piggyback a driver in, and the switch is mounted on the driver. It's a uni-body design, so to work on the LED wires, you have to solder downward quite a ways. I posted bout my F6 fiasco here: https://budgetlightforum.com/t/-/26078, in the hacking thread. I can sort of get the end tailcap to stick now, and modded the "Convoy F6" to get 3A (added 2 R100's), upgraded the wires to 22 AWG, added 22 AWG wires to bypass the springs, upgraded the LED to a XM-L2 T5 5D3 and get bout 800 lumens now with a nice look'n tint, but still with the stock driver.

I see what you mean with little details. Seems like a few folks did the full F6 mod, but not much in details around.

That info about the driver cavity is all I needed to know. Thanks

Not to be denied, I spent 10 or so minutes scraping the solder mask off the spring side negative ring that OSH left on, then cleaned the board and oven reflow soldered all the parts on. This is a very easy board to populate, with lots of room for the Pomona programming clip.

I used TK’s BLF-A6 firmware, with two PWM channels. I tweaked it to match the Battcheck results (for accurate battery testing and LVP) and turned the memory function on.

I haven’t done much testing on a battery, but it’s putting out 5.8 amps at 4.2 volts on my no-sag power supply.

Thanks Wight and everyone else who made this possible.

Cool! Just finished a build using my 2nd one - I ended up scraping off the remaining solder mask on the spring side ground ring, so not as much scraping as you Smile. I put this driver in an e-switch light, the UltraFire F35 from FastTech, piggyback'd in. I drilled a hole straight thru the LED+ pad, then drilled a hole in the stock, stripped driver, used as a contact board (and switch mount), through the center + pad, so my LED+ 22 AWG wire runs direct from the bypass wire of the spring to the LED. Used 3 22 AWG wires in between the stock driver and A17HYBRID-S, and got the better SIR800DP FET. I got an amazing 6.45A measured with a LG HE2 @4.18v, using a XM-L2 T6 4C/Noctigon from Richard. Think I got lucky with a low resistance LED or something... I'm using a PWM value of 4 for moon mode on the 7135 channel, phase corrected not FAST. Occasionally I see a flicker when turning it on in moon mode. Heres' my PWM values for the 5 modes:

#define MODES 0, 0, 0, 0, 80, 255 // Must be low to high, and must start with 0

#define ALT_MODES 0, 4, 24, 255, 255, 0 // for secondary (7135) output. Comment out if no secondary output

#define MODE_PWM 0,PHASE, FAST,PHASE, FAST,PHASE // Define one per mode above. 0 tells the light to go to sleep

With those modes using the T6 4C on a Noctigon with an extra 3/4" copper disc reflowed to the MCPCB, I got the following results:

Mode Amps Lumens

Moon: 0.01A 0.6

Lo: 0.03A 9.5

Med: 0.35A 126

Hi: 1.89A 595

Turbo: 6.45A 1547 @start, 1431 @30 secs

Throw on turbo: 25 kcd (316 meters)

It's a pretty decent spread for 5 modes, but I'd make the Lo a little bit higher - I'd target 12-15 lumens. I can't explain the moon mode amps=>lumens results. Moon doesn't scale to Lo very well, but at these low levels, maybe my light box, or maybe resolution of the amp readings is deceptive.

Nice mod Tom :), 6.47A & 1549 lumens from a T6 4C is a lot :slight_smile: good to know those from lower flux bins XM-L2’s & warmer hue bins can reach as high as an XP-L.

Maybe it was only a extra good LED you got or……

Maybe those XM-L2 T6 4C’s is more easy to get good results from than the U2 & U3, i always though my X6-SE T6 4C looked brighter than my X6-SE U2 3C, and i remember some reported similar in the GB thread.

Thanx! Dunno - it is pretty weird - I'm only using 22 AWG wire too, but maybe you're right - something hot bout the T6 4C's... Yes, I thought the X6 SE's (3) I got with the T6 4C's were doing pretty darn good too. And I like the tint of the 4C's - definitely more neutral than the 3C's when look'n at them side by side, and no rosy tint as some of the D's have.

I find it so interesting how we all seem to like so different tints the best :slight_smile:

I can’t get enough of the 3D and i really like that hint of rose, but find the 3C & 4C way to yellow and/or green.

In NW anything below the BBL i like much more than anything over, i haven’t seen enough WW to know if that trend continuous there also, but i wouldn’t be surprised if it does.

And i know some people genuinely like 1A the best, but in my eyes it feels like someone is slapping my eyeballs with that harsh blue light :smiley:

I wonder if it is some natural bias in our eyes that does it or the environment we live in or something else that is making certain tints look more pleasing to us.

It ticks me off you got 1 more amp than me. I even went back and fed the driver with the twisted pair 12 ga wire right off my PS, bypassing the clicky switch. 5.8 amps @ 4.2 volts. Is that new FET THAT much better? I’ll have to buy some. I have no idea the bin XM-L2 I have on my bench heatsink. It’s probably in the 5000K color range. It was a pull from a donor flashlight.

On another note, I got rid of the blinkies in the hidden mode. Now there is just battcheck. All I did was change this:

#define NUM_HIDDEN 4
#define HIDDENMODES BIKING_STROBE,BATTCHECK,STROBE,TURBO
#define HIDDENMODES_PWM PHASE,PHASE,PHASE,PHASE

To this:

#define NUM_HIDDEN 1
#define HIDDENMODES BATTCHECK
#define HIDDENMODES_PWM PHASE

Now a medium to long press at ML goes straight to battcheck. I’d like to say I did this all on my own, but TK told me where to look.

Hmm I’d suspect your lower current has more to do with your emitter than your fet

The FET doesn't make hardly any difference, since we're only talking about 0.001 to 0.002 ohms difference between the 0.004 and 0.005 ohm variants to the SIR800DP. I have tested them back to back, and I found very little difference---so little in fact that I couldn't say that it was statistically significant.

Some of it is probably from the spring bypass, but the rest of it is probably due to the LED's vF. Also, LED Boatguy, you probably will never see the same amps on that huge heat sink that you'll see inside of most actual lights (the heat sink keeps the LEDs cooler; heat decreases vF significantly). There is also a huge difference between different LEDs.

Boy, I gotta double check, and verify with different cells, but I did do mutiple tests and saw in the 6.45A range multiple times. Could be one super low Vf LED - hit the LED lottery or something...

I've seen this before, where I get great #'s, then can't reproduce it with the supposedly the same components, same design. I agree with Richard - my #1 cause would be the LED. However, it's still interesting that I thought these new L2 LED's melt/die over 6A, so this one appears to not only obtain high amps, but survive it. Edit: maybe I'm wrong - maybe the new L2's peak early but don't blow up - can't recall, djozz tested this out I believe... I could be getting output beyond the max peak at 6.45A...

The ones that blow up "early" won't usually even hit 6A direct drive with a single cell. You can usually only blow them with a buck driver or other higher voltage power supply, since a single cell just can't provide enough voltage.

I have seen some LEDs lately that are pulling more amps than those from a few months ago. There seems to be some batch to batch variation going on. It's hard to know what you're going to get.

No GND ring on both sides (component and spring) on all three driver boards that I just received from OSHPark.

I see that I am not alone since LED Boatguy and Tom E have had soldermask problems also. I’ll email OSHPark support as suggested above.

Any idea why our boards show such variation in fabrication results? Just curious.

This driver is my first attempt at a driver build. I’ve been watching this thread and the EE A6 BLF thread quite a bit.

Thanks!

I almost wonder if there is something wrong with the Gerbers. Too many problems just to be a coincidence at this point, unless you all got them from the same panel.

fyi, this was the ticket communications I had with "Dan" at OSHPark. Interesting read:

-------------------------
Hi!
For order #nnnnnn, for these boards: https://oshpark.com/shared_projects/7sNNyipn, all three came with an error where the bottom should have exposed copper circle of 10mm, instead, it's all covered in purple. Others have ordered these same boards without this problem. Any ideas what happened? Can anything be done to replace the ones I received with corrected ones, or is there something wrong with the project files?
Thanks!
Tom
-------------------------
Hi Tom,
That's really weird. I'll get a new set of those shipped out to you. Doesn't appear to be anything specifically wrong with the project files, although Alex's boards have a slight history of confusing the bajeeezus out of the fab. I think this is the 4th design of his I've seen that has some random issue that comes up. Fortunately, they seem to pop up once, and never happen again.
-Dan
-------------------------
The replacement boards I received had no problems, but RMM - you could still be correct, there's something goin on confusing the fabrication, probably based on where in the panel it is, and/or what it's around.

Just received an order of these boards, and all seems to be well. I could've overlooked something, but everything seems to be where it should.

I know that in the past there have also been mask issues with some of wight's other boards as well, but I can't pretend to understand what is going on there since the gerbers and renders look alright.

I have ordered 3 batches of this board and personally haven’t had a problem with them, but I have seen inconsistencies with other boards from Oshpark.

I will say, Oshpark customer service is the best I’ve ever talked to. Knowledgeable, personable, and non-robotic.

I have had huge differences in silkscreens (precision and cutoff) and sometimes some minor milling errors, but never a huge mask issue like this, and I've ordered thousands of boards.