DQG 26650 driver hacking

1 RUN B+ in all modes
2 Ith 1.52mV/226mV/380mV/.754V
3 FB 1.247V/1.234V/1.250V/1.289V
4 FREQ .623V/.627V/.639V/.680V
5 M/S 5.17V/5.19V/5.21V/5.26V

AO4468 gate:
24mV/.739V/1.791V/2.375V
58khz/696k/657k/657k

9426 gate: B+/0Hz in all modes

Correct.

edit: Resized/aligned images and turned it into a mouseover:

So what you're wondering is, is it just using a different resistor for each mode? Wouldn't that be a terribly weird way of doing it? Although I guess that would explain why there's also no detectable PWM at the LED+/-...

That's exactly what it does.

Shorting across the 2R0 makes medium brighter. Shorting the R500 makes high brighter. And (no bonus points for guessing this one!) shorting the R120 makes turbo brighter. Dunno where the low mode is coming from, since none of the MCU pins are active for the low mode it must be the default current supplied by... something else.

And a mirrored x-ray of the battery side, so the vias line up with the topside pic (most of them, anyway - what's going on under the LTC1871 is still a mystery).

Uhm, well then. I guess we know how to control this thing then. Does jumpering the turbo resistor affect the other 3 modes?

Do the 3 transistors which control this action get hot at all? I just can’t see the circuit well enough to see if these sense resistors are the inline kind (I think current can be sensed without putting a resistor inline with the load in some circumstances, but I’m not certain. If these are in line with the load, seems like they’d get hot.)

Here is the STAR firmware modified to run this light.
~~http://photo.jesusthepirate.com/blf/DQG-26650-TRIPLE/SRK_no_ramp_1.0\_\_DQG-26650_v002.c~~
EDIT: switch pin not set as input.

I added the 3 flashes on stepdown code from the clicky firmware.

If it works as is I’ll be a little surprised.

http://photo.jesusthepirate.com/blf/DQG-26650-TRIPLE/SRK_no_ramp_1.0\_\_DQG-26650_v005.c

  • Fixed SWITCH_PIN setup.
  • Fixed MODEx_PIN issues.
  • Moved function declaration to remove warnings.

I haven't noticed them making heat, but I haven't specifically checked them for that, either. The only part I noticed trying to burn my fingers was the big diode down at LED+. These are just limiting resistors. I could swap the resistor locations around, and change the mode order.

This must be the section handling the low mode, since the other 3 are already known:

The little 180 (or 18C?) is the limiter, or the 010? What's a '010' anyway? My go-to decoder page says it's not a valid marking. Through that 010, there's a direct path to the #3/FB pin at all times. I haven't tried, but I'd assume that shorting the 010 resistor there would affect all modes, unlike the other 3 resistors that all work independently (in mid, shorting the H or T resistor has no effect; in high, shorting M or T has no effect, etc.).

Resistors like R120 and R500 are definitely sense resistors, not limiting resistors.

You may be able to directly measure the resistance of those two resistors you’ve got in the red box. What does that show? I’d measure both directions on each. I’m not sure how the transistor (or whatever) & capacitor there affect the operation of that section.

Basically the whole board could be “low mode stuff” since low mode is just a baseline and all the other modes are higher. Almost any of the small components could be involved with setting the baseline current and voltage… But yeah, that section is a suspect.

Your latest pictures show that there should be a break in the green line in your colored-in pic. Each of the sense resistors connects (through a dedicated transistor) to the big copper pour on the back which attaches to the FETs.

Due to the camera orientation, I cannot see where the via between the big cap and the red box you drew connects.

Immediately to the left of your red box there is a via, this appears to head down to the diode marked XL, through a resistor, and then on to Vin (Pin 9). I don’t think that’s what’s really happening, that’s just what I see.

Personally I am much less interested in this driver board now that I realize that it does not use a dedicated LED controller. It is very complicated for what it achieves.

FWIW, the FB pin is for setting voltage, SENSE pin is for setting current. I think.

Is there a path between LED- and the big copper pour on the back of the board when:

  • the light is off
  • the light is in low mode?

obviously there’s a path for M/H/T

If they are sense resistors, which component on the board is reading their voltage drop? If the LTC1871 were controlling the modes I could see it, but we know it's not. It looks to me like it just boosts up to whatever set point, then the path to ground is routed through the appropriate resistor via each mode's little FET.

The big cap between the inductor & the LTC1871 is on the main B+ pad on the R/H side, its L/H side goes to a longish fat trace that goes down to a medium size cap, the thing marked 82LE, and a tiny cap. There's a big horizontal B+ trace at the top of the inductor. Easiest seen in this pic:

Anyway, I'm about certain there's not enough room in the light to fit a piggyback board on top to hold the attiny - depth in the head is only 5.8mm off the top surface of the original driver, and the inductor takes up 3mm of that already. The only workaround I can think of at the moment is something like gluing the attiny upside-down onto the board where the original MCU is and use jumper wires to route the legs to the right pads. :(

Funny you should mention that. I worked on this some but decided it was a stupid idea. You have to reflow the adapter PCB onto the main PCB of course - either a big hoof tip or some chipquick might do OK, or just hot air. Then you solder your regular ATtiny on top. Total height should be around 3.5mm

So as to this stuff…

My bad on sense-vs-limit. Now that I think about it I’m not certain that they aren’t limiting resistors. At this point I could go either way, without a schematic it’s hard to guess. Gotta figure out what’s hooked up to the SENSE pin to know for sure. This thing can do a weird sensing setup as well as a more normal one. I’m not certain they aren’t used for sensing, but I’m leaning towards you being correct.

Thanks for the clarification on that path, I see it now. The LH side of that cap is a ground of some type, even if it’s not real GND. The medium sized cap you mentioned hooks up to INTVcc (pin8), so it must be a smoothing cap for that. (Therefore the other side, which connects to the big cap, is ground).

Is the mode separation as good with the MT-G2 as it is with a triple?

Just from memory, I think the output currents were something like just a few mA in low, .25A in M, 1A in H, and 1.5A turbo. Of course now we know what they do, those levels can be adjusted with different resistors. I don't know how that compares to when it's running the 3S XPG2s, though.

Sorry, now I forgot what I planned to learn from that. :frowning:

Anyway I went ahead and put an outline on that adapter PCB and uploaded it to Oshpark. It has to be slightly large in order to meet minimum size restrictions, you can file it down.
~~http://oshpark.com/shared_projects/WtKCjCqt~~ EDIT: see post #57

I also thought about putting a via through the middle of each pin. You’d need to file up to halfway through the via, but it would make the board easier to hand solder.

just burnt my driver, bridging the r120.
one of the two mcus, that they are on the spring side, burnt.
If I try to power again the flashlight, this mcu burns again.
I measured 11 amps on high with a KK 26650.


Those are FET’s, not MCU’s. They are just in the SO8 package.
I only know the model information for one of those two. It seems you are lucky, you burned the one we know about.

Digikey has it in stock:
http://www.digikey.com/product-detail/en/AO4468/785-1038-1-ND/1855980

eBay is probably much cheaper for you, luckily there are plenty on there:
http://www.ebay.com/sch/i.html?\_trksid=p2050601.m570.l1313.TR11.TRC1.A0.H0.Xao4468&\_nkw=ao4468&\_sacat=0&\_from=R40

EDIT: fixed Digikey link.

Thanks, is there any which can handle the extra current?

Possibly, but I don’t think it’s a good idea. 11A is 30-35W from a single 26650. 11A looks like it’s pretty rough on a KK.

We don't really understand how this driver works. Swapping FETs could cause big problems. That said, maybe a PowerPAK SO8 style FET would be a good choice. For example, the SIR462DP can handle more current and has much lower RDS(on) figures than the stock FET. You'd have to take a look at what is under the FET and make sure that you weren't bridging anything. I wouldn't do it.

AO4468 is also used on other drivers, most common one is probably the 3-toroid SRK drivers. If you have any dead drivers laying around it's worth a look.

edit: Also, I doubt there is any other FET out there that will survive having the resistor bridged. Being a boost driver, the input current isn't self-limited by the LED voltage like in a DD setup, and that's why these same parts survive fine in other driver designs but not in this one.