Don’t hold your breath. I ordered mine (the same as yours) on the 7th of April. It was out of stock. They got some in and shipped mine on the 15th of April. My driver is the one you see in the original post- NOT the 3 terroid one. Yours sounds like it’s coming from the batch right after mine. By the way mine was super easy to open up. Yeah the centering ring is different for sure, because 2 of mine aren’t what you’d call “centered.”
Come on smart people, I'm slowly figuring this out... well, more like guessing. Probably going completely in the wrong direction.
AMC7136 is being used basically just to control the AO D436 FET, and not supply LED current directly? And the 3-resistor set, with the added jumper, effectively overrides the current limiting on HI mode and runs it direct drive, is that right?
that 3 resistors not jumpering. it has very little resistance.
LED current sense by 3 resistors and voltage drop feedback to CS input of AMC7136 which turn off FET when CS pin voltage goes above 50mV.
This is a linear constant current regulator / current limiter
AMC7135 is also same. but it has internal sense resistor.
This driver is complete garbage.
L1+/L1- 2.20A
L2+/L2- 2.60A
L3+/L3- 2.08A
I think the torroid driver was in the 6.5A range, so this one is similar; with full batts. The original boost version would maintain full output longer. This looks like a cost reduction to me, coils cost a lot more than cheap FETs.
That’s too bad, I was considering a FT SRK when they had the torroid driver. Now it is a host with a glorified contact board. 
And the old driver managed to run all 3 LEDs at the same current.
edit: And it's not really usable as a host, either, the machining is a complete joke. The FandyFire version is still OK (the $45 one, I haven't seen the cheaper version) but only available in gold.
With the '000' not-resistors removed it runs at really low output, like 450mA. And if I jumper across the resistor array it goes direct drive, so I don't know how this is actually controlling the output. What kind of current is present here, can I use pots to set/balance the output in place of the '000's?
And since the resistance value will be something very close to but not quite zero ohms, I'm guessing it would need to be something like a multi-turn 0-5 ohm pot, does such a thing even exist? I only have stupidly high value pots (5k and up) here I could test with but I don't know if those will have enough resolution for this down at the low end.
The FETs have internal resistance (called RDSon) that acts as a current limiting resistor. I’d leave the resistors shorted. It should be fine like that.
No, it's obviously not fine like that. Maybe my labeling up there wasn't clear.
There's three drivers here on a single board. The section that runs LED #1 is putting out 2.2 amps. Section for LED #2 is putting out 2.6 amps. And the section for LED #3 is putting out 2.08 amps.
More than half an amp variation between LEDs isn't acceptable.
OK, I see. That’s the nature of FET on resistance; quite variable from part to part. Also, I doubt the FET resistance is actually being used for feedback, probably just a current limiter now.
This will not be an easy thing to fix. Since one LED is getting only 2A, to even it out you need them all to be 2A. This means a 0.025 Ohm resistance in place of the sense resistors. You can use four R100 resistors to get that. Might be worth a try. Good luck with it.
I will note that any lights that run LEDs in parallel from a single driver probably have a similar emitter current imbalance due to slightly varying emitter Vf. It may be even more drastic there.
I just ordered a $32 SRK from WB, hoping these ones (still) have the good driver. Fingers crossed.
Those numbers were with the same LED connected in turn to each driver output.
Try the other way; each LED in turn to the same driver output. That should indicate any difference between LEDs. It doesn’t really matter too much for you, just an exercise in knowing.
I fixed it. Well, not really fixed, just made it irrelevant.
You can't imagine how much I wanted to just pound it with a hammer. POS.
I like it. You have just put a smile on my dial. Thanks comfychair. 
I would have done the same thing, after the hammer 
How hot does the host get now?
How many coulombs/sec are flowing now? 
I wonder how many lumens these new drivers are putting out now.
Thank you Comfychair for your determination to get to the bottom of how to get this driver performing better. Your conclusion that it can’t be readily improved seems reasonable to my amateur electronic brain.
My driver is different from everyones. It has one toroidal and a large through-hole type capacitor. It also has only one LED + and - (all LEDs connect to the same contact pads). I did a quick current reading when I had it apart and it was shy of 6 amps. Got mine from brotherhaug or something like that on ebay.
Received my SRK on Friday. It has the same, new DD driver (picture of driver contact plate for identification without dismantling the flashlight). I’m disappointed at fasttech, because I especially asked if it has 3 terroid driver before they sent it out. PWM is noticeable annoying to my eye. LEDs are not 100% centered, but much better then on mhaneln’s SRK. Opened up mine and removed the reflector just for interest what’s under there. When I put it back, they were out of center just like mhanlen’s but I rotated the reflector to different position to get best results.
Overall quality is good enough. Came in thick carton box with instructions sheet, spare O-ring and lanyard. Bezel is made from aluminum and I can’t see no big dents or imperfections.
I have other cheap SRK on way from Wallbuys, hopefully with a good driver. Plan is to mod it with XP-G2s on copper for more throw.
Exactly what i got from Fasttech a few weeks ago.
Really happy with the service from them, but not about the differences between my other SR Kings.
Same driver in my Wallbuys SRK.
