Another one, this time with the stock resistors still functional.
Remove the stock controller, optionally place it on the floor and smash it with a hammer for proper disposal.
These are the 3 pins that go to the 3 gate drivers that go to the 2 buck controllers that run the 3 FETs. What an elegant design! :Sp
OCD? What's that?
See? Now they're mounted properly.
While still hot, smoosh the heatshrink down flat. Holds the wires in place tight enough that no epoxy is needed.
Power for the brain is picked up from the inboard tab on the big yellow capacitor (the cap's outboard tab goes to ground). The brain's ground comes via the copper mounting frame. The single black wire connected to the big driver's three PWM pins goes to any of the 105c pads that used to control the 7135s.
The only thing left to do is attach two pins to the back of the switch PCB, and corresponding female connectors to the ends of the switch cable. Makes disassembly so much nicer when you don't have to futz around with that stupid switch bezel just to remove the driver.
I measured L1=2.48A, L2=2.47A, L3=2.35A. All 3 channels have different resistor combos because of the vastly different PCB traces, but odd that the low one only has a single R010 resistor, while L1 has a R010 & R110, and L2 has a R010 & R068. Every one I've seen of this style has those same resistor values in the same locations, but I guess that's pretty close for a $5 driver.
The uH rating is less important than the current capacity, since everything that goes to the LEDs has to go through the toroid. Which of these would be least happy having 15 amps shoved through it?
I have not yet tried the little red one. Ones with fewer turns (less total length) or multiple or fatter conductors can carry more current with less resistance than ones with more turns or smaller/fewer conductors. As we know from selecting 'good' FETs, a few milliohms can make a big difference when the input voltage is so close to the LED's Vf.
The thing that has me stumped is, I haven't found any listings that allow sorting by number of conductors, and just looking through thousands of pictures on Digikey isn't really getting anywhere - everything I find that might be suitable is non-stocked. :(
You're right, they are all listed by uH (inductance) and amp rating. There are lots of stocked models on Mouser. The uH rating has got to matter, right? Isn't that the whole reason for using the toroid? Otherwise we could just stack a resistor with the same resistance as the toroid, but we know that the inductance is what's making the difference here.
Problem is that without an oscilloscope I have no idea how to measure the inductance.
D = Drain (negative wire from the load, here that's LED-)
S = Source (battery ground)
It goes between LED- and ground just like a 7135, except it doesn't limit itself to a fixed current. Whatever your cell can deliver is what the LED gets (minus a tiny amount from the FET's internal resistance, this one is around 0.0045 ohms).
Yeah, glued on upside down with JB Weld, and flattening the legs and rebending them is probably a bit outside the intended design parameters, but you know... :evil:
It doesn't get hot though, on any mode. It might, in the other versions using the inductors since (I think?) in that setup the FET spends more time inbetween fully on and fully off, which is where FETs can get a little unhappy and start making some reasonable amount of heat.
A copper plate wouldn't help here anyway, since if the tab touches anything that could carry heat away the FET is bypassed altogether (LED- to GND).
Correct me if I miss something, but at say 5amps and an internal resistance of 0.0045 ohms, it will only generate 0.1 W, that will easily dissipate through the connectors.
Yes. Except I'm not selling it (yet). Comfychair's work was really what brought this all about in the first place. DrJones & Werner have provided excellent yet simple free firmwares for anyone to flash their own drivers.