The case has to be negative. Reason: Cathode Direct Thermal Path emitters. You know, those Nichia 144As, for example. 
This is one of the things DavidEF pursues to address, I believe. I remember hinting at these kind of issues months ago on a discussion related with the 144As, but no one seemed to care much by the time. It was about time. :-)
Go go Gadget ultra-high efficiency anode infinitely variable regulated output gazillion disco modes buck/boost plus direct driver ffs!!! 
Cheers ^:)
Hmm… maybe im missing something but I don’t think it works like that. Our drivers now have a pwm signal going to the 7135 gate pin. This signal allows current to pass through the 7135 across the other two pins. Whatever side of the LED this regulation is on will still require a separate pwm signal from the mcu that tells the 7135 to let the current through. No?
If you “reverse connect” a 7135, you would have to feed the battery + to the OUT pin of the 7135, and the 7135’s GND pin to the LED +. The PWM signal has to be higher than what’s on the 7135’s GND pin or it won’t do anything. So will it be higher with this “reverse connected” configuration? Why don’t you just solder a 7135 with a LED and try it out?
Perhaps I’m the one missing something. This is the whole reason I started this thread. I didn’t think the LED would care if it is sharing the positive lead with the AMC7135. And I didn’t think the AMC7135 would care either, as long as it is otherwise connected properly. I may have to prototype this idea on my breadboard and see for myself. It seems we need to know conclusively IF it will work before we can figure out HOW to make it work in a regular driver.
If you connect an unPWMed voltage to the 7135 gate pin will it will remain open all the time. So if this pin is shared with the positive lead from b+, the positive lead will essentially “let itself through without concern of the PWM signal. But again, I may not be correctly visualizing what you would like to do. Can you sketch it up quick?
Let me know if these pictures don’t show up. I really need a new photo storage/sharing app. Google is no longer nice to share inline like BLF does.
Some confusing (at least for me) results from your use of “pwm” in the different places. I’ll understand that the driver output has been “pwm’d. Within the circuit though it would help to only use pwm as the signal designation not the resultant led current and only apply it to the mcu>FET/7135 control pin(gate)connection.
In your diagram it appears that you are boosting the pwm signal(giving it more current varying capability) by using an fet, then powering the led and limit the current with an amc using that same signal? If so you are still regulating the current on the neg side. Then the amc becomes unnessesssry because you can control the current with the fet.
Actually, the diagram is very much simplified compared to what I’m actually wanting to do, which is to feed multiple remote emitters from one driver, and use the (also remote) AMC’s to regulate them in groups. The reason for that is because if I send one big hunk of amps to the remote LEDs, some of the LEDs will take more power than others (those that are closer to the source) and the difference will be seen in the light output. Grouping them allows better consistency of light output. The FET, then, is the part that is actually redundant. But, if it works the way I’ve drawn it, then the FET saves me a third wire, which is my purpose. So, that makes it useful to me. It’s true that I’m still ultimately regulating the current on the negative side.
The added complexity isn’t necessary if you don’t mind running a third wire off the driver to the remote AMC’s. Then, it becomes like my first drawing, with the “driver” board supplying PWM+ and BATT+ and BATT- along three wires to the remote location(s) where the AMC (s) and LED (s) take care of the rest. Going back to my original question, is there a way to do this without having to add either the third wire or the extra driver complexity? I don’t think it could be done without a FET on board, but maybe the transistor could be removed, if we can get a PWM- signal from the MCU instead of PWM+ signal.
RBD, I was going to use a P-channel FET instead of our usual N-channel FETs for this. The added transistor is only there to reverse the polarity of the PWM signal, so that the FET can be reversed, so that I can send only two wires to the remote AMC’s and have them still work. None of this would be needed, and I could use literally any normal driver, if there was such a thing as a “reverse polarity” AMC that I could use at the remote location(s).
Feeding the pwm as power over long wires will get you very ugly EMC conditions leading to major problems
You will need 3 wires
Ah, ok. Thanks!
EMC ≈ Electro Magnetic Coupling???
Sometimes all of you use too much abbreviations, and honestly, without a proper forum table of definitions, can get a bit hard to understand some stuff.
PWM already is a hideous PoS when used as a lightsource output modulator, so no big deal imho. Due to the relatively high frequency (KHz) used and the typically low duty cycle (strobe), you may as well call it “KHz strobe”, or something. :-)
Cheers ^:)
Somehow I don’t think he’s referring to the visible strobe effects. Sounds more like device problems.
I thought I understood this stuff until I saw those diagrams, lol It's simple. On all these switching/regulation devices we're using, the control voltage, or gate voltage must range (depending if on or off) from equal to, to a few volts higher (more positive) than than voltage on the current path going through it. When that current path is a ground return, that's easy. If that current path is a feed from a positive battery terminal, then what is higher than that? The solutions are to
a) generate higher voltage with some kind of gate driver/level translator (they exist).
b) use a different switching device that works upside down (they exist).
c) Make the case positive and attach the positive lead of the led as the DTP lead so you can still switch the negative. That means turning the battery backwards.
I think that about covers the main flavors of choices. All of them are possible, but they all have some drawbacks.
