USB battery dedicated to the task, add 5 bucks.
First I was thinking just get rid of the AAA set entirely, briefly considered a fallback scheme (since the USB simply pops out of existence, not great in the dark lol)... But space issues forced me to abandon the idea. Back to bypass, and it's flaky (logic control vs the auto-on sensing going on in the battery), but can be turned on with a little magic handshake lol...
Now it's cutting off "randomly", and I'm catching a whiff of "not so happy". Nothing seems excessively hot so I push on to see the deal... Ruled out cable disturbance, it was heat... Somehow.
Turns out the two key components in all this are closely matched for where it runs out of pedal, and as the LED heats up, draws more current, pushes the battery into shutdown.
Confirmed this by letting it get it's juice from a heftier port and the LED began to blue. Oops. Seems ok, we'll see.
Ok so told you that to tell you this...
The original battery box has the PWM and what not in it and so for that moment, had decided it was best to just connect the USB across the battery terminals and well... They also fight, as mentioned.
The heat indication with only 0.5V increase tells me they were relying purely on cell impedance to limit current, and that since it's piping 100% "PWM" on high, with no ballast, and I'm really only interested in running it on high, and there's a magic trick to turning it on?
Pfft. And the battery can handle turning it on and off and everything! Lol
I'm thinking I can ditch the box, throw a 1N4001 in series with it and go raw and probably be just fine. I may want a small bit of real resistance in there but I think the wiring being so thin helps there some. Don't need to lose much
It'll also look a bit cleaner lol