I did a similar thing with a rack of test equipment. I had a PC power supply on each shelf, and used an NPN transistor to turn on the power supply whenever the USB port connected from the main computer to that shelf rack was turned on. Crude, but effective, out of band communication.
The local USB hub on the shelf was powered from the PC power supply on that shelf, so a single long USB cable from the shelf with the control computer could command the shelf (including the USB hub on the shelf) to fully power up or down, and then control several test devices through the USB hub, and none of the USB devices on the shelf (some of which were bus-powered) needed to worry about not getting sufficient voltage through a long cable and a hub.
alt Hacker News
I did a similar thing with a rack of test equipment. I had a PC power supply on each shelf, and used an NPN transistor to turn on the power supply whenever the USB port connected from the main computer to that shelf rack was turned on. Crude, but effective, out of band communication.
The local USB hub on the shelf was powered from the PC power supply on that shelf, so a single long USB cable from the shelf with the control computer could command the shelf (including the USB hub on the shelf) to fully power up or down, and then control several test devices through the USB hub, and none of the USB devices on the shelf (some of which were bus-powered) needed to worry about not getting sufficient voltage through a long cable and a hub.