I am not sure if there is a protocol for being a power only device on a USB hub. It is a good question. Here an image of the pinout from "source":http://www.connections-usa.com/usb_pinout.html
Well, First you have to consider the load and the current draw from the USB port. You first have to have to find out the specs on the corrects on the deices you are going to be running.second are you making a interface cable??..No, yes..?
It is true that the limiting power is 500mA, although many ports will expect a lower power (typically 90mA) as standard and will give the familiar 'USB power surge' message if yiu try to draw more - although it'll probably still work - it's just a bit messy.
If you are adding any sort of interface to the USB cable then the current required can be specified during the enumeration process (up to 50mmA - not sure what happens if you ask for more).
At first introduction, USB interface design can seem pretty overwhelming, but a really good (and cheap) entry method is to use one of the communication management chips (such as the FT232) from manufacturers like FTDI (http://www.ftdichip.com/index.html q.v.).
I've used these in several commercial products involving PC interfaces and have found them both easy to use and precise in their performance.
I agree with Jeff on the FTDI chips. It's surprising how many semiconductor manufacturers use them on their demo and eval boards anytime they need to accommodate a USB interface to a PC.
You're not borrowing power from USB since there's no mechanism for giving it back. The USB power line is not a storage device.
If you are just looking for a little power at 5V, then simply connecting up to the USB power will work on most desktop machines. Most desktop motherboards simply have a polyfuse between the USB power out and the internal 5V supply. The maximum a USB device can draw is 500mA, so if there is just a polyfuse it must be set to somewhat above that.
Laptops and other mobile devices, with their need to manage power more carefully, can be different. A USB device is guaranteed 100mA in normal operation. It can ask for more during enumeration, but the OS could decide it's asking for too much and shut the device down. There are also special standby modes where the device must draw considerably less than 100mA. So if you're thinking of stealing power from a laptop via USB, it might be compilcated or you don't get much without enumerating. It will also vary between laptops.
True, but by itself that statement is a bit misleading. A device is only guaranteed 100mA during normal operation unless it requests and is granted more, for example.
The USB 2.0 port is limited (as has been said) to 500mA. I had a project during university that required power over USB and I think depending on host it can provide more or less units of power, one unit being 100mA.
I suggest like Jeff did, if it also involves data transfer over USB you should use FTDI chips; their well documented and easy to use.
If you want to test how well it works with the port USB and if it's compliant I guess there are some developer tools to help with that over here (haven't used them and don't know if they're still valid but I did stumble upon them while I was researching USB myself):
This project will only involve the +5v/Gnd pins on the USB port. So, as long as the system under use does not decide to shutdown the USB port under use, then all is good. Of course, the device will be within the USB parameters.
This is good to know. Although, I was hoping for a simple passive solution to "steal" a little more power from the USB port or rather trick the OS to allocate more. Maybe version 2...
