Douglas A. Tutty wrote:
Then how does my USB light get power; surely it doesn't "request" enough power to run an LED? I can also charge my Palm with the computer off; there's nothing running to receive any "request".
The USB spec allows a certain amount of power to be sent (below 200 mA I believe - I forget right now) at all times, and if a device needs more power it's *supposed* to request it.
Different laptops and chipsets implement power delivery to USB devices differently, but generally devices like lights and other small things don't bother to "ask" for power, they just use what they can get from the default power pin at the default (low) current draw level from the laptop/USB host.
Larger devices (like iPods) would have to "ask" for power before they would "connect" their power bus to the USB, because 200 mA won't be enough. That would be implemented in the iPod itself, not the laptop. The designer would know if his device could run properly at the 200 mA level, and if not, the device wouldn't attempt to charge or connect to the USB power bus until it had done a "handshake" with the laptop/USB host. If the USB driver is down on the host, and no handshake can happen, no power will flow.
So it's all device and chipset implementation dependent. There are devices that "misbehave" and pull more than their fair share of power. Some laptops put up with this, other's don't and cut off USB power to any device drawing more than its requested current level.
The iPod has pins on the bottom connector for different power sources, if I remember correctly (Apple's not too forthcoming with the connector specification, since it's proprietary and they license manufacturers who use/make iPod accessories) where one is meant to be power from a computer source, and the other pin is meant for things like car power cables, airplane power cables, etc.
Then the iPod itself decides what it can do depending on where it's seeing input voltage.
It also might be possible to trick the iPod by plugging it into a powered USB hub and then connecting that to the laptop, instead of directly to the laptop. I've never tried, but thinking about it -- that could exhibit different behavior than directly-connected.
Nate