Another question about power.

Most USB chargers are built to deliver around 5.5V, to maximize the amount of power available to MP3 players or phones without destroying "normal" USB devices that gets connected to them.

But the USB connectors on a PC are powered from the PSU of the PC. This PSU is not adjusted to always give a slight over-voltage. Let's say that it may give 4.75..5.25V. Then there is losses in the cables and connectors and and in any fuses or intelligent overcurrent protection. The units I work with have an intelligent overcurrent protector having a maximum Ron of 125mohm which means a potential voltage loss of 62mV at 500mA.

In the end, it's important to separate maximum current from maximum power. That is also a reason why there are several proprietary work-arounds to have a USB connector to identify that it is special and supports more current for charging. For example Blackberry are using a special solution with specific voltages on D+ and D- to recognize their charger from a normal USB connector, but they are not alone. We have quite recently got a standardized USB charger port where shorted D+/D- (max 200 ohm between them and no connection to + or GND) is used to identify a charger port from a USB host port. Alas, portable USB chargers don't know what "trick" the specific connected device needs to be able to draw more than 500mA.

Mobile phone manufacturers have been required to do a lot of magic to get their phones to be able to initially charge the battery at a really low current until the phone has enough power to boot and turn on the USB support and negotiate for 500mA. Luckilly, a number of smart charger chips are getting out on the market, that will initially take ownership of D+/D- and then on command release ownership to the processor at a later time - all just to support negotiation and optional charge at 100mA (or sometimes even trickle-charge at 0.1mA) while at the same time also be able to detect a USB charger port allowing extra-fast charging.

Another thing is that a USB device that deregister don't even have 100mA. If the USB host says "take a hike", I think the limit is 100uA - enough to be able to "taste" the D+/D- signals while waiting for a bus connect again.

One problem here is that even the standardization body takes quite lightly on overcurrents from a PC. Several documents may talk about up to 800mA from a standard connector, ignoring the fact that some PC motherboards will fail at these currents. Originally, people did decide based on 500mA and it is quite recently the USB organization have noted the large use of USB connectors as chargers and have somehow turned around. Too bad that their changed attitude is incompatible with a lot of older hardware.