USB 2 Hub

[tw]

Originally Posted by zippyt

How do you know how much total load you can put on the USB ports?

USB solved many problems with the RS-232 ports. For example USB devices can be powered from that serial port and all connections / disconnections can be performed 'hot'. All devices today (including the power supply inside all computers) must neither cause damage nor interruption of devices upstream during a short circuit. And USB addresses another problem (irrelevant to this thread) solved by differential mode signaling.

I'm doing this from memory so take them as ballpark numbers. The USB port can provide 100 ma to each port or 1 USB load. All USB devices start in this 'low power' mode. A USB device must than ask for more power while only using less than 100 ma. USB port then enables more power and USB device then shifts to a 'high power' mode. Designs / standards have already determined what is sufficient and what therefore can work. Most USB devices reviewed at IC level have two power levels - 100 ma and 500 ma - even though USB devices can ask for various intermediate power levels.

If a USB device draws too much power, then USB port typically informs (interrupts) computer which then orders power limited. Suspend mode is typically less than 1 ma. To a user, the USB port has failed with a message about a 'surge'.

Above describes a USB device powered via its USB plug. USB can be daisy chained to operate up to 128 devices. Obviously too many devices daisy chained from one USB root port might overwhelm that root port - cause excessive power consumption and shut down that port.

So we have alternatives. Each hub (that daisy chains out to other USB ports) can be self powered. A self powered hub will not draw too much power from the root port. Many self powered hubs use power bricks (wall warts).

A solution for busterb solves that wall wart problem by drawing power, instead, directly from computer's power supply. Obviously this solution also costs less money. It also puts USB ports on computer front panel where connecting and disconnecting is easiest. Another alternative is a PCI card which then puts USB connectors on a more congested IO port in back and requires software configuration (which normally - but not always - would occur automatically). Both solutions solve wall wart congestion.

Meanwhile, USB devices can also provide their own power - such as cameras. Then USB port would not have to provide high power.

USB must be designed so that user can make or break connections 'hot'. 'Hot popping' is involves special design considerations - be it a serial port or a computer PCI card. "Hot popping' permitted by USB. For example, if a device first connects or if a device requests high power mode, then a sudden inrush of current might cause overcurrent. It might even cause a computer crash - if not designed accordingly. USB design makes 'load changing while hot' irrelevant to users.

Some have upgraded their software or connected some devices only to suffer the 'power surge' message and a USB port shutdown. Well, some designs are too subjective - and therefore did not provide sufficient margin of error. Some upgraded software responded too quickly to an excessive 'inrush' current - programmers who don't learn about hardware realities. IOW the user is now at the mercy of an analog world, confused by too many without a full hardware / software grasp, blaming others, and leaving a user with few solutions.

The 'power surge' failure message is not AC electric. It is a USB subsystem doing something often traceable to a problem you can only fix by speculation and shotgunning. USB so simple that when it fails, you have few solutions. But USB is made so simple that such failures should be rare or directly traceable to a human buying only on price.

One related topic not relevant here - USB 1.1 verses USB 2.0. QED - which I also will not define.