LWN.net Logo

Advertisement

Interested in hardware, diags, validation, Linux, C, ARM, Microcode and low level programming and blazing networks?

Advertise here

The ins and outs of USB (IBM developerWorks)

The ins and outs of USB (IBM developerWorks)

Posted Apr 28, 2005 23:21 UTC (Thu) by jwb (guest, #15467)
Parent article: The ins and outs of USB (IBM developerWorks)

The article rambles. The author should have addressed the real weaknesses of USB, namely, very bad implementations, and power. USB is marred by a proliferation of fourth-rate implementations. Many USB hubs simple do not work correctly. For example, if I plug my smartcard into the powered USB ports on my Dell monitors, the card reader refuses to work. It works fine when plugged into other hubs or into the root hub. You just can't know in advance if the latest random taiwanese USB gadget will work.

Power is the other big problem. USB simply doesn't provide enough of it. IEEE1394 specifies 1.5A at 12V per port, while USB provides 500mA at 5V on certain weekdays, with a stiff tailwind, under a full moon. If you plug an iPod into a computer with a USB cable, the battery will be quickly discharged. If you use FireWire instead, the battery will be swiftly *charged*. End-users appreciate the difference. USB often fails mysteriously, either because the user plugged a current-hungry device into an unpowered hub, or because the aforementioned random taiwanese USB port is simply incapable of supplying the needed current. There exists considerable confusion in the public about powered, unpowered, bus-powered, and self-powered USB hubs.

Another considerable nuisance of USB is its total inability to provide isochronous end-to-end transmission. The interfaces -- hardware and software -- simply do not expose the needed functionality. I have recently acquired a multichannel digital audio interface supporting 24-bit, 192kHz data rates. It uses FireWire because that protocol has rich isochronous capabilities. The interface provides satisfactory low-latency, high-bandwidth performance. It couldn't have been done with USB; USB just doesn't allow for high-performance isochronous transfer.

Clearly USB is an ideal interface for cheap devices like keyboards, mice, joysticks, blinking LEDs, and so forth. But that doesn't explain the continuing existence of USB2.


(Log in to post comments)

The ins and outs of USB (IBM developerWorks)

Posted Apr 29, 2005 4:01 UTC (Fri) by beoba (guest, #16942) [Link]

It sounds like you have more issue with USB itself than with the article. Not that I disagree, few things are more irritating than attempting to plug in a USB keyboard when you can't see the socket.

Printing and scanning is a bigger problem

Posted May 6, 2005 12:24 UTC (Fri) by rwmj (guest, #5474) [Link]

I find a bigger issue is lack of standardisation for the protocols for printing and scanning.

I can plug in just about any storage device (camera, memory card, phone, ...) and it appears as "USB mass storage".

Why can't I do the same for printers and scanners? It's not as if printers and scanners differ from each other so much (rectangular page, the only major difference is the resolution). Yet I need to install special drivers for every type of printer. Certain printers (esp. HP crap) the drivers are very invasive, whatever platform you're on.

Rich.

Copyright © 2012, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds