Why Powered USB Is Going to Fail

An anonymous reader writes "Patrick McFarland, famous Free Software Magazine author, has written a two part article about why Powered USB is not taking off at home. (part 2 is also available) He includes a lengthy history on why USB took off in the first place, and then continues on to explain what we gain by allowing Powered USB to power all our devices."

Re:I agree

[bigtomrodney]

If I could pass back one comment on the article, there seems to be a few grammatical errors in there. The article itself is very interesting but poor grammar detracts from its impact.

By 2000, some computers were not shipping with hardly any legacy ports at all

That's not just a double negative, I don't think that even makes sense.

I will tell you why Powered USB will never be widely excepted

I'm sure you meant accepted there.
I apologise if I come off sounding like a grammar nazi but I find it difficult to read an article and take it seriously when it is presented in this way. That aside I would completely agree with you on the point of powered USB. Splintered standards are no standards at all.

Re:I agree

[blixel]

I also noticed this:

"Second, I suggest the USB Working Group should releases USB 3.0 already."

Assuming the word "releases" was just a simple typo and that he really intended to say "release", it still wouldn't have sounded right.

Re:I agree

[repvik]

Bzzzt! Wrong. There's no guarantee that 2.5" drives do not require more than 0.5A @ 5V. I've got four drives in front of me, and three of them require 0.7A.

Not only that, but the 2.5" drives are more expensive, slower, and has way less storage capacity.

Re:USB needs to add more firewire like stuff

[Hal_Porter]

like offloading work from the cpu as the older and slower fire wire 400 bus is faster then the usb 2 bus and it can be used to link 2 systems together with out a special cable

I don't think USB will ever support peer to peer. USB 1.0 was designed to have a smart host and dumb devices, to make sure that low end Asian manufacturers could make mice and keyboards with a couple of man-weeks of labour, probably only a few man hours once they get up to speed. Later on people started to use it for storage, and USB 2.0 was needed to up the speed. But it was never designed to do firewire type things like peer to peer networking, because Intel thought that Firewire would always do those. Theoretically, you could have two USB OTG hosts linked together, one as "device" and one as "host", but support for OTG is non existant on desktops.

But remember that USB is popular because every PC has it, and there are loads of sub $30 USB peripherals. If it had started off like Firewire, that wouldn't have happened.

Re:USB Power Strips as notebook power supplies

[AndrewNeo]

Hooking up a DVI cable to a video camera wouldn't be very size efficient. And there'd be no audio transfer, so while HDMI would be better, the Firewire camera protocol isn't just sending video like a monitor does, it's actually sending the raw video (kind of like how HDTV sends MPEG2 over the air) as well as support for bidirectional communication other than video, like playback commands and tape position, for example.

Re:USB needs to add more firewire like stuff

[sjames]

You can sort of fake peer to peer with USB. For example, a device that is 2 USB devices that look like network or serial devices back to back in a device. It's not REAL peer to peer since the host can't use that to mediate direct data transfer betwen (for example) a storage device and a printer, but it's close enough for most people.

I fully agree that it was a decent tradeoff for the bazillions of sub $30 devices that resulted from an easy to implement standard.

I just wish there was a standard for scanners rather than letting them weasel out with the vendor specific interfaces. If any mouse, keyboard, storage device, serial, parallel, etc device can work with any OS that supports USB without having to think about it or load proprietary software, scanners can do that too. The least they can do is support a fully standardized basic mode to do a full resolution color scan and let the software reduce it as needed. Ideally they should have no problem selecting a resolution, gray or color and a region using a standard interface.

Re:I agree

[that this is not und]

Nowadays a special expansion card or external device are necessary to get back the same capability. The cheapest are around $100

You're correct, but to be fair in comparision, back when those ports were popular on the IBM-PC, the parallel, serial, and joystick ports were themselves expensive add-on cards.

The PC and PC-XT had NO built in I/O. You had to plug in EVERYTHING as expansion cards, includng floppy controller card, hd controller card, serial card, parallel (which you could get built on with the MDA monochrome text-only video card), game port card, video card. None of these were built in 'on the motherboard.' On the original PC it wasn't hard to tie up all five expansion slots, since you also didn't get 640K on the motherboard, so had to plug extra RAM in _brace_yourself_ an expansion card in other I/O slot.

Thus the rise of the 'AST Six-Pack' and other multifuntion cards, which gave you seral/parallel/memory/realtime-clock/etc. all on one card (a card that cost about what people pay now for a whole system at Wal-Mart.)

Re:I agree

[jwdav]

I think you have it backwards, unless you have a PC or FireWire Card that doesn't support bus master. It is not uncommon to find marginal FireWire support on PC's, but most consumer & pro electronics, as well as all Apple products offer full Firewire support.

USB requires a host CPU; FireWire does not.

FireWire uses a "Peer-to-Peer" architecture in which the peripherals are intelligent and can negotiate bus conflicts to determine which device can best control a data transfer

USB 2.0 uses a "Master-Slave" architecture in which the computer handles all arbitration functions and dictates data flow to, from and between the attached peripherals (adding additional system overhead and resulting in slower, less-efficient data flow control)

USB 2.0

1.5 Mbit/s 12Mbit/s 480Mbit/s supported.
USB controller is required to control the bus and data transfer.
Cable up to 5 m.
Up to 127 devices supported.
Power supply to external devices is 500 mA/5V (max).
Full compatibility with USB 1.1 devices.

FireWire (IEEE1394)

100 Mbit/s to 800Mbit/s supported.
Works without control, devices communicate peer-to-peer.
Cable up to 4.5 m.
Up to 63 devices supported.
Power supply to external devices is 1.25A/12V (max.).

Powered USB will not fail

[XNormal]

Powered USB will not fail. It is already a success - in the Point Of Sale market for which it was intended.

The reason why the USB-IF did not adopt this is because of IBM's patent schticks. Without this issue they probably would have ratified the 12v version for general consumer use. The other voltages would have remained specialty items for the POS industry that normal users would never encounter. The multiple voltage versions make a lot of sense for the tightly integrated and cost-sensitive POS market.

BTW, the designers of USB were not dumb. Sure, they made compromises, but if you started with the same constraints you would have reached more-or-less the same results. Consider the fact that it is impossible to build a $2 firewire mouse.

Re:I disagree

[Stewie241]

Agreed... but I think a balance is necessary - increased power efficiency, but there are some devices and applications that need more power than the standard allows.

A typo? I have a 17" MacBook Pro that supports FW800 just fine.
Not a typo... poster was only referring to the fact that many PC manufacturers don't include add-ons that most people never use. Mac's having this hardware is the norm, for PCs, it is not. For a PC to have it, it means that it is at least gaining some common use.