My backup device is a USB 2.0 250 MB Iomega Hard Drive.I'd like to put
it in my garage, requiring about 20 to 25 feet of wire. The official
USB specs say 5 meters max; if further use a hub/repeater.

I'm wondering how realistic the 5 meter spec is. I recall that parallel
printer cables were supposed to be limited to 10 feet or so; 75 foot
cables worked just fine. Same thing with the specs for RS-232 cables.

Does anyone have any real world experience with long USB 2.0 cables?
I'm thinking of building mine using unshielded CAT5 cable, one twisted
pair for power, another for data. Wise? foolish?

As I understand it, most usb drives are powered by the usb socket on computer. Hence there will be a considerable loss of signal and drive voltage if the cable is too long, hence suggestion of hub/repeater.

Fingees is right. "Signal strength" is not the real issue in this case - 5Vdc is. As in the past, MUCH depends on the quality of the connectors and cable - and I personally feel quality control was much better back in the old days.

Self-powered hubs are cheap - don't you be!

The good news is wireless usb is on its way. At 3 meter, full 480 mb/s. At 10 meter 110 mb/s. Which by the way, in your case, the more you add to the distance by lengthening your cable as you described, the lower your speed will get. Yes, RS232 and all of them work with cable length longer than specified, but performace starts degerading. And it all deponds on how much outside intefrence get into the cable. In some cases, you won't notice anything adding a few feet, in some other cases adding the same amount will make a big difference.

http://www.usb.org/home/

http://www.usb.org/developers/wusb/

Interesting.

The "active" cable below claims that the USB signal can be extended by as much as 80 feet via connecting five 16-foot active cables in series. Although, this is going to be much more expensive than just buying the powered hub.

http://www.usbgear.com/USB2_extensio...ion-cable.html

WIRE GAUGE # 20: 0.1015 OHMS OF RESISTANCE PER 10 FT.
WIRE GAUGE # 24: 0.2567 OHMS OF RESISTANCE PER 10 FT.
WIRE GAUGE # 28: 0.6490 OHMS OF RESISTANCE PER 10 FT.

The power supply level out of the USB port is nominally 5V dc, but allowances for voltage drops means that the device must return device/configuration data down to at least 4.4V dc. The normal low level for full operation is 4.75V. The USB specification allows wire gauges from 28 to 20 AWG to compensate for losses.

Looking at the specification in the link above for the active cable, the power groundwires use no.20, the data wires are no. 28. Toto25 wants to use cat5 cable. Cat5 is twisted no. 24. As far as voltage drop is concerned, no. 24 is more than 2.5 times better (has less loss) than no. 28 wire, which is about the same factor in terms of distance that Toto5 wants to add to the USB cable length. (5=meter is equal to almost 16.5’-and distance to his garage is almost twice that). I say, use no. 20 wire for power, and no. 24 for signal. The signal wire must be well twisted to minimize the build up of any common mode differential voltage, in order to reduce built up of noise and interference.