DIY USB Extension Cables Using Cat5/6?

TheDarkRogue asks: "Ok, I have a small USB device that I need to position a distance from my computer, more then the cable, at least 30 feet. The thing is that I don't think it can go that far very well. I looked around for a solution and found a few products I could get for a price way out of my price range, so now I'm looking at this as a doit-myselfer. How can I extend the distance using Cat5/6? Does anyone know of any schematics for something or is that in its self more then is needed?" Extending a USB cable to this length (or more) can't be this easy, can it? What's the maximum length of USB cabling one can get in a single stretch?

Comments

links

[bellings]

http://www.google.com/search?q=usb+maximum+cable+l ength+specification

http://www.google.com/search?q=usb+extension+cable

Long cables

[orangesquid]

Can long cables really be this easy? Maybe...

I dunno, I needed to network two machines QUITE a distance apart, so I made a crossover UTP network cable out of an existing UTP patch cord and several hundred feet of speaker wire, as two pairs (send pair and receive pair).

Works perfectly. Blazing fast, too. I would say that I just got lucky but I do these sorts of things all the time....

You may very well be successful with the Cat5/6 option. My excessively-long ethernet cable made out of speaker wire has been functioning flawlessly for months (in fact, I built another to network hubs at vertically and horizontally opposite corners of my household, and it has also worked like a charm).

Larry Flynt, hero of the people

[orthogonal]

Ok, I have a small USB device that I need to position a distance from my computer, more then the cable, at least 30 feet.

Hot chick moved in next door, huh?

And you've found the perfect perch for your webcam, right outside her bathroom window?

And you don't want to get a wireless X-10 camera, for fear of being spammed and popped-under every day for the rest of your life?

"For a low speed device the limit is 3 meters."

[Futurepower(R)]

"For a low speed device the limit is 3 meters (9 feet 10 inches)."

From USB.ORG's USB Info: Frequently Asked Questions

In practice, manufacturer's technical support representatives often say that their devices will not work with a cable longer than 2 meters (6 feet), and they supply a cable of that length. In many office situations, a cable that short is useless. But they are right, a longer cable doesn't work, and a hub doesn't work.

Ethernet and FireWire are examples of properly designed computer communication interfaces. They are trouble free.

USB 1.0 is an example of a poorly designed and poorly implemented interface. USB often interferes with computer hibernation, to give another example.

Re:"For a low speed device the limit is 3 meters."

[evilviper]

Ethernet and FireWire are examples of properly designed computer communication interfaces. They are trouble free.

Quite right... What people don't usually realize is that FireWire is just a form of serial SCSI, meaning you should probably be saying 'Ethernet and SCSI'. The same is true of FiberChannel, Serial SCSI, and perhaps even standard parallel SCSI.

<rant>
If you think about it, manufactures NOT putting Firewire in their systems makes a good deal of sense. IDE is used rather than SCSI, so USB would be used rather than Firewire. It just makes sense.
</rant>

I have two complaints.
1. SCSI standards are too divergent. If they could just come together and make one controller that does Serial SCSI, Fiber Channel, and Firewire, all three would get quite a boost from the others.
2. Not enough devices support Firewire and Ethernet. I was looking at laser printers recently, and only the extremely hi-priced ones have NICs. (A 100Mbps Linksys NIC can be had for $5, so I don't see the problem) Perhaps more manufacturers should make their devices (printer, scanner, camera, et al) modular. That would be the best way to encourage easy interface upgrades (buy a new module, not a whole new device)

Re:"For a low speed device the limit is 3 meters."

[afidel]

The problem is that instead of just being a layout engine a printer with ethernet interface becomes an embeded pc, it now talks tcp/ip and has to convert the incoming signal(s) into document layout signals. It also has to be able to cope with multiple simultanous documents because now anyone on the network can throw a job its way.

Re:"For a low speed device the limit is 3 meters."

[sconeu]

Because SCSI is now divided into two parts -- physical and logical.

SCSI-3 is a protocol that can run on top of:

Parallel SCSI cable
Fibre Channel
FireWire
TCP/IP (iSCSI)
etc...

Re:"For a low speed device the limit is 3 meters."

[MrResistor]

I was looking at laser printers recently, and only the extremely hi-priced ones have NICs. (A 100Mbps Linksys NIC can be had for $5, so I don't see the problem)

A broadband router+4-port switch+print server can be had for under $60, so what's the big deal?

Re:"For a low speed device the limit is 3 meters."

[rakslice]

"USB often interferes with computer hibernation, to give another example."

Are you talking about the Windows 2000/ME/XP hibernation feature? It doesn't require any particular hardware support, only that the drivers are designed with it in mind so that they can save state properly.

ACPI suspend-to-disk is another matter, however.

Re:"For a low speed device the limit is 3 meters."

[cloudmaster]

I was looking at laser printers recently, and only the extremely hi-priced ones have NICs.

Get a Brother HL-1270N. They ran about $300 a year ago when I bought one, have on-board ethernet, and specifically mention linux in the setup guide. The print speed's good, the web interface is good, and the price is pretty darned good (IMHO).

Re:"For a low speed device the limit is 3 meters."

[evilviper]

The print speed's good, the web interface is good, and the price is pretty darned good

I like laser printers because the ongoing investment is much less than inkjets, not to mention that the print quality is much better (I dare you to print a full page, front and back, in fontsize 2 text on an inkjet). With ink, you know you're getting ripped off in any number of ways (which is why you don't see an estimated page count on the packages). With toner, you used to be safe, but lower cost lasers have led to toner scams.

For instance, at work, one printer I maintain uses a $150 toner cartridge that prints tens of thousands of pages. This isnot mentioning that, with several toner-saving features, that is stretched into many times more.

My home laser printer, uses a toner cartridge that costs about the same, yet only prints about 5,000 pages...

When I am looking for a new printer, I could care less about the interface, and very little about the speed... My biggest concern is how many pages I'll get with the toner cartridge, per-dollar. The printer you mention doesn't do much better than my own. I'd be willing to pay a load of money up front, if necesary, to minimize the ongoing cost of tones, (and the drum for that matter).

Unfortunately, it's fairly rare that a printer or toner cartridge is listed together with it's page count, so it's hard to search for. If you've got a suggestion for a good printer under THAT criteria, I'm all ears.

Re:"For a low speed device the limit is 3 meters."

[evilviper]

That still gets you 12,000 pages for your $150 figure - but this printer's a "desktop" class machine.

I have been looking at printers lately. After comparing the prices for lower-end laser printers & toner, I've noticed that it will take (at best) around 20,000 pages before the lower priced toner makes up for the price of the machine (and that's not counting the cost of drums). So, I'm not going to half-ass this thing, I'm going to look around for some time, and go for the gold.

I do have a hard time believing "tens of thousands" from a single toner cart, but whatever.

Kyocera is a brand I've been looking at closely. The toner cart for the FS1700 laser printer costs $93, and yeilds 20,000 pages.

get a used Tektronix (now Xerox) Phaser 850. It uses wax blocks

I'd be willing to spend the money, and I have heard great reports about wax printers from at least a dozen people already. However, I'm very skeptical of any offer like that (e.g. free for life). There is ALWAYS a very large catch. If you can direct me to some web page with the offer, I'd be happy to read more about it.

Re:"For a low speed device the limit is 3 meters."

[evilviper]

Appreciate the info, but I've found a bit more on my own:

It seems you can order a 3-pack of ink sticks for $0... If true, what ordering limit do they impose? If 'none', they would be broke by now.

Their Phaser 860 is available for $2,000, with 'Free' Ink sticks as well.

Their Phaser 8200 sells for $500 less, without the 'Free Black Ink' deal. (but read below)


The 'Free Black ink' is expensive!

For the 860: '2 cyan + 1 black Free' prints 2,800 pages, and costs $92.

For the 8200, '2 cyan' prints 2,800 pages and costs $80.

Is it just me, or does it sound like you are paying $12 for the '+ 1 black Free' ink stick?

You can verify those prices here: http://www.officeprinting.xerox.com/perl-bin/produ ct.pl?product=860&page=supl#<b>colorstix®_solid_in k</b>

And here: http://www.officeprinting.xerox.com/cgi-bin/produc t.pl?product=8200&page=supl&cache=miss#<b>colorsti x®_solid_ink</b>

I've looked at other printers they've got for sale as well. Their Phaser 1235 is $2,700, but the black toner ends up only costing $0.0097/page. It would take some time to make up the printer's cost, but it would be a good deal if I get a refub. In any event, I've requested additional information from Xerox and might end up buying one of those three printers. So, I appreciate the information, but the 'Free Black Ink' deal continues to look like snake-oil to me.

Profit!

[Karma Farmer]

Here's a plan!

1. Cut ends off existing USB cable, and solder to longer cable.
2. ...
3. Profit!

But seriously -- what exactly are you asking? If want to know if a longer cable will work, then just solder up a longer cable and test it. If you want to know if twisted pair will help with electrical interference, or what wires to choose to run to which pins to give the greatest benefit from the twisted pair, then ask that instead. If you're looking for inexpensive passive or active extensions, ask that. If you want to know where to find cheap cable ends from a reliable vendor, ask that. If you want to build your own USB repeater (or whatever they're called), ask that. But I can't figure out what your current question is. If your question is "How can I extend the cable length using a longer cable," then the answer might be, umm... maybe you could use a longer cable?

viola!

[Detritus]

cello!

Probably isn't going to work.

[CMiYC]

You can't just go hacking into a USB cable and make it longer. Probably the #1 reason is that USB uses a differential signal. That means that however you extend the cable you have to do it exactly the same on both D+ and D-. If you don't, part of your diff signal gets skewed and becomes corrupted at the receiver. Secondly you have to maintain an imepedance match. Anytime you introduce an impedance mismatch you generate reflections. The wire used in a USB cable has a certain about of reactiance (X). The X of a twisted wire pair is completely different.

In order to do what you want, you really need to build a fairly preceision amplifier/receiver setup. The difficulty is the fact that you are dealing with differential signals. Each "side" of the diff signal must be treated the same way. It is almost impossible for you to make the exact same size solder joints, exact same size wires, and have the exact same impedances throughout. Its especially important on long runs because of the highly immentant signal attenuation.

Re:Probably isn't going to work.

[rcw-home]

Ethernet is also a differential signal - signal skew is simply not that big of a deal, especially at USB speeds. Light travels a long ways in one clock cycle.

Impedance also isn't a big deal. According to this, it's supposed to be 90 ohms, and the spec allows for 15% variation. Cat5 is about 100 ohms.

What is important is timing. According to this, the low-speed timing limitation is 18ns, or 300000000*.000000018 = 5.4 meters. Significantly under 30 feet.

Size of solder joints... sheesh.

Re:Probably isn't going to work.

[CMiYC]

Yes size of solder joints will matter. The impedance mismatch of the solder joint can be enough to generate reflections that will destory the signal.

Yes skew if a very big deal for differential signals of any speed. Sure maybe adding an inch on a USB cable won't hurt, but more than 3 inches will.

Impedance mismatches will be the largest problem of them all.

These are all issues if you do not know that they can cause problems. If you've never soldered something, don't know what a differential signal is, or do not know anything about tranmission lines, these are all big issues. From the sound of the question, my guess is the person asking doesn't know (and maybe not understand) these issues. If you do, then no, they may not be a big problem.

Re:Probably isn't going to work.

[rcw-home]

Yes skew if a very big deal for differential signals of any speed. Sure maybe adding an inch on a USB cable won't hurt, but more than 3 inches will.

The wavelength of a 24mhz signal (f.e. USB) is 12.5 meters. Let's round 3 inches up to .1 meters - it's .8% of our wavelength, or 2.88 degrees. 1 - cos(2.88) = .0012, or .12% of RMS. Have you been living in wonder at how people make a 250mhz signal go across these lines? (gigabit ethernet)

These are all issues if you do not know that they can cause problems.

Ahh, so that's why having a technician look at a computer problem instantly makes it unreproducible.

You'd be surprisedwhat you can do

[brink]

Stranger things have been done. Once my brother-in-law and I were SOL on finding a serial-to-rj45 cable for the management port of a router. All we could find was a cd audio cable and a cat 5 patch cable, so we stripped one end of the cat 5 and plugged it into the pinholes of the audio cable, then pulled the wires out of the plastic on the other end of the audio cable and just shoved them onto the serial port pins. Worked like a charm!

'Course, it probably helped that the total length was only like 2 feet, but still.

Re:You'd be surprisedwhat you can do

[Jonny 290]

I'll agree. Back when I was a young sprite, I brought my computer to a friend's house and we set up in the shed out back. Didn't have a phone line, so I spliced together speaker wire, 300 ohm twinlead, 75 ohm coax, and household zipcord to make a two-conductor cable long enough to get a POTS line out there.

I never broke 26.4k connects at home normally, but through that line I got 51k to 53k connects all night long.

Electrical Limitations

[SagSaw]

From The USB FAQ

Q3: Why can't I use a cable longer than 3 or 5m?
A3: USB's electrical design doesn't allow it. When USB was designed, a decision was made to handle the propagation of electromagnetic fields on USB data lines in a way that limited the maximum length of a USB cable to something in the range of 4m. This method has a number of advantages and, since USB is intended for a desktop environment, the range limitations were deemed acceptable. If you're familiar with transmission line theory and want more detail on this topic, take a look at the USB signals section of the developers FAQ.

Essentially, the USB cable is a mismatched transmission line. As a result, the signal reflects off of the ends of the cable, resulting in a stair-step pattern whenever the voltage level on the line changes. As the cable gets longer, it takes more time for this stair-step pattern to settle to the final voltage.

Use an active USB extension

[no_such_user]

I read this story earlier and just happened across this device a minute ago:

16 ft. USB Active Extension Cable

I have ZERO relation to this company (though I have bought from them before) and have ZERO experience with the product. But it looks like what you're after. I'm sure there are other vendors who supply this device as well.

Good luck!

Re:Is 80 feet enough?

[geoswan]

As an anonymous coward has pointed out this is a mis-reading of the KDS page. To get the 80 foot extension you have to connect 5 x 16 foot extension cables. I am repeating the anonymous coward's observation as I have a higher karma rating.

Depends... but yeah.

[Mike1024]

Hey,

I wanted to play about with an extended USB cable, so I could have a webcam at my door. Here's what I did:

1. Get a short USB cable.
2. Cut it in the middle.
3. Get some STP Cat5 cable (I used this because I had spare cable in work).
4. Splice the Cat5 into the middle of the USB cable (Note: If you don't have much electronics experience, it might be worth employing the help of a friend).
5. Bind the joints in insulating tape.
6. Light two black candles, and sacrafice a live chicken.
7. Test cable.

Of course, I only wanted a cable for a cheap USB webcam, and it was the only thing on the bus. The specification calls for much shorter cables, and that's presumably for a reason. Your mileage may, or course, vary.

I've had success though. Good luck.

Michael