Ethernet cable (CAT 5?) questions
Mar 5, 2006 at 12:52 AM Thread Starter Post #1 of 10

Zorander

Headphoneus Supremus
Joined
May 14, 2004
Posts
5,493
Likes
36
Location
Sydney, Australia
I've read some mentions about making interconnects out of these cables. I happen to have one run lying around uselessly so I guess it's off to my 'butcher's table' it goes. Upon stripping one end, I saw that it has 8 separate cores/conductors. Since I'm trying to make a mini-to-mini, how should I do it? Do I take three of the conductors for each channel, and the remaining two as the ground (assuming the plug can fit all of it - which I doubt)? Or do I just make do with one core for left/right/ground, leaving the other five cores unused?

Cheers!
 
Mar 5, 2006 at 2:55 AM Post #2 of 10
I just built a mini-mini from cat5 about a month ago. I took the wires out of the bigger cable. The took 3 of the small wires, and used those for left, right, and ground. I also braided them. You could do 4 wires and do a litz braid, but then you'd take 1 to left, 1 to right and 2 to ground.

I have heard reports that just using cat5 as-is, doesn't produce great results, but I haven't tried it. I assume you're making a short cable so the extra steps of braiding shouldn't be that much of a chore.

For mysetup, I also used expandable sleeving, and heatshrink on both ends to finish it off.

-Jeff
 
Mar 5, 2006 at 3:56 AM Post #3 of 10
I'm planning to make a long run (over 3m) so I'll be stuck with using the cable as is. Just wondering though what configurations you all use when converting a CAT-5 cable into ICs.

Talk about Litz Braiding, is there a site that gives a clear and not-too-confusing explanation on how it can be achieved? One site I stumbled onto (forgotten the link) has a very technical explanation on Litz braiding and it might as well they wrote the whole page in Greek since they lost me so easily.
smily_headphones1.gif


Thanks!
 
Mar 5, 2006 at 8:09 AM Post #4 of 10
Quote:

Originally Posted by Zorander
I'm planning to make a long run (over 3m) so I'll be stuck with using the cable as is. Just wondering though what configurations you all use when converting a CAT-5 cable into ICs.

Talk about Litz Braiding, is there a site that gives a clear and not-too-confusing explanation on how it can be achieved? One site I stumbled onto (forgotten the link) has a very technical explanation on Litz braiding and it might as well they wrote the whole page in Greek since they lost me so easily.
smily_headphones1.gif


Thanks!



Well, Litz braiding's effects are mostly in the RF frequencies which are waaaay out of the human audible range. It's not necessary for interconnects, save for the fact that it keeps the wires organized. Ask some of the actual engineers on the board if the parasitic capacitance of the braid could possibly produce non-linear distortion with wires that thin. I really have no idea myself.
 
Mar 5, 2006 at 1:57 PM Post #5 of 10
Quote:

Originally Posted by Twombly
Well, Litz braiding's effects are mostly in the RF frequencies which are waaaay out of the human audible range. It's not necessary for interconnects, save for the fact that it keeps the wires organized. Ask some of the actual engineers on the board if the parasitic capacitance of the braid could possibly produce non-linear distortion with wires that thin. I really have no idea myself.


OK, lets try to steer clear of some potentially harsh debates on the effects of Litz Braiding, and get back to the original question.

I found Chimera Labs to be the best Litz Braiding tutorial. You practically don't even have to read the instructions, just follow the picture along with the correct wires they state and you should be Litz'ing in no time. http://www.chimeralabs.com/diy_braid.html
 
Mar 5, 2006 at 7:27 PM Post #6 of 10
Quote:

Originally Posted by Zorander
I'm planning to make a long run (over 3m) so I'll be stuck with using the cable as is. Just wondering though what configurations you all use when converting a CAT-5 cable into ICs.


In TP CAT5 cable, each colored wire is a TP (twisted pair) to the wire with that color stripe. Given the length and that you want the wire whole, I would use all 8: two colors (wires) for L, & R channel, the twisted pair wire for each of those being a ground return.
 
Mar 5, 2006 at 7:32 PM Post #7 of 10
Sadly the Chimera Labs (why are these places always "labs"?) tutorial doesn't actually create a real Litz braid. But then again most audio places don't either. Although the tutorial does touch on the history, it does fail to understand what is going on, and has the usual slab of pseudo-science and mumbo-jumbo.

Bottom line, you can build your interconnects out of Cat-5. This cable is made of four twisted pairs. (As is any Cat-3,4,5,6.) The pairs are designed to reduce cross coupling and are used in differential signalling.

You could choose a number of possibilities, and run the range of possible trade-offs.

One, you made the classic mistake of implicitly assuming that somehow the "signal" part of the cable is more important, and that the ground, is somehow less so. Ain't so. It is simply a matter of convention in what we call signal and ground. The electrical energy doesn't know you called what wire what, and it will obey the usual conservation laws and travel in each equally.

So, you have a few choices.

For a stereo interconnect you need essentially four conductors, but since the plug combines the grounds, it becomes three at each end. But you should always recognise the send return nature of the circuit for each channel.

But choices are to use both wires in a pair, solder them together at each end, and call it a single conductor (the audiofools might call this two wire Litz). You lose the inherent noise immunity offered by the twisted pairs, but reduce the capacitance.

Alternatively, take one wire from each pair, combine these together as the ground, take two of the remaining conductors, combine these as one channel, the remaining two the other. This provides the best noise immunity at the cost of the highest capacitance.

Not recognising the way in which Cat-x cable is built as four twisted pairs, and treating it as 8 separate conductors will simply result in close to random results.

If you have a very high impedance circuit you might just be able to hear the difference in cable capacitance in changes to the high frequencies, but with most modern designs you would be hard pressed to do so.

Consider that Cat-6 cable is capable of sending Gigabit Ethernet over distances of 100 plus metres. Gig-E needs a bandwidth of close to 10GHz in order to transmit this. Even Cat-5 reticulating 100Mb/s Ethernet needs 1GHz bandwidth, and can run for much longer than 100 metres.
 
Mar 5, 2006 at 8:22 PM Post #8 of 10
Quote:

Originally Posted by firefox360
OK, lets try to steer clear of some potentially harsh debates on the effects of Litz Braiding, and get back to the original question.

I found Chimera Labs to be the best Litz Braiding tutorial. You practically don't even have to read the instructions, just follow the picture along with the correct wires they state and you should be Litz'ing in no time. http://www.chimeralabs.com/diy_braid.html



I know I've been confrontational about the subject of cable tweaks in the past (and I seem to be doing it again), but braiding several tiny CAT5 wires is something of a time-consuming undertaking and I thought it would be prudent to broach the subject of cost versus benefit in this case. What benefit entails here can include other purely qualitative factors such as aesthetics and sense of accomplishment; however, when speaking in terms of a quantifiable effect or lack thereof on the transmitted signal, it would be a good idea to at least consider the opinion of an establishment before going through with this project.

I'm not forcing you to agree with me, nor am I forcing the original poster NOT to braid his wires. I can't somehow snatch away the human element of choice. That's yours alone.
 
Apr 2, 2006 at 1:20 AM Post #9 of 10
I decided to use only 2 of the pairs (4 conductors in total), leaving the other 2 pairs unused, for the mini-to-RCA interconnect I made. On the mini-plug side, I combined the 2 white wires together as ground and each 1 of the coloured wire as signal for each channel. On each RCA end, it's 1 white wire as ground and 1 coloured wire as signal (both wires belonging to the same twisted pair). What implications, noise-rejection and capacitance-wise, does this setup bring? It sounds good to me, accurate-sounding with a dry presentation coming to my mind.

My next mini-project is to make speaker cables out of these cables. Has anyone done this before? Are there certain techniques to achieving this other than simply twisting all coloured wires together for one end and all white wires as the other end?

Cheers!
 
Apr 2, 2006 at 4:26 PM Post #10 of 10
Aside from all the technical issues (and mumbu jumbo if you will)....

I made a cat5 interconnect and I didn't like it simply because I think the wires are very fragile. Especially for a portable cable or some other use where the cable is flexed a lot, it just seemed to me that those tiny solid core wires would tend to break. Since they are no fun to fit into small connectors, I would not want to have to revisit a connector with a broken wire.

I am curious if anyone has used cat5 and had a problem with broken wires over time.

Personally I just bought some 4 conductor starquad and use that, simply because it is easier to deal with and flexes better.
 

Users who are viewing this thread

Back
Top