Cat 5 network cable for balanced audio and power?

[miallen]

Hi,

I've been toying with the idea of using standard computer network cables for wiring small audio devices. These cables are laying around all over the place, you can buy the sockets at home depot [1], their shielded (usually) and are composed of 4 twisted pairs which means you could use balanced signals.

At the moment I'm just thinking about using them just to deliver DC power to things like opamps that need dual supply but before I decide on a wiring scheme I thought I would ask if anyone has considered sending signals over these cables. I know people have bundled them to make speaker cables but I'm thinking about using individual pairs to send analog audio.

So my question is ultimately, how would you get the best SNR from a Cat 5 networking cable twisted pair? Would you send the signal at line-level? Would you use the pair to send balanced (inverse phase) signal? Would stranded wires be preferred (not sure if Cat 5 comes with stranded wires).

So far the wiring scheme I was thinking of is the following (the letters indicate the twisted pairs):

1 A Signal A
2 A Signal A
3 C +15VDC
4 D 0V
5 D +9VDC
6 C -15VDC
7 B Signal B
8 B Signal B

So this offers +9 VDC, +-15 VDC, +30 VDC and two channels of balanced signal.

I'm thinking this would be great for guitar pedals because you can have one connection for everything including power and send and return signals.

What do you think? Is this idea nuts or what?

Mike

[1] The RJ45 wall jacks neatly clip into a rectangular hole and if you know how to take them apart you can reduce the internal footprint to less than a centimeter within the enclosure.

[panomaniac]

Cat5 does come in stranded. Most "jumpers" are stranded. That's what you'd want, the solid core stuff does not last long if flexed a lot.

Sounds like it would work, as long as current draw is low.

[eclectic2k]

yes, most ethernet "patch" cables are stranded. "Infrastructure" cabling (bulk) is normally solid. Solid has better high frequency specs, but either will work fine for this app...

Most cat 5 is UTP. *unshielded* - Shielded (STP) is available, but much less common. Shielding is normally not needed as the balanced signaling and the twisting gives it fine noise rejection. (each pair is twisted at a different rate as well)

If the audio signal is balanced you should be OK just running it down one pair like you suggest. If it were not balanced it'll pick up noise for sure, and may still pick up some noise from the other pairs anyway...

Jan had a short thread over in Digital about AES & Power over Cat5 with some useful comments maybe:
AES over Cat5e?

oh, and if you buy pre-terminated cables, be mindful that they are pinned out such that: pins 1/2, 3/6, 4/5, 7/8 are the twisted pairs. (looks like you may know this - shouldn't matter by your planned pin out.)

[Atilla]

Power is often provided over the free pairs in a Cat5 cable for network components - routers, switches, bridges and whatnot. You need to be careful with the current though, it can't go too high, as ethernet wires are quite thin.

If you're sending balanced signals over STP cable you won't have any problems, it's as good as it gets.

I've got a huge supply of cat5 cables and I use them virtually everywhere, they're fantastic. Solid-core ones are more rare to find, since it's mostly used in structural cabling. It's hard to bend and easy to snap, so I'm not sure if it's worth it.

[AndrewT]

Hi,
if you require flexible 4pair, then PATCH cables are available in lengths from ~0.5m to 3m.

If you are planning a fixed installation, then go for the solid core version, available off the roll of 1000feet.

If you use the spare cores to deliver PSU then plan to use regulators at the remote end to lower the impedance seen by the circuits/amplifiers/opamps etc.

[SY]

FWIW, I use CAT5 for balanced phono cable. The tight twist is very useful.

[miallen]

Originally posted by AndrewT
If you use the spare cores to deliver PSU then plan to use regulators at the remote end to lower the impedance seen by the circuits/amplifiers/opamps etc.

Hi Andrew,

Is this really necessary? If the power going over the cable is DC then does impedance really factor into anything? I'm not an EE and I'm still struggling with the whole idea of impedance but I thought impedance was pretty much an AC phenomenon.

Otherwise, what chips do you recommend for +9, +15 and -15 respectively (power would probably be very low)?

Actually this leads me to another question: If I have a distribution box that supplies power to multiple devices, should each output use it's own voltage regulator? If I use the output of one regulator to power to multiple circuits, couldn't I get cross talk?

Or perhaps this is what you meant - I should just use one regulator for all outputs and one on each of the remote inputs?

Mmm, dual voltage regulators - that will probably have low ripple!

Mike

[coloradosound]

Keep the twisted sets together for each signal!
(to avoid crosstalking to other pairs)

[panomaniac]

CAT5 seems to be the "miracle cable." I thought about opening a thread with this title. We've got a good start here.

[Atilla]

Originally posted by miallen


Is this really necessary? If the power going over the cable is DC then does impedance really factor into anything? I'm not an EE and I'm still struggling with the whole idea of impedance but I thought impedance was pretty much an AC phenomenon.

Otherwise, what chips do you recommend for +9, +15 and -15 respectively (power would probably be very low)?

Actually this leads me to another question: If I have a distribution box that supplies power to multiple devices, should each output use it's own voltage regulator? If I use the output of one regulator to power to multiple circuits, couldn't I get cross talk?

Or perhaps this is what you meant - I should just use one regulator for all outputs and one on each of the remote inputs?

Mmm, dual voltage regulators - that will probably have low ripple!

Mike

To a certain extent you could look at regulators as buffers, I guess. If you're running power over a long cable, which has stray capacitance and inductance, it can do you good. Providing decoupling caps and regulation at the end of the cable will definitely improve the power supply you've constructed, especially if you've got long cable runs.

I don't know what "very low" would mean, but national's 317 and 337 should be able to deal with currents of about 1.5A. You'll need heatsinks and the current output can be enhanced by using additional transistors. At any rate, those ICs are among the most versatile devices made, I love them!


If you get output for multiple devices from the same supply, you should consider providing decoupling caps for all of them, that way you'll minimize any interaction between them. The next step would be to combine that with some additional isolation - a simple JFET like in Tangent's PPA, or a voltage regulator.

Dual regulation is nice, but do keep in mind you'll need to drop more voltage. For LM317/337 to be effective, you'll need to drop a good amount of voltage, which would double with double regulation. If you want that, maybe LDO chips would be a better idea...

I don't think you can really get it wrong though, it's just how good you can make it