Sending Decent Quality Audio (stereo)

[Black Hole]

OK folks, anybody with a technical bent or relevant experience welcome to pitch in:

Problem:

Currently have a mixer balanced-line stereo output sending to a pair of self-powered (built-in amp) speakers over about 10m (each) balanced line XLR. I want to move the mixing desk to the far end of the hall so the sound man (me) gets a better impression of the sound balance (all the sound sources are local to the mixer, so there is no issue with mic & instrument runs or anything like that). Obviously that could be done with a pair of 50m XLR leads, but have you seen the cost of those??? And they would need to be routed out of the way, so maybe 80 or 100m is more realistic.

Ideas:

What about Cat5e UTP? Four pairs, and I have loads of that. I have seen unbalanced audio to Cat5e passive balun adapters, but surely the noise pick-up would be awful, not to mention the bandwidth?

What about using the Cat5e as a point-to point wired Ethernet network, and actually sending the audio digitally? The problem is that the mixer is analogue so the audio would have to be digitised, transmitted, and converted back to analogue... but maybe that wouldn't be too bad or expensive (see TOSLink adapters/converters). The issue here is availability of (cheap) equipment, TOSLink itself might not be a bad idea, but it is not designed for those distances and cheap fibre is plastic with a limited length due to its poor transmission.

Wireless? Licence-free bands are very noisy, and with wireless there is always a risk of something causing a break in reception which is absolutely not acceptable in performance.

Perhaps there is an existing solution to this problem I have not come across. Thinking it out like this, I have nothing to lose trying out sending two balanced signals over two pairs in a length of UTP. I don't expect the result to be wonderful, but you never know...

 

[prpr]

What about Cat5e UTP? Four pairs, and I have loads of that. I have seen unbalanced audio to Cat5e passive balun adapters, but surely the noise pick-up would be awful, not to mention the bandwidth?

Why would they be awful? The whole point of balanced audio is that it is immune to the common-mode interference. You need a lot more than 100m before you need to worry about equalisation (off the top of my head... haven't needed to do this for a long time!).

What about using the Cat5e as a point-to point wired Ethernet network, and actually sending the audio digitally? The problem is that the mixer is analogue so the audio would have to be digitised, transmitted, and converted back to analogue... but maybe that wouldn't be too bad or expensive (see TOSLink adapters/converters). The issue here is availability of (cheap) equipment, TOSLink itself might not be a bad idea, but it is not designed for those distances and cheap fibre is plastic with a limited length due to its poor transmission.

You could use pro. stuff such as Sonifex PS-Send/PS-Play but that's going to be over £1500 for the pair.

Wireless? Licence-free bands are very noisy, and with wireless there is always a risk of something causing a break in reception which is absolutely not acceptable in performance.

Yes, non-starter really (IMHO).

I have nothing to lose trying out sending two balanced signals over two pairs in a length of UTP. I don't expect the result to be wonderful, but you never know...

I think you might be surprised.

 

[Black Hole]

I think you might be surprised.

I do hope so, your confirmation is good news!

Yes, I understand the theory of balanced lines, and a hall I used to use had a built-in multi-core trunk cable from the stage to the control box, but no doubt that was good quality cable and screened whereas I'm taking about crap unscreened cable (albeit twisted pair), and I did think there might be some differential pick-up and high-frequency roll-off.

Anyway, I will get around to a test soon.

Even if the UTP isn't very good, the best option seems to be to bite the bullet and shell out for a length of good cable.

 

[Trev]

HF roll off????? How fast Ethernet will it support? 1 GHz? That's a tad higher than the 20kHz or so needed by audio.

 

[Black Hole]

HF roll off????? How fast Ethernet will it support? 1 GHz? That's a tad higher than the 20kHz or so needed by audio.

Sure, but digital transmission doesn't require an even frequency response. I'm not doubting you but I have no thumb for this (being a digital engineer not an audio engineer).

I have seen unbalanced audio to Cat5e passive balun adapters, but surely the noise pick-up would be awful, not to mention the bandwidth?

Why would they be awful?

Well, I couldn't see how a passive adapter could turn an unbalanced signal into a balanced one, but I suppose if the adapters contain transformers... (clearly the cheap ones don't)

 

[Trev]

Fed into an el cheapo differential amplifier would 'do' bal-unbal without a transformer. (but you would need a DC supply)
But if it had an HR rolloff at (say) 50kHz (more than enough for audio unless you start messing with sub-carriers). How would you get the reasonably sharp edges required for of 100GHz

 

[prpr]

a hall I used to use had a built-in multi-core trunk cable from the stage to the control box, but no doubt that was good quality cable and screened

Probably star-quad if it was used for mic. level stuff. You can get away with a lot more at line level.
You could also try using a pair for each wire on your XLRs, and thus use all 4 pairs for stereo, a bit like star-quad.
Pro. analogue audio was always wired up with stuff like this!

I couldn't see how a passive adapter could turn an unbalanced signal into a balanced one, but I suppose if the adapters contain transformers...

Exactly. Rep. (repeating, I think) coils. Just a 1:1 transformer.
Analogue audio used to use bucket-fuls of the things.

 

[Black Hole]

Pro. analogue audio was always wired up with stuff like this!

What, unscreened?

You could also try using a pair for each wire on your XLRs, and thus use all 4 pairs for stereo, a bit like star-quad.

Rely on the twisted pairs being twisted around each other? It's an idea. I was going to use a spare pair for the common pin (XLR has three conductors and an optional screen).

 

[prpr]

What, unscreened?

Totally. As I said before, you're relying on the balance to cancel out the rubbish, not the screen.

Rely on the twisted pairs being twisted around each other?

Yes. It may help, it may not. I've never tried it.

 

[Black Hole]

As I said before, you're relying on the balance to cancel out the rubbish, not the screen.

Sure, but it's gotta help if you reduce the rubbish before cancelling it. Too much will exceed the common mode capabilities of the input amp.

Anyway, that works well enough to run with (1 pair +/-L, 1 pair GNDL, 1 pair +/-R, 1 pair GNDR). After careful consideration I have settled for 50m.

Even better: if, instead of wiring XLRs directly onto Cat5 I make a couple of break-out boxes with cables to XLR on one side and RJ45 sockets on the other, I can fit standard RJ45 onto the length of cable and then it will dual-purpose as a long-reach Ethernet when I want.

 

[prpr]

Even better: if, instead of wiring XLRs directly onto Cat5 I make a couple of break-out boxes with cables to XLR on one side and RJ45 sockets on the other, I can fit standard RJ45 onto the length of cable and then it will dual-purpose as a long-reach Ethernet when I want.

Spoken like a true engineer. And what I would do too!

 

[rpb424]

Pro. analogue audio was always wired up with stuff like this!

Indeed, and still is. I've wired up literally thousands of metres of this between IDC ('Krone') blocks (often multi-metre runs between blocks on larger frames) over the last 30 years. And my wiring went to air on BBC1/2, so can't have been too shabby!

 

[Black Hole]

Initial experiments were encouraging, but (while I wait for parts for the adapters) I have discovered that although my mixer has balanced inputs it doesn't have balanced outputs! Therefore the initial experiments were only pseudo-balanced (and remarkably good considering).

My injection adapter will now include audio transformers!

 

[Black Hole]

Bingo, working well.

In case anyone's interested, here's what I did. The input jacks are arranged so that a balanced line input simply connects + and - across the transformer input (tip and ring, plus 0V on sleeve), whereas an unbalanced line input (signal on tip and 0V on sleeve) applies the signal across the transformer input because the ring contact will short to sleeve.

The audio transformers are there to create proper balanced signals to send down the long twisted pairs (Ethernet cable). Might be overkill, but theoretically just launching an unbalanced signal into the twisted pair can't achieve full noise rejection because the signal has a different impedance from the 0V connection. The downside is that the transformers themselves could couple ambient magnetic hum into the line.

 

[prpr]

Might be overkill, but theoretically just launching an unbalanced signal into the twisted pair can't achieve full noise rejection

Not at all overkill. It's really important to use a balanced source to maintain balance down the line. You also need a balanced destination of course otherwise the same thing happens.
JOOI, what transformers did you use?

Bingo, working well.

 

[Black Hole]

Not at all overkill.

I got decent results in initial experiments without the transformers (and an unbalanced source).

what transformers did you use?

EL14

 

[Black Hole]

Went live last night, with no noticeable impact on sound quality and any pickup (there was just a very little) well below the perceptible threshold in the performance environment.

There was a temporary problem with buzz in one channel, but reseating a jack fixed that. It seems the sockets I have used for a quick build are not as reliable as I would like, but now the concept is proven I don't mind redoing it with more expensive parts and over a longer period of time.

 

[Black Hole]

Can anyone shed any light on this:

I would have thought it wasn't necessary to link the 0V reference ("screen") right through to the input jacks. With an audio signal coupled to the input side of the transformers, my gut feeling is that the output side doesn't have to be referenced to anything and can happily float to whatever voltage the balanced inputs at the far end dictate.

And yet, during experiments, I found that not connecting the reference resulted in worse hum.

 

[Black Hole]

Just for my own satisfaction: here's what I ended up with (having built PCBs yesterday, and shown exploded to see the parts):



The circuit diagram is almost the same as shown in post 14, with the addition of a stereo input socket which is enabled if there is nothing plugged into the separate L/R sockets.

I embarked on a rebuild because the ¼" sockets I used on the prototype proved to be unreliable, so I obtained these gold-plated PCB-mount ones (with switches, giving me the opportunity to make it L+R or Stereo). I could still have hand-wired it, and would have done without this lock-down, but PCBs are more reliable. That involved new machining for the boxes (therefore obtaining new boxes). The PCBs also meant I could use IDC termination for the wires from the RJ45 sockets - another reliability improvement. And I rearranged the conductors in the Cat5 cable: if the patch lead happens to be a cross-over type, all that happens is L & R swap over instead of not working at all.

So: the input can take balanced or unbalanced separate L and R, or unbalanced stereo, and sends balanced stereo out on the Cat5, then the receiver box splits the Cat5 out to balanced L and R.

All I need to do now is colour-code the sockets and add labels, and the job's a good 'un.