How robust is AES/EBU digital?

[Pete Alcock]

I have an Behringer DCX crossover/DSP box which lives in my amp rack and has always worked perfectly. Usually at FOH I have an Ultracurve for house eq feeding the DCX via the multicore. As an experiment, and to avoid 2 sets of A-D & D-A I thought I'd take the digital output from the Ultracurve and put it straight into the digital input of the DCX. I did this and it seemed to work fine (I couldn't hear any difference, but something told me it had to be better.

 

Question is, how robust (or sensitive) is an AES/EBU stream of music? Does it really need a cabe with a characteristic impedance of 110 ohms or will any well screened cable do? Put another way, if 10 is zero error/perfect transmission, and 1 is audible glitching, where is my 33m multicore likely to lie?

 

Thanks as always.

 

Pete.

[Brian]

Question is, how robust (or sensitive) is an AES/EBU stream of music?

Well a few years ago Canford were demonstrating transmission of AES/EBU over wet string!

[J Pearce]

Knowing the hiss on Behringers DACs I'd say this isn't a bad idea if you can get it to work.

 

Best bet is try it, as I suspect it would vary wildly between cable and shielding types. I'd keep the multi channels and cable spare to run it analog, just in case you roll into a gig and have trouble with the digital link though.

[Jivemaster]

Prob a bit like DMX down audio cable! It will work but not necessarily to full spec, distance etc

 

Keep the tested system to hand while you try out the new system.

[broadcast_techie]

Firstly:

http://www.canford.co.uk/blueprintonline/b.../wetstring.aspx

 

Secondly,

Remember, just because it's digital it doesn't mean it has to be better, AES will either work or not, no gradual loss of quality, etc.

 

As you're running 110 balanced AES you should be fine over a reasonable distance. We send audio, (both analogue and digital) over 200m without problems, so your average multi should cope.

 

Kris

[Fletch]

The only real way to know what's going on is to measure it!

 

Have a look at this article about measurements: Digital audio measurements for broadcast

 

I think the main concern is jitter, does your cable increase the jitter level, if so you may be on the verge or errors, but of course you wouldn't know until they become audible (which would be too late) unless you can measure it!

 

From what Kris has said, and what I've seen in use, I would imagine it'll be OK down a multi.

 

Also, as Kris rightly points out, just because it's AES doesn't mean it's better. Your ears are probably the best guide to this, if you can't hear a difference then there probably isn't one!

 

 

Cheers

Fletch

[Mark Payne]

Hi Guys

 

Pete (the OP) PM'ed me for advice in this area as I have a bit of experience. I have also done a whole bunch of testing to evaluate the issues if running AES/EBU over multicore distances. I hope Pete does not mind me responding here as I guess this will be helpful to many.

 

Contrary to advice in this thread, you are not going to be running AES/EBU for 200M.... well not without some techniques..... read on...

 

First I really suggest you read this article: http://www.dbaudio.com/pub/live/TI315_1.0EN.PDF

 

Ok read that?

 

Ok you can use 110Ohm cable but once you go much over 100M it can all go a bit wrong independant of what cable you use. Not all cables are the same and they all fall apart at different lengths. One surprising thing I learnt is that VDC's standard mic cable (not "110 Ohm") is better than their green series 2 pair digi grade 110ohm multicore. Opps

 

The good news Pete is that I am confident that you will be able to get AES/EBU to work reliably over 35M of twisted cheese string. Any old crap will carry it that far.

 

Two things happen over distance..

1. We loose voltage / signal level and more importantly...

2. We loose the definition of the edges of the signal... the digital square waves become rounded off and look more like saw tooths ... I know, I have taken some time scoping this stuff.

 

Now .. standard proceedure is to rebuffer the signal each time we go through a device (the d&b D12 amplifier does this for example, the digi output is electronically rebuffered from the input). This restores the signal level and reforms the edges and you would think you are good for another 100M run but no.....

 

The problem is that the sample syncronisation (the digital work clock if you will) is being carried as time line data along with the audio info. The reshaping of the saw tooth wave form by electronic rebuffering introduces sucessive errors into the time line and corrupts the sync. The clock becomes useless and you end up with drop outs at best and no audio at worse.

 

So a better form of restoration is to take the digital info in, decode it, reclock it, re sync it, buffer it up and send it out again..... AES/EBU distribution amplifiers exist that do this (all the good ones do it) but we cannot use them in live applications (ok we can but be warned) as they successively introduce signal latency at every stage. Imagine that once you have daisy chained through 20 digital amplifiers.....

 

Ok down this far?... I am getting bored myself now so cutting to the chase:

 

1. Do not ysplit AES/EBU ... run it from source to reciever. Any onward daisy chain sends must be buffered.

 

2. AES/EBU sounds great and does not degrade over distance. If your consoles are digi and your amps are digi (containing DSP) then going AES/EBU gets rid of a load of DAC and susequent ADC and you can HEAR the difference... trust me. I cannot hear the difference between 96K and 48K but I can hear this.

 

3. Extend your AES/EBU cables and test it... do it till it goes dodgey and you get dropouts that you can hear.

Now accept that in the real world you will only want to use up to 70% of this distance and you will have reliablity.

 

4. If you need more distance and better reliability use these......

http://www.neutrik.com/content/products/le...d=CatMSDE_video

This converts from 110Ohm balanced line to 75 Ohm Video coax ... you then run your long cable run on 75 Ohm coax cable (cheap as chips) and convert it back again.... its all passive and the littel convertors are about £25 a shot.

 

If this has been helpful.... who's your daddy?

 

Peace

 

Mark

[nonik]

I completely agree with Mark except for the last advice

 

4. If you need more distance and better reliability use these......

http://www.neutrik.com/content/products/le...d=CatMSDE_video

This converts from 110Ohm balanced line to 75 Ohm Video coax ...

you then run your long cable run on 75 Ohm coax cable (cheap as chips) and convert it back again....

its all passive and the littel convertors are about £25 a shot.

 

A balanced line is better than running an unbalanced Coax. This is also true for digital audio. Remember that even computernetworks, with a far higher bitrate, use twisted pair connections ( CAT-5 CAT-6 ) instead of the RG-58 coax.

 

Even worse is to use 2 transformers in the proposed chain. These transformers are passive devices, introducing insertion-loss.

 

An AES-EBU connection is intended for running long distances. The only drawback here is running it over a line that is not 110 Ohm, and for that the advice to experiment with chaining more and more

lines of the snake until the sound interrupts and then use max. 70% of that length is a perfect idea.

 

btw.

The BBC experimented in the 90's running AES-EBU over CAT-5 cable ( that is 110 Ohm as well ) and it worked up to 700 meter. !!!

 

Andre

[Mark Payne]

Hi nonik

 

Good discussion, not sure what you want to disagree with.

 

Fact 1 ... My AES/EBU solution would not hold together for a 110Ohm balanced line "Green Series" digi cable much over 100M.

Fact 2 ... If I do the long run on 100M of 75 Ohm unbalanced line (that we have in stock as video coax) using the convertors I have indicated, I can then daisy chain on from there with another 50M or so of 110 Ohm stuff and my solution works.

 

I have not tried running AES/EBU down 2 cores of the CAT5 ... are you suggesting this will be better than 110 Ohm "digi grade" cable from whoever? I am not adverse to trying it. Maybe I could have saved money on the adaptors. I just have this thing about 20,000 listening to a gig that is reliant on an RJ45 connector. Why would the twisted pair in a cat5 cable be better than the twisted pair down a green thing? I am asking myself..... Hmmmm damm I am going to have to test this now and I am sooooo going to 'ave you if its crap! ;-)

 

The electronic engineer in me just said the characteristic impedance and capacitance of a 75 Ohm video coax was going to be better even though unbalanced. Anyway, it is better... I tested it, it works and the 110 ohm cable "made for the job" does not (again over 100M here). My gig happened.. it all got very wet and was faultless.

 

BTW the insersion loss of the adaptors is small (1db or so) and not seen on a scope. In the go/nogo world of digital its kind of irrelivant.... also transformer issolation is your friend when your systems are on different phases.

 

Good stuff

 

PS Wish I knew all this two months ago, it cost me two days of invested time and a lot of head scratching.

 

M

[nonik]

PS Wish I knew all this two months ago, it cost me two days of invested time and a lot of head scratching.

 

Ha ha, knowing everything, two months before will make you the king of the world...... and rich !

 

The BBC / CAT-5 story is presented in a preprint of the Audio Engineering Society, the AES of the AES-EBU

 

Transformer isolation ( wich indeed is GOLD in pro-audio, analog and digital ) should be provided by a devices

output or input. Ideal would be that all outputs are transformer balanced because there is no need to

chain 2 transformers when 1 will provide the electrical isolation.

 

Looking into a typical studio or PA-setup, microphones at the input side are not mains-connected so need

no isolation. there are more inputs, fewer outputs, audiotransformers are expensive. So let the outputs have

transformers. this applies also to digital outputs.

 

Can it be that your original 110 Ohm cable has a short ? 110 Ohm cable gets it characteristic impedance

partly from the isolation material "polyethylene" that is kindoff foamy and therefore easily damaged.

 

btw. !

Hum, in an unbalanced or even balanced connection ?

Insert a 10 Ohm / 2W resistor in series with the ground-connection. ( can put it inside a connector )

For the ground loop, 10 Ohm is a large value that minimizes the ground-loop current.

For the audio with typical 10 KOhm impedance, 10 Ohm is next to nothing.

 

On RJ-45 reliability:

I have installed audio in radio and television-studios and there are many AES-EBU connections made

with CAT-5 cable using RJ-45 connectors. Besides this, all audio is played by workstations that get

the audiofiles from a server over these standard network connections. In an office-envirement everybody

trusts these.

And then, Neutrik has XLR-like connectors to put an RJ-45 in. !

 

Doing gigs with 20k audience gives you a big responsibility and it is wise to test beforehand.

I hope you will do the same with my advice.. When it works, you'll buy me a beer, when not I come over

to the UK on a fine sunny day and buy you a beer.

 

Andre