Watch SFL Video testing Ether Sound to the Limits

[Mark Payne]

Hi Guys

 

I wanted to share this with you for comments before I bang it out to the wider world.

 

It's quick and dirty video shot in our training room but it shows you the reliability we see in Ether Sound.

It also gives you a look at the M7CL ES board.

Find out what happens when I stretch the CAT5 beyond 100M and start pulling bits of the system out while audio is running.

 

Enjoy

 

Mark

 

http://www.sflgroup.co.uk/technical/Ether%20Sound.mov

 

This is a 40MB file so its broadband only and you may have to wait a min for your browser to download the start of it.

You will need QuickTime Player

 

Comments here or to me mark@sflgroup.co.uk

[Bobbsy]

Thanks for that, Mark--useful video.

 

Did you do any testing with silence and/or continuous tones to see if there are any clicks, pops or momentary dropouts when the cable length is above the 100M spec?

 

As you say, I wouldn't want to push things too far but it's good to see that the spec is resilient enough to put up with a few extra patch cables, adapters, etc.

 

Finally, in your ring topology, is there any problem if you wanted to have one stage box near the mixer (say if you wanted radio mic receivers or more local sound sources for lots of SFX in a theatre setting) or (as we do with our Pro6 at the local theatre) scatter the stage boxes in the pit and one on either side of the stage?

 

Again, thanks for sharing the results of your testing.

 

Bob

[paulears]

That's the kind of info you really wish the manufacturers would tell you - real practical tests that don't even really need your conclusions - it's a kind of here is the evidence, work the limits out for yourself thing.

 

One question, not being familiar with this system - you run it with the extra cable for redundancy, but does the system let you know one has failed - does each line have an LED or other indication, and does this go out when the audio stops, as in the last bit of your demo. I'm thinking that if you got into the area of working for whatever reason, too close to the distance limit - is the sudden lack of audio indicated somehow, to save you having to check the rack, or the amps, or cables etc?

[Pete McCrea]

Great Video. Thanks for taking the time to post it. Not too sure Craig will appreciate the comments on the crappy CAT5 though :-)

[Mixermend]

A very interesting demonstration of the capabilities of Ethersound - thanks Mark!

A clever part of the design of any digital audio system is to decide at which point to mute the audio - clearly this is when the digital error rate becomes unacceptable. But at what point is this?

A few years ago I did some research into extending the distance AES/EBU data could be sent, with the aim of producing a regenerator to extend this distance. It was possible to correlate the digital errors over the link with audio harmonic distortion of a signal sent over the system. I was able to use all sorts of different balanced cables to simulate 'worst case' scenarios - and there was a massive difference between types and makes (don't use quad mic cable for AES/EBU data!).

I wouldn't expect such massive differences with CAT 5 cable, but there will certainly be differences between makes and types.

Mark - do you have a test set to measure harmonic distortion? It would be valuable to see how this rises as you increase the cable distance until the point of audio muting occurs.

Ground breaking research being done here - thanks again Mark!

[Sully]

Fascinating, 100m is only the spec for copper based transceivers anyway of course but I've never seen it pushed *quite* that far, I wonder if they're doing something clever to squeeze a bit more performance out of the cable and wonder how more generic ethernet kit would perform over the same cabling. The failure mode shown in the video is pretty typical to most ethernet faults on data cabling in my experience, it normally works or it doesn't though it is possible to have interfaces drop back to a 10M half-duplex sometimes leaving the appearance of working whilst giving full performance. If it drops back to half-duplex on only one end you can kiss throughput goodbye, does Ether Sound give any feedback as to interface speed/error rate?

 

If you put in non-proprietary media converters such as:

 

http://www.alliedtelesis.co.uk/products/line.aspx?pid=1

 

You can go kilometres over multimode fibre. Put in singlemode fibre and you could go tens/hundreds of kilometres.

The biggest difficulty for a temporary fibre install might be cable fragility but the pre-terminated multimode fibres we've used at work for a few installs weren't *that* fragile (bend radius being the biggest concern) so it's possibly a toss-up for temporary installs depending on the distances.

 

For fixed installs over 100m there's no reason you couldn't put in fibre which is probably a better option than pushing the spec on copper cabling.

 

David.

[Mark Payne]

That's the kind of info you really wish the manufacturers would tell you - real practical tests that don't even really need your conclusions - it's a kind of here is the evidence, work the limits out for yourself thing.

 

One question, not being familiar with this system - you run it with the extra cable for redundancy, but does the system let you know one has failed - does each line have an LED or other indication, and does this go out when the audio stops, as in the last bit of your demo. I'm thinking that if you got into the area of working for whatever reason, too close to the distance limit - is the sudden lack of audio indicated somehow, to save you having to check the rack, or the amps, or cables etc?

 

Hi Paul

 

Yes there are LED's that show the status of both lines (In and Out) but they are connectivity "go/no go" indicators and show noth the TX and RX pairs so it is testing all 4 lines of the CAT5 (This IS 802.3 so I am guessing the other lines are not used). I think they are sensing carrier of some description. Its more of a cable connection sanity check rather quality of service indicator.

 

What I mean is even when I made it fail due to distance, the LEDs say its ok because the cables are basically connected.

 

 

 

 

 

A concurrent post has been automatically merged from this point on.

 

Great Video. Thanks for taking the time to post it. Not too sure Craig will appreciate the comments on the crappy CAT5 though :-)

 

Hi Pete

 

Yes, I regret that comment. I just did this live to camera with no edit and no script. We also have some crap CAT5 in the Audio dept also! We would not put any of it on a gig ;-)

 

Craig, my appologies. I might edit that out ;-) Sometimes I am not very politically correct. Its a failing on my part.

 

 

M

 

A concurrent post has been automatically merged from this point on.

 

Hi Bobbsy

 

Good idea on the tones. I will do that when I get back to this. My window for "play time" has closed for a few weeks but I will get to it and report back.

 

Every time you go through an ES device the signal is rebuffered (electrically) so we have the 100M limit between devices, the full ring could be much larger. How much larger? Well this is hard to answer. I already know from extensive AES distance tests that the full solution distance question/answer is a composite function of (No of repeating devices) vs (inter device length) vs (total overall length) vs (transport media).

 

None of these technologies can be allowed to sample-store-reclock-send because the latencies would be too large and would keep growing.

 

Short answer ... yes you could lob boxes around your stage and orch pit etc.

I will do some more example layouts using your suggestion.

 

M

 

Thanks for that, Mark--useful video.

 

Did you do any testing with silence and/or continuous tones to see if there are any clicks, pops or momentary dropouts when the cable length is above the 100M spec?

 

As you say, I wouldn't want to push things too far but it's good to see that the spec is resilient enough to put up with a few extra patch cables, adapters, etc.

 

Finally, in your ring topology, is there any problem if you wanted to have one stage box near the mixer (say if you wanted radio mic receivers or more local sound sources for lots of SFX in a theatre setting) or (as we do with our Pro6 at the local theatre) scatter the stage boxes in the pit and one on either side of the stage?

 

Again, thanks for sharing the results of your testing.

 

Bob

 

 

A concurrent post has been automatically merged from this point on.

 

Fascinating, 100m is only the spec for copper based transceivers anyway of course but I've never seen it pushed *quite* that far, I wonder if they're doing something clever to squeeze a bit more performance out of the cable and wonder how more generic ethernet kit would perform over the same cabling. The failure mode shown in the video is pretty typical to most ethernet faults on data cabling in my experience, it normally works or it doesn't though it is possible to have interfaces drop back to a 10M half-duplex sometimes leaving the appearance of working whilst giving full performance. If it drops back to half-duplex on only one end you can kiss throughput goodbye, does Ether Sound give any feedback as to interface speed/error rate?

 

If you put in non-proprietary media converters such as:

 

http://www.alliedtelesis.co.uk/products/line.aspx?pid=1

 

You can go kilometres over multimode fibre. Put in singlemode fibre and you could go tens/hundreds of kilometres.

The biggest difficulty for a temporary fibre install might be cable fragility but the pre-terminated multimode fibres we've used at work for a few installs weren't *that* fragile (bend radius being the biggest concern) so it's possibly a toss-up for temporary installs depending on the distances.

 

For fixed installs over 100m there's no reason you couldn't put in fibre which is probably a better option than pushing the spec on copper cabling.

 

David.

 

Ok yes,

 

It is a point well made that ES will allow the use of IT industry fibre repeaters. And they are cheap. However....

We do a lot of DVI over fibre on the video side and its a bit of a pain to be honest. We keep breaking stuff. The crew keep pulling expensive ends off stuff.

 

For now we want to introduce ES at the copper level unless there is a special project that needs fibre.

 

There is no error rate data on the boxes .... BUT yes I hear you. I will run the ESControl management software to see what this can show me.

 

I need that playtime window.....

[Ben Lawrance]

Mark,

 

How do the pre-amps in the stage box compare to the pre-amps in the old flavour of the M7?

 

 

Cheers

Ben

[northernaudioman]

Hi,

 

Interesting demo Mark. A few extra bits of info to add to the pool:

 

- We've been using the Allied Telesis AT-MC102XL on fiber runs of upto 600m for shows for a while without any problems, the cable we use is Belden Brilliance Tactical Fiber Optic Cable which seems rugged enough to cope with live use.

 

- Although redundant mode is great on the new ethersound M7 it's slightly less intuitive when using the ethersound MY cards as you can't connect the computer to do any patching or check on system status whilst at the same time using the ring / redundant topology ( that's without using an extra Auvitan box ).

 

- Auvitran have launched a method recording direct from an ES stream by just patching a CAT5 in via one of their boxes ( http://www.auvitran.com/www/index.php?page=AVM500-ES ) which also acts as an interesting ES matrix box.

 

- We've been using Belden 1305A Deployable Stranded CAT5 for all our ES systems for some time and not experienced any problems, our longest multi is 100m, so we've not tested it beyond this. It is on the ethersound list of tested cables at 105m.

 

- Each ES node ( eg: SB168 ) adds 0.0014ms ( about 0.5mm ) in time delay to the signal path, so the latency by adding boxes is negligible in all practical applications.

 

- ESMonitor has a network integrity monitor which will show the quality of the connection between devices, so you can asses different cabling types and structures. Would be interesting to see the signal integrity when using 200m of variable quality CAT5 cable.

 

Kind regards

 

Oliver.

[mackerr]

It is a point well made that ES will allow the use of IT industry fibre repeaters. And they are cheap. However....

 

Will it? My understanding is that ES uses all 4 pairs of the CAT5, with downstream on 2 pairs and upstream on the other 2 pairs. This would mean it cannot pass through a standard Ethernet switch in both directions, nor can it pass both directions through a standard media converter. There are ES compliant switches and media converters, but they are not the cheap ones you can buy at any data systems supplier.

 

Mac

[Sully]

Will it? My understanding is that ES uses all 4 pairs of the CAT5, with downstream on 2 pairs and upstream on the other 2 pairs. This would mean it cannot pass through a standard Ethernet switch in both directions, nor can it pass both directions through a standard media converter. There are ES compliant switches and media converters, but they are not the cheap ones you can buy at any data systems supplier.

Media converters including some Allied Telesys products are listed on EtherSound's website, the fact they list "compliant" kit (which all appears to be generic ethernet kit) just appears to be what has been tested as working by them. If a product claims to be ethernet based but is doing something else entirely that doesn't interoperate that would be absurd and would deserve to be ridiculed. They claim "Special care has been taken to ensure full Ethernet IEEE802.3x compliance"

 

The fact the data systems stuff is cheaper is down to cut-throat market forces rather than them being "cheap" in a lot of cases, I've had a couple of media converters listed running at the core of my employer's network for several years without a break in service.

 

It would seem a false economy to buy a cheap no-name switch to connect together kit worth x thousands of pounds but as ethernet isn't proprietary you have a choice of various decent kit that should do the job reliably.

 

David.

[mackerr]

Will it? My understanding is that ES uses all 4 pairs of the CAT5, with downstream on 2 pairs and upstream on the other 2 pairs. This would mean it cannot pass through a standard Ethernet switch in both directions, nor can it pass both directions through a standard media converter. There are ES compliant switches and media converters, but they are not the cheap ones you can buy at any data systems supplier.

Media converters including some Allied Telesys products are listed on EtherSound's website, the fact they list "compliant" kit (which all appears to be generic ethernet kit) just appears to be what has been tested as working by them. If a product claims to be ethernet based but is doing something else entirely that doesn't interoperate that would be absurd and would deserve to be ridiculed. They claim "Special care has been taken to ensure full Ethernet IEEE802.3x compliance"

 

The fact the data systems stuff is cheaper is down to cut-throat market forces rather than them being "cheap" in a lot of cases, I've had a couple of media converters listed running at the core of my employer's network for several years without a break in service.

 

It would seem a false economy to buy a cheap no-name switch to connect together kit worth x thousands of pounds but as ethernet isn't proprietary you have a choice of various decent kit that should do the job reliably.

 

David.

 

On the EtherSound topology page and again on the signal routing page it is shown in diagrams, or mentioned in the text that the signal only passes through a switch on the "downstream" side. As I mentioned earlier, I believe this is because the downstream network in on the normal Ethernet pair and the upstream side is on the other 2 pairs. I believe there are switches and fiber converters available from Auvitran that do work in both directions.

 

Mac

[northernaudioman]

 

On the EtherSound topology page and again on the signal routing page it is shown in diagrams, or mentioned in the text that the signal only passes through a switch on the "downstream" side. As I mentioned earlier, I believe this is because the downstream network in on the normal Ethernet pair and the upstream side is on the other 2 pairs. I believe there are switches and fiber converters available from Auvitran that do work in both directions.

 

Mac

Hi,

 

Ethersound uses standard ethernet protocol and hence only uses 2 pairs in the cable, but you can't use a switch if the ES stream is bi-directional, 'this causes more than one 100MB stream of ES to arrive at the same device at the same time and it overloads it.' The Auvitran AVM500ES does enable you to combine ES networks, but it's much more than an ethernet switch, it's a huge matrix switcher enabling routing between 4 or 5 ES networks.

 

Also note that the ethersound protocol always sends all data for all channels irrespective of use, so if you use 2 or 64 channels it makes no difference it's still sending the same amount of data. Less audio channels does not result in less bandwidth.

 

The media converters are standard ethernet devices and work bi-directionally, they don't need to be Auvitran products, standard data ones work fine - but it's best to stick to ES tested products. Yamaha use a HP ProCurve 2510-24 which is rackmountable, has internal PSU, and can use fiber converters, it also enables you to set up V-Lans so that you can use other data on the same network if required.

 

Kind regards,

 

Oliver.

[mervaka]

the ethersound programmers manual states that it operates on 100BASE-TX full duplex. from this, I would imagine that it operates on pairs 2 and 3 (orange and green, pins 1&2, 3&6)

[mackerr]

On the EtherSound topology page and again on the signal routing page it is shown in diagrams, or mentioned in the text that the signal only passes through a switch on the "downstream" side. As I mentioned earlier, I believe this is because the downstream network in on the normal Ethernet pair and the upstream side is on the other 2 pairs. I believe there are switches and fiber converters available from Auvitran that do work in both directions.

 

Mac

Hi,

 

Ethersound uses standard ethernet protocol and hence only uses 2 pairs in the cable, but you can't use a switch if the ES stream is bi-directional, 'this causes more than one 100MB stream of ES to arrive at the same device at the same time and it overloads it.

 

This seems like an arbitrary limitation. Every other audio network I am familiar with manages to pass both upstream and downstream data simultaneously. Among the networks I am familiar with that do pass bi-directional audio through a switch are Dante, Telos Axia, and Wheatstone E2. With GB switches pretty widely available 100MB bidirectional operation shouldn't be an issue.

 

Mac