ProSoundWeb Community
Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => LAB: The Classic Live Audio Board => Topic started by: Andrew Broughton on December 03, 2019, 08:02:50 PM
-
What’s the deal with L’Acoustics, Meyer and Avid choosing AVB?
What was it that Dante couldn’t do that AVB can?
-
AVB is an open IEEE standard, which anyone can use without licensing.
d&b audiotechnik, L-Acoustics, Luminex and Meyer Sound have been developing Milan (for AVB) for several years now. They don't want to be 'limited by one company’s vision and its future development and support decisions for its technology'. I bet you they were thinking about a certain Australian company...
Wondering where that leaves AES67 though.
-
Funny how the companies with the most proprietary infrastructure systems all want the open standard.
Except that AVB, or TSN as it's now called, isn't even a finalized standard. Which is why Audinate went and implemented Dante....Because they got tires of waiting around for AVB to be finalized. That was like 10 or 12 years ago now. And itsy still not finalized.
Once they finish their video implementation for Dante, I bet they surpass AVB and Milan, as well as AES67 and Raveena all put together by a country mile.
Dante is just so easy and flexible to use and only requires off the shelf networking products where as AVB has to be specifically built into the switches.
-
We're talking about some of the most expensive products in the business. The cost of a Dante License wouldn't even show up as a rounding error on their products.
What I'm getting at, is what is it about AVB TECHNICALLY which made them decide to use it over anything else?
I've not seen anything on the 'net about it except them touting how great it is, but not listing anything that couldn't be done just as easily with Dante.
A big downside seems to be Redundancy. I'm not sure if there's a way to have a Primary and Secondary cabling system with AVB.
-
From the discussions I've have with both L-Acoustics and Meyer people, they seem to dislike something about the latency or clocking of Dante. Somehow they think that it's not tight enough or has some errors inherent to it. Or something.
But you are right, the lack of redundancy is a HUGE drawback to AVB. Also, the lack of being able to use off the shelf networking products is another HUGE drawback.
It seems to me that console and interface manufacturers are jumping on to Dante, while the speaker and amp manufacturers are jumping on the AVB bandwagon. Not sure why that dicatomy exists though.
Once upon a time, Dante was supposed to be fully AVB compliant. Then I think AVB changed their protocol to purposely exclude Dante.
-
We're talking about some of the most expensive products in the business. The cost of a Dante License wouldn't even show up as a rounding error on their products.
What I'm getting at, is what is it about AVB TECHNICALLY which made them decide to use it over anything else?
I've not seen anything on the 'net about it except them touting how great it is, but not listing anything that couldn't be done just as easily with Dante.
A big downside seems to be Redundancy. I'm not sure if there's a way to have a Primary and Secondary cabling system with AVB.
In those situations where different latency to different receivers could be an issue (arrays where varying latency would change phase interactions for instance) AVB would be preferred if directly feeding multiple amps, etc.
You could always feed analogue or AES to those devices as well so there are other ways to get around the changing latency creating a problem.
Lee
-
But you are right, the lack of redundancy is a HUGE drawback to AVB. Also, the lack of being able to use off the shelf networking products is another HUGE drawback.
Woah, hold up. You absolutely can do a redundant network with AVB/MILAN.
And as far as network switches, Cisco, Luminex and Extreme Networks are the ones that I'm familiar with that have AVB compliant products. There may be others but I haven't looked in a while. At one time you had to add on a license to the Extreme switches (may still? it's been a couple years), but I believe the Luminex are off the shelf compliant. So no, no Office Depot level switches, but certainly accessible products in terms of networking, and definitely nothing "custom".
-
Woah, hold up. You absolutely can do a redundant network with AVB/MILAN.
SIDE BY SIDE REDUNDANT AVB NETWORK
https://www.l-acoustics.com/en/hotnews/l-acoustics-launches-avnu-certified-ls10-avb-switch/
-
Woah, hold up. You absolutely can do a redundant network with AVB/MILAN.
I wasn't aware that there was redundancy possible with AVB, thank you for clarifying.
I'm interested to know whether L'Acoustics MILAN system can be set up to be redundant, i.e. can their Amps and processors be set up to use redundant AVB cabling?
The AVB literature tout's prioritized transport (something Dante can do as well) and sub-microsecond clocking, but the standard latency seems to be about 2ms. At least Dante can get it down sub-1ms...
And as far as network switches, Cisco, Luminex and Extreme Networks are the ones that I'm familiar with that have AVB compliant products. There may be others but I haven't looked in a while. At one time you had to add on a license to the Extreme switches (may still? it's been a couple years), but I believe the Luminex are off the shelf compliant. So no, no Office Depot level switches, but certainly accessible products in terms of networking, and definitely nothing "custom".
Yeah, I don't care that much about the additional cost of AVB switches. Decent Dante switches (like the Yamaha) cost too. I think the MOTU AVB switch is probably still the cheapest.
Still not grokking the attraction to AVB for these manufacturers. I will dig in further and speak to them if I can get through to someone in the know at these companies, but so far just seeing what y'all knew about it...
-
Is there an AVB virtual sound card? DVS is big plus for Dante. I know Macs have AVB compliant Ethernet hardware.
Is there a Yamaha AVB MY-card or a Dante -> AVB conversion box?
-
Looks like there is an AVB VSC, but not sure of the price...
http://www.audioscience.com/internet/products/avb/hono_avb_vsc.htm (http://www.audioscience.com/internet/products/avb/hono_avb_vsc.htm)
Apple was SUPPOSED to support it natively, but I guess there's been issues with every MacOS after El Capitan, i.e. 2012?
https://motu.com/avb/using-your-motu-avb-device-as-a-mac-audio-interface-over-avb-ethernet/
-
We're talking about some of the most expensive products in the business. The cost of a Dante License wouldn't even show up as a rounding error on their products.
What I'm getting at, is what is it about AVB TECHNICALLY which made them decide to use it over anything else?
I've not seen anything on the 'net about it except them touting how great it is, but not listing anything that couldn't be done just as easily with Dante.
A big downside seems to be Redundancy. I'm not sure if there's a way to have a Primary and Secondary cabling system with AVB.
Hi Andrew-
I use Dante so I inquired one of these mfgs. About the possibility of adding Dante compatibility in the future for a project I am working on for one of their DSP products. This was the response I received from a higher up support person:
We’ve put our eggs in the Milan AVB basket. I’ve seen some major clock and phase drift in Dante systems that starts to sound like wind.
He said it was confirmed with measurements.
-
Thanks, Kent.
I've never heard of any of these Clocking issues he speaks of. Would love to see the measurements!
Also, surprising that Dante couldn't rectify this issue?
-
What I'm getting at, is what is it about AVB TECHNICALLY which made them decide to use it over anything else?
Layer 2 vs layer 3.
What they've done with DANTE is extremely impressive but, when it comes to timing, it can't compare to AVB/TSN operating on layer 2 with it's deterministic capabilities.
One thing that confuses the situation is that a lot of AVB information on the net right now is from several years ago and often refers to 100btx implementations. The latest specs, including calling it TSN, is largely pushing into automotive and industrial control applications. The latency and synchronization specs in the newer gigabit networking protocols are really tight.
That said, using layer 3 means that DANTE can work in a lot of pre-existing infrastructure and with cheaper networking gear. That and it's strong in terms of history, support and market penetration. I'm curious what's going to happen in the next few years wrt to this.
-
Motu and Presonus make the cheapest switches I know of. The Netgear and Cisco and others require a licence to unlock the AVB features, which I feel is a shame.
-
In terms of clock drift/timing issues, I have a colleague who swears that his active linearray sounds better with analog input compared to AES input, supposedly because the AES inputs has sample rate converters so the timing is off between boxes.
I have no idea if this is possible, but he's very adamant about this being "A Big Issue".
-
If these clock issues are as bad as they're saying, why haven't more users or OEMs spoken up about them?
Googling around, the only issues related to clock drift/desync I was able to dig up were attributed 1) improperly setup networks, 2) improperly configured network equipment, 3) multiple devices set as master wordclock or 4) the master wordclock device itself being faulty.
I have personally never experienced the issues described above with our EAW rig which we have been using for well over 2 years now. It's entirely possible that the scale of our system has something to do with that (we only have around 20 devices at most), but the events we do has it operating for 20+ hours a day, which seems ample time for any gremlins to crop up.
-
Funny how the companies with the most proprietary infrastructure systems all want the open standard.
It does seem funny. At least I'll get to buy and replace a bunch of stuff when Dante isn't compatible anymore :o
I've never experienced any of the drift issues that have been mentioned, and run large Dante networks regularly. I also tour with a Dante based system. Sure there are issues when people don't know how to set things up properly, but isn't that the case with any gear? Perhaps Ive been lucky, and most of the people I know and work with have too.
Having a new standard is going to be a step back for many people and honestly cause more issues than it fixes IMO.
JF
-
It does seem funny. At least I'll get to buy and replace a bunch of stuff when Dante isn't compatible anymore :o
I've never experienced any of the drift issues that have been mentioned, and run large Dante networks regularly. I also tour with a Dante based system. Sure there are issues when people don't know how to set things up properly, but isn't that the case with any gear? Perhaps Ive been lucky, and most of the people I know and work with have too.
Having a new standard is going to be a step back for many people and honestly cause more issues than it fixes IMO.
JF
As far as I'm concerned, these are both "new standards".
If you are using DANTE on a dedicated network with no other traffic then you are far less likely to encounter an issue. One of the biggest concerns that is addressed with AVB/TSN is that it guarantees the timely delivery of the signal by giving it priority at a fairly low level in the networking protocol. When functioning at Layer 3, it is far more difficult to guarantee the timing of the data. Of course, it could be argued that if you are using a dedicated network you may as well operate at layer 2 but, the finer details of that are swamped by all sorts of other issues. Suffice to say, engineering is usually about balancing priorities. DANTE is great in a number of areas but has drawbacks in others. I hope that both protocols will exist into the future. Rather than looking at them as competing, we should probably see them as useful options to solve our 'problems'. You can find info about AVB at the AVNU alliance as well as at some of the manufacturer's websites. Biamp in particular seems to have a lot of info.
-
I remember when the audio industry was talking about going digital, it went something like this:
"Currently the analog world has all kinds of differences, dBu, dBV, 600 ohm, balanced, unbalanced etc. But don't worry, when we go digital, everything will be the same and everything will work with everything else, you just plug it in and go".
YEAH RIGHT!!!!!!!!!!!!!!
It is FAR worse than it was-as far as components and compatibility.
-
I remember when the audio industry was talking about going digital, it went something like this:
"Currently the analog world has all kinds of differences, dBu, dBV, 600 ohm, balanced, unbalanced etc. But don't worry, when we go digital, everything will be the same and everything will work with everything else, you just plug it in and go".
YEAH RIGHT!!!!!!!!!!!!!!
It is FAR worse than it was-as far as components and compatibility.
Just because everyone seems to have a different interpretation of "the Standard"....
Chris.
-
Just because everyone seems to have a different interpretation of "the Standard"....
Chris.
The nice thing about "standards" is that there are so many to choose from :)
-
The nice thing about "standards" is that there are so many to choose from :)
" We've upped our standards, now UP YOURS!"
-
On one hand I want to say 'hats off' to the Australians. They've done something truly spectacular with their tax dollars.
But on the other hand, if we're going to have a standard (de facto or not) it really needs to be open and available. At this point, DANTE requires their proprietary physical chips. I don't believe that any of the under the hood details are openly published. As of a couple years ago Audinate was not interested in working with smaller manufacturers which is really a shame since audio has been a home to dozens and dozens of tiny companies for decades.
I don't know how Audinate treats the education world.
AVB/TSN is open. But, like other open standards, things can move very slowly. Some of the silicon companies (Xmos for one) who could really help in making a new standard a reality, jumped in several years ago but have since back-pedaled. Presumably this is because it started to appear that AVB wasn't being adopted by the industry after all. Thankfully there's been more movement after the automotive industry took notice.
Again, I hope this isn't another VHS/Beta situation where consumer confidence steers the ship. DANTE and AVB/TSN can both exist but so far it's been confusing for end users who largely view them as direct competitors to each other.
And all the while, I'm sure some are still mourning the other network audio standards that have already died. :-\
-
Ok, we're starting to get somewhere now.
[rant]
In the world of PROFESSIONAL LIVE audio, I'm getting sick of hearing how this or that multi-channel audio can "co-exist" on other networks, and how you can use "consumer grade" networking components. How and why did this become a thing?
In the "good old days", we BARELY would piggyback anything on an analogue multicore (maybe some returns, but not DMX or COM if you knew what you were doing). Do you buy consumer preamps and mics? Why shouldn't the networking be as professional as the rest of the gear? Why would you not run your own network for your audio? You had to run your multi-$k multicore by itself in the past, so why are you trying to share these inexpensive networking cables with some other network? Multi-cores cost $1000's, why shouldn't you spend $1000's on your networking too? It shouldn't be about saving money and lowering quality, it should be about the convenience and flexibility that comes with Digital.
Who cares if the networking is Layer 2 when it's running on it's own? As long as there are bridges between protocols that allow you to send/receive digital audio between different formats, that's all that matters. Just like you used to run your OWN Multicore, run your OWN network cabling for the audio.
Argh.
[/rant]
If there is truly a clocking issue with Dante and AVB is the solution (I really don't think companies like Midas and L'Acoustics would choose a protocol lightly), then the missing part of AVB currently seems to be bridges to convert formats, OS support for a working AVB ASIO/CoreAudio Driver, and IO cards for companies that use them (Yamaha, Midas, Behringer, Soundcraft, etc.)
-
As far as I'm concerned, these are both "new standards".
If you are using DANTE on a dedicated network with no other traffic then you are far less likely to encounter an issue. One of the biggest concerns that is addressed with AVB/TSN is that it guarantees the timely delivery of the signal by giving it priority at a fairly low level in the networking protocol. When functioning at Layer 3, it is far more difficult to guarantee the timing of the data. Of course, it could be argued that if you are using a dedicated network you may as well operate at layer 2 but, the finer details of that are swamped by all sorts of other issues. Suffice to say, engineering is usually about balancing priorities. DANTE is great in a number of areas but has drawbacks in others. I hope that both protocols will exist into the future. Rather than looking at them as competing, we should probably see them as useful options to solve our 'problems'. You can find info about AVB at the AVNU alliance as well as at some of the manufacturer's websites. Biamp in particular seems to have a lot of info.
I would hardly call DANTE a new standard. It's been around since 2006 and I haven't been to an A level gig in years where it wasn't part of the audio distribution, console audio, or RF path. For those of us using things like Axient or ULXD and Yamaha desks having to adapt to a new standard (which has admittedly also existed for a while, but never really caught on to my knowledge or markets I visit worldwide) there needs to now be a bridge and new card set to consider. Also with all the desks that have DANTE capability via card to interface with current networks (AH, Midas, Soundcraft seem to be prime suspects here) the argument that DANTE has a lot of issues seems to be refuted by the popularity of it's use. But hey, I'm always interested in something better. I am not interested in solutions looking for a problem though, and I have yet to have a real problem that wasn't self imposed in the DANTE realm.
-
...but so far it's been confusing for end users who largely view them as direct competitors to each other.
I wouldn't exactly say that's the case, I had no idea AVB even existed before this thread. Granted, that's primarily because I don't live in any of the OEMs' ecosystems that are developing it, and I would say that others ignorant of AVB would share that same reasoning.
If anything, I see this as a first to market type scenario where Audinate got their product out first, and have gained significant market share for over a decade because of it. Like John just mentioned, it's everywhere you look now on products ranging from mixers and active cabinets to wireless rx and amps/processors across multiple tiers of products from MI gear all the way to arena touring. Trying to pry people away from it at this point is going to take a paradigm shift either in terms of performance, cost, or both.
-
Again, the most damning evidence for there being a problem with Dante is that companies that can afford to do WHATEVER they want, including creating their own protocol from scratch, are NOT choosing Dante. Cost and licensing cannot be a factor at this level.
I hesitate to mention that I've heard the difference on a K2 rig switching between running AVB and Analog (not AES, unfortunately) and the difference was quite audible. The reason I say that I hesitate to say this is that it wasn't any kind of proper laboratory test, no double-blind, etc. and I've been a big Unbeliever in things like external wordclocks and so on, with the attached claims of changing people's lives, etc. So, I don't put much weight to the "evidence" from that one listening test.
-
Again, the most damning evidence for there being a problem with Dante is that companies that can afford to do WHATEVER they want, including creating their own protocol from scratch, are NOT choosing Dante. Cost and licensing cannot be a factor at this level.
That's not universally true; Yamaha certainly could have done whatever they wanted to, but they chose Dante.
I always figured that it was about control (and not being beholden to a third-party): AVB lets you roll your own chips and firmware, Dante doesn't.
[...] You had to run your multi-$k multicore by itself in the past, so why are you trying to share these inexpensive networking cables with some other network? Multi-cores cost $1000's, why shouldn't you spend $1000's on your networking too? [...]
I've never understood this mentality. We work in a niche industry (or a niche of a niche, if you will), so almost nothing we do has the ability to benefit from economies of scale; yet when the opportunity presents itself to allow us to take advantage of other industries' economies of scale, so many of us balk at the idea.
An unmanaged network switch is, at its heart, a pretty simple device that does a pretty simple job. For most professional live sound applications (i.e. ones that don't need touring-grade physical durability) the difference between a $100 switch and a $1000 is $900 (that could be put towards something more useful).
In addition to saving money, network sharing and commodity hardware leave a lot more options on the table for solving those "oh shit" moments. They also leave the door open to doing things that would otherwise be much more difficult (getting audio between buildings on IT Department infrastructure is a heck of a lot easier and cheaper than burying your own fibre).
Basically, there's more to professional live audio than top-tier touring with top-tier budgets. If the audio is getting from A to B reliably, why spend more than you have to?
-Russ
-
I disagree about using consumer grade hardware and sharing networks, but let's not derail the conversation any more than I already have! :-)
-
Drew said: ...but so far it's been confusing for end users who largely view them as direct competitors to each other.
I wouldn't exactly say that's the case, I had no idea AVB even existed before this thread.
Well, that's exactly the topic of the beginning of this thread. Andrew saw that some companies went with what looked like a different direction and wondered why. So at least he fit my description. I assume that most people in the live sound world, who have heard of AVB at some point in the past, have been assuming that VHS (DANTE) won and BetaMax (AVB) is on it's way out. If they've kept watching, they see that's probably not the case.
Granted, that's primarily because I don't live in any of the OEMs' ecosystems that are developing it, and I would say that others ignorant of AVB would share that same reasoning.
We should be aware that live concert sound is a pretty small industry. This AV over network thing is going on more strongly in the permanent installation/ system integration world where there are many more dollars at play. Think conference rooms, schools and various public spaces. And that's a pretty small world compared to the automotive and industrial applications where TSN is heading. Hopefully our little industries can hang onto those coattails.
If anything, I see this as a first to market type scenario where Audinate got their product out first, and have gained significant market share for over a decade because of it.
This is part of the story. Audinate started off with a large amount of cash to do a lot of the groundwork necessary to make their stuff work. Some smart people there recognized early on that feature rich low latency layer 3 audio (and eventually video) would attract a lot of customers as AV gets more integrated into IT. By the time some other folks were starting to realize the potential for this, Audinate were already shipping product. As the name got around and the capabilities became more understood, they garnered a positive brand image. And the popularity continues to snowball. Users ask for it in their new products. Manufacturers [of a certain size] oblige because Audinate has made it fairly easy. Rinse and repeat. I think that it can be argued that there are systems that would be better for a live music sound system but if you have a dedicated network, everything is probably fine. That said, read Andrew's posts. Several live concert oriented manufacturers have based their systems on AVB, probably with an eye to implement some bridge devices to DANTE and other network schemes where necessary.
There are a few other AV over LAN schemes out there. QSC's QLAN for instance. But if all you do is live music, you probably haven't heard of it.
-
There are a few other AV over LAN schemes out there. QSC's QLAN for instance. But if all you do is live music, you probably haven't heard of it.
And well before any of the current crop there was Telos Livewire. I began using Livewire in the early 2000s. It was a similar system to Dante, using off the shelf switches and proprietary audio nodes. Routing was done by a web based login to each node, which provided a pull down menu of every automatically discovered source available to each destination on the 16x16 audio nodes. Telos had at one point over 10,000 installations in radio stations worldwide.
Mac
-
Unfortunately in the digital world, "standards" become quickly outgrown (and replaced) as technology advances.
-
And well before any of the current crop there was Telos Livewire.
Mac
Any idea what the latency was with Livewire?
-
Thanks, Kent.
I've never heard of any of these Clocking issues he speaks of. Would love to see the measurements!
Also, surprising that Dante couldn't rectify this issue?
For those interested in Dante clocking, please check out the following power point.
Dante Advanced Configuration (https://www.audinate.com/sites/default/files/PDF/advanced-dante-networking-avnw-2015-audinate.pdf?fbclid=IwAR1PiLIUfxO4nPJ5ZiI3mlpOy6sRVcUQ-wNI47Te9nCEPd0iKIWOWdVyHyU)
Specifically where you will find on page 17:
"The sync tolerance of Dante devices is guaranteed to be +/-1μs (microsecond)
In practice we find it to be more like +/-0.2μs
One clock cycle at 48kHz is 20.8μs
So we have a realistic sync within 1/100th of a sample!
All outputs from the network will be w/c aligned!"
That last line seems to dispute Lee's comments earlier about arrival differences for difference end nodes on the network and issues with say line arrays.
In those situations where different latency to different receivers could be an issue (arrays where varying latency would change phase interactions for instance) AVB would be preferred if directly feeding multiple amps, etc.
You could always feed analogue or AES to those devices as well so there are other ways to get around the changing latency creating a problem.
Lee
Does anyone know the sync tolerance spec on AVB?
-
Didn’t livewire essentially become aes67?
And well before any of the current crop there was Telos Livewire. I began using Livewire in the early 2000s. It was a similar system to Dante, using off the shelf switches and proprietary audio nodes. Routing was done by a web based login to each node, which provided a pull down menu of every automatically discovered source available to each destination on the 16x16 audio nodes. Telos had at one point over 10,000 installations in radio stations worldwide.
Mac
-
For those playing along at home, here's too much info about AVB/MILAN:
https://avnu.org/Milan/
https://meyersound.com/download/avb-networking-guide/?ind=1572643301219&filename=avb_networking_guide_a.pdf&wpdmdl=539613&refresh=5debf527c0eb81575744807
Interestingly, the bit depth is 24bits. While just fine for normal audio, I do like that Dante can be 32bits which helps when a multiple consoles are being used off 1 set of preamps and a lot of digital trim is needed. Certainly not a common situation, but a difference nonetheless.
I wonder if whatever clocking issues are overcome by AVB will be negated if one were to use a Dante to Milan bridge?
Maybe clocking isn't the issue as much as companies not needing to rely on a single supplier like Audinate?
(Remember Ethersound??)
-
That's not universally true; Yamaha certainly could have done whatever they wanted to, but they chose Dante.
Yamaha actually did do whatever they wanted, and it's called TwinLANe. But it would have been too expensive for the CL series, and so it's only used in the Rivage system.
Interesting side note that for anything Yamaha the Dante part is a removable module. So if the worst happened and Audinate stopped supporting Dante and manufacturing those modules, Yamaha could make their own replacements.
-
Interesting side note that for anything Yamaha the Dante part is a removable module. So if the worst happened and Audinate stopped supporting Dante and manufacturing those modules, Yamaha could make their own replacements.
Also interesting side note... Yamaha is, or at least was a 10% shareholder of audinate. At least that is what I think pg 58 means.
2017 Audinate Investor Report (https://investor.audinate.com/FormBuilder/_Resource/_module/U31UphySGkWm4tEdvC_Xbw/file/AD8-2017-Annual-Report.pdf)
-
For those playing along at home, here's too much info about AVB/MILAN:
https://avnu.org/Milan/
https://meyersound.com/download/avb-networking-guide/?ind=1572643301219&filename=avb_networking_guide_a.pdf&wpdmdl=539613&refresh=5debf527c0eb81575744807
Thank you for posting these links.
Interestingly, the bit depth is 24bits.
Here's a quote from a resource paper written by Jeff Koftinoff in 2015.
"An AVB Stream can contain different media formats including encoding, sample rates and channel
counts. Some formats that are supported include: integer 24 bit audio, integer 32 bit audio, floating
point 32 bit audio, MIDI, SMPTE, MPEG video with audio, MJPEG video, SDI video with meta data, or
control data. "
-
For those interested in Dante clocking, please check out the following power point.
Dante Advanced Configuration (https://www.audinate.com/sites/default/files/PDF/advanced-dante-networking-avnw-2015-audinate.pdf?fbclid=IwAR1PiLIUfxO4nPJ5ZiI3mlpOy6sRVcUQ-wNI47Te9nCEPd0iKIWOWdVyHyU)
Specifically where you will find on page 17:
"The sync tolerance of Dante devices is guaranteed to be +/-1μs (microsecond)
In practice we find it to be more like +/-0.2μs
One clock cycle at 48kHz is 20.8μs
So we have a realistic sync within 1/100th of a sample!
All outputs from the network will be w/c aligned!"
That last line seems to dispute Lee's comments earlier about arrival differences for difference end nodes on the network and issues with say line arrays.
Does anyone know the sync tolerance spec on AVB?
David,
I was reiterating what I have been told by manufacturers such as L 'Acoustics. they have directly told me that is why they chose AVB over Dante.
I have never tested it myself.
Lee
-
I did a show last night on a EAW Adaptive rig, it's Dante all the way. The rig performer flawlessly as I would expect.
Asked the guy owning it if he ever have had any issues with it, he just came off mixing audience sound for a 10 week music show on tv using that rig. Never had a issue with it during his regular tv shows or other stuff he's been doing. In fact, he finds Dante so reliable that he has dropped using secondary for most shows.
Same thing with all the Yamaha owners I talk to, they usually run Dante primary only.
I'm not saying that it's not possible to have Dante failures, but it seems that you would need more usage time than I have to experience failures or I've been lucky. There could be a ton of cases that I haven't heard about.
-
From what I have been able to gather, the crux of the issue, as I understand it, relates to running Dante to each speaker cabinet in a line or point source array, and that there can me phase alignment issues due to the latency inherent in the Dante system when trying to keep each cabinet aligned.
I think it's a much, much less of a problem if you are only running Dante to the DSP or the amp and then using traditional analog wire for speakers.
At least that is how I understood it when I pressed a guy from L-Acoustics during a demo at a local dealer.
In my mind, if I was running a large array of speakers, I wouldn't use Dante as the audio transport to the cabinets. I might use it to get from the console to the DSP or from the DSP to the amps. But not the individual speakers. Unless I was just deploying a small SOS system with a few cabinets for specific areas, which would make Dante a super easy and flexible solution.
-
From what I have been able to gather, the crux of the issue, as I understand it, relates to running Dante to each speaker cabinet in a line or point source array, and that there can me phase alignment issues due to the latency inherent in the Dante system when trying to keep each cabinet aligned.
I think it's a much, much less of a problem if you are only running Dante to the DSP or the amp and then using traditional analog wire for speakers.
At least that is how I understood it when I pressed a guy from L-Acoustics during a demo at a local dealer.
In my mind, if I was running a large array of speakers, I wouldn't use Dante as the audio transport to the cabinets. I might use it to get from the console to the DSP or from the DSP to the amps. But not the individual speakers. Unless I was just deploying a small SOS system with a few cabinets for specific areas, which would make Dante a super easy and flexible solution.
I’ve been told something similar by a rep from one of the other speaker brands. I have never observed the issues they mention, though latency in a Dante network can be complicated. Each device has its own latency setting; not all devices support all latency options; and Multicast always operates at 1ms latency. These are the things I see as potential ‘gotchas’, not some hypothetical drifting.
Running redundant Dante links to each cab sure would mean you’d have a lot of data cables coming from a hang though.
I think the practical advantages of running AoIP direct to line array speakers is modest however. Probably far more valuable for installation type speakers.
-
From what I have been able to gather, the crux of the issue, as I understand it, relates to running Dante to each speaker cabinet in a line or point source array, and that there can me phase alignment issues due to the latency inherent in the Dante system when trying to keep each cabinet aligned.
I think it's a much, much less of a problem if you are only running Dante to the DSP or the amp and then using traditional analog wire for speakers.
At least that is how I understood it when I pressed a guy from L-Acoustics during a demo at a local dealer.
In my mind, if I was running a large array of speakers, I wouldn't use Dante as the audio transport to the cabinets. I might use it to get from the console to the DSP or from the DSP to the amps. But not the individual speakers. Unless I was just deploying a small SOS system with a few cabinets for specific areas, which would make Dante a super easy and flexible solution.
We've been solely using Dante on our Radius rig for the past 2+ years and have never run into the phasing issues described. This is with two hangs of 6 cabinets each. We tested early on using different network topologies where each hang had its own small switch as well as linking them in one long chain from the end of one hang to the beginning of another and saw no difference in sound quality and zero glitches/digital artifacting. The latency shift was so minimal we didn't even need to adjust anything in the controller to make up for it.
At this point I'd almost look at these claims as mud slinging, but with only anecdotal evidence from either side we'll probably never know until someone fully documents this behavior past 'hearing it from a friend of an A1' or taking a sales rep's words at face value. If there IS an issue, then it needs to be known so that operators can either compensate for it, or hold the developer accountable for implementing a fix.
-
At this point I'd almost look at these claims as mud slinging, but with only anecdotal evidence from either side we'll probably never know until someone fully documents this behavior past 'hearing it from a friend of an A1' or taking a sales rep's words at face value. If there IS an issue, then it needs to be known so that operators can either compensate for it, or hold the developer accountable for implementing a fix.
I really think you nailed the issue here. This industry is becoming more and more technical, and the typical audio sales rep isn't keeping up on the technology. I regularly have to correct sales people on features sets within their own products. Some companies have technical sales folks that really know their stuff, others not so much...
-
"Sound rig sounds wierd, switched from dante to analog, all good" is not a much seen topic. Can it be one of those examples just like a 2" cab can do 144 decibels? Measure at the best/absolutely worst constructed setting?
-
Part of the problem is there's no real way to test this in a scientific way. The products have to be built by the manufacturers to either include the feature set or not. You can't set up a 12 per side hang of Leopards or Karas and test wether the latency or phasing issues exist with Dante unless the manufacturers actually build them.
I do wish that they would at least include it as an option and let the end users determine if the feature meets the cost:benefit ratio for ourselves.
In other words, let the market determine if there is a problem.
-
In other words, let the market determine if there is a problem.
Indeed, far too many variables and unknowns in what is being reported. In the end, time will tell and the adoption rate of AVB will either be its saving grace or nail in the coffin. I'm all for 'the next big thing' if it makes my life easier, but so far this seems more like a pissing match than touting any kind of real innovation.
-
It could be the manufacturers decided that AVB presented less risk of user error that could reflect badly on their product.
-
I'm rather surprised one of the network engineer types on the list hasn't visited this thread to set us straight...
DANTE operates at layer 3. This means it's at the "IP address level" which means it has the sometimes huge advantage that it can pass through 'regular' network switches and can be routed through most network configurations (wifi not included). This kind of data transfer system was not intended to carry time sensitive payloads and has some inherent issues. DANTE is a combination of various "solutions" to those issues.
TSN/AVB/MILAN operates at layer 2. This means it's at the "MAC address level". Even layer 2 was not designed for time sensitive data. BUT, because it avoids the IP level, it avoids several issues. When AVB compliant switches are used, there is a framework for prioritizing packets and defining the timing of streams between talkers and listeners at a fairly low level. But TSN/AVB is also changing, in particular to meet the needs of various industries. And eventually it may be that every manufacturer implements it somewhat differently such that they're all somewhat proprietary. That said, in any form, it is also a combination of various "solutions" to inherent issues involved with packetizing time sensitive data and sending those packets over a network. But those solutions happen at layer 2.
I'm sure there are lots of people in this food chain that don't understand many technical nuances of this (myself included) and there are probably some people taking advantage of the chaos. But assuming that it's just Pepsi vs Coke is missing part of the story here.
So:
A) Yes, for a lot of customers, both of these do the same thing. They allow you to send your audio over a Cat5/6 cable. In this respect, this topic is probably Coke vs Pepsi. But there's also Fanta, 7-UP, Green River, RC and quite a few others that also send audio over a Cat5/6 cable.
B) Yes, there is a fundamental difference here between a proprietary system that is driven by an independent company that has all the advantages of an independent company like moving quickly, unified marketing, private R&D, scales of economy, etc etc AND the completely separate situation of a protocol that is trying to be an open standard for industry and inclusive of as many independent companies that it can entice including all of the issues that this involves.
C) But, there is also a fundamental technical difference between these two systems. As Audio and IT converge, we as audio dudes must learn some of this weird IT stuff. When we understand these things better we'll perhaps know when to use which tool for which purpose.
-
Hello all, Scott Sugden from L-Acoustics, I wanted to comment on some of the questions that have come up in the post.
Why has L-Acoustics specifically chosen AVB as the infrastructure for networked audio?
- Unified network infrastructure: same network cables are used to pass traffic of all types (TCP/IP and digital audio samples). TCP/IP traffic will be automatically rate-limited before it can ever disturb the digital audio traffic.
- Plug-and-play: no network management ever needed, no matter the size of your network.
- Multicast transmissions: any number of devices can listen to the same digital audio stream.
- Low latency: the playback latency is optimized automatically according to the network topology. By “playback latency”, we mean the provision for the maximum time an AVB packet takes to go through the network from the talker to the farthest listening device. In the simple case where all your devices are connected to one single switch, your playback latency can be reduced down to 250 microseconds. If 6 cascaded switches separate the talker from the farthest listener, the playback latency can be set to 1ms.
- Automatic overload prevention: whenever a segment of the network is approaching its physical bandwidth capacity, AVB will automatically prevent any new stream to be added on this segment.
- Determinism: performances of AVB are guaranteed, they are not the result of statistical phenomenon; either it always works, or it never works. “It works” means that your digital audio samples are going through the network and reaching their destination on time. “It doesn’t work” means that you have no audio passing at all.
- Sound quality: AVB relies on state-of-the-art time synchronization protocol that results in very low jitter on the audio clocks. Such an amazing performance is achieved by AVB precisely because it requires all switches to support AVB.
- Playback synchronization: all devices listening to the same talker are able to play the received digital audio samples at the exact same time within +/-1us.
The MILAN solution, which is based on the AVB technology, brings additional benefits:
- Common format for the digital audio traffic.
- Plug-and-play synchronization of audio clocks through the network itself.
- Inter-operable seamless redundancy.
- Unified control of AVB devices from a central controller software.
Scott Sugden
Product and Technology Outreach Manager
L-Acoustics
-
Hi Scott, thanks for the info. Would you care to talk about what specifically L-Acoustics doesn't like about Dante?
-
Hi All,
These are great questions. I encourage you to make this post at the Milan Proboards forum hosted by Avnu. http://milanavcommunity.proboards.com
From there you can get responses directly from the engineers who have created the standard.
Tim Boot
Meyer Sound
I'm rather surprised one of the network engineer types on the list hasn't visited this thread to set us straight...
DANTE operates at layer 3. This means it's at the "IP address level" which means it has the sometimes huge advantage that it can pass through 'regular' network switches and can be routed through most network configurations (wifi not included). This kind of data transfer system was not intended to carry time sensitive payloads and has some inherent issues. DANTE is a combination of various "solutions" to those issues.
TSN/AVB/MILAN operates at layer 2. This means it's at the "MAC address level". Even layer 2 was not designed for time sensitive data. BUT, because it avoids the IP level, it avoids several issues. When AVB compliant switches are used, there is a framework for prioritizing packets and defining the timing of streams between talkers and listeners at a fairly low level. But TSN/AVB is also changing, in particular to meet the needs of various industries. And eventually it may be that every manufacturer implements it somewhat differently such that they're all somewhat proprietary. That said, in any form, it is also a combination of various "solutions" to inherent issues involved with packetizing time sensitive data and sending those packets over a network. But those solutions happen at layer 2.
I'm sure there are lots of people in this food chain that don't understand many technical nuances of this (myself included) and there are probably some people taking advantage of the chaos. But assuming that it's just Pepsi vs Coke is missing part of the story here.
So:
A) Yes, for a lot of customers, both of these do the same thing. They allow you to send your audio over a Cat5/6 cable. In this respect, this topic is probably Coke vs Pepsi. But there's also Fanta, 7-UP, Green River, RC and quite a few others that also send audio over a Cat5/6 cable.
B) Yes, there is a fundamental difference here between a proprietary system that is driven by an independent company that has all the advantages of an independent company like moving quickly, unified marketing, private R&D, scales of economy, etc etc AND the completely separate situation of a protocol that is trying to be an open standard for industry and inclusive of as many independent companies that it can entice including all of the issues that this involves.
C) But, there is also a fundamental technical difference between these two systems. As Audio and IT converge, we as audio dudes must learn some of this weird IT stuff. When we understand these things better we'll perhaps know when to use which tool for which purpose.
-
Hi All,
These are great questions. I encourage you to make this post at the Milan Proboards forum hosted by Avnu. http://milanavcommunity.proboards.com
From there you can get responses directly from the engineers who have created the standard.
Tim Boot
Meyer Sound
Welcome, Tim. Don't forget to go back and change your username to your real name, as required in these forums!
-
Hi Scott, thanks for the info. Would you care to talk about what specifically L-Acoustics doesn't like about Dante?
From the sounds of, it, Dante does not do any of these things "out of the box", and they feel AVB/Milan is more idiot-proof...
Certainly Dante is somewhat plug-and-play, until things get complex and you want higher performance.
I'm starting to get it.
-
From the sounds of, it, Dante does not do any of these things "out of the box", and they feel AVB/Milan is more idiot-proof...
Certainly Dante is somewhat plug-and-play, until things get complex and you want higher performance.
I'm starting to get it.
Easy there partner. Buy the right switches... cough... Yamaha... cough... like you have to with AVB. It will sure do a whole heck of a lot of that list “out of the box”. Or use a ton of other switches with a little setup... the choice is yours.
-
Hi All,
These are great questions. I encourage you to make this post at the Milan Proboards forum hosted by Avnu. http://milanavcommunity.proboards.com
From there you can get responses directly from the engineers who have created the standard.
Tim Boot
Meyer Sound
Please go to your profile and change the "Name" field to your real first and last name as required by the posting rules displayed in the header at the top of the section, and in the Site Rules and Suggestions (http://forums.prosoundweb.com/index.php/board,36.0.html) in the Forum Announcements section, and on the registration page when you registered.
Mac
admin
-
Hi All,
These are great questions. I encourage you to make this post at the Milan Proboards forum hosted by Avnu. http://milanavcommunity.proboards.com
From there you can get responses directly from the engineers who have created the standard.
Tim Boot
Meyer Sound
Let's continue the conversation here! I'm not interested in joining that forum. It does seem from some of the responses that the interest is more in idiot proofing with AVB than having and end user make a mistake in DANTE rather than an actual issue with the protocol.
JF
-
What’s the deal with L’Acoustics, Meyer and Avid choosing AVB?
What was it that Dante couldn’t do that AVB can?
Sorry to derail your post, Andrew...but I am working on a sound system design that will be using Dante console and stage boxes. How do I interface Dante to the AVB speaker DSP? Seems I saw a Meyer Sound write up for a project not too long ago that did such. I know there are some boxes that do this [with lots of other formats included]. This is a pretty simple though, just need Dante [probably 8 streams -not multicast] to AVB DSP.
Anyone have an experience with this?
Thanks.
-
Sorry to derail your post, Andrew...but I am working on a sound system design that will be using Dante console and stage boxes. How do I interface Dante to the AVB speaker DSP? Seems I saw a Meyer Sound write up for a project not too long ago that did such. I know there are some boxes that do this [with lots of other formats included]. This is a pretty simple though, just need Dante [probably 8 streams -not multicast] to AVB DSP.
Anyone have an experience with this?
Thanks.
Kent,
The issue is not so much the conversion. There are a handful of DSP manufactures that will sell you boxes or cards that can translate between Dante and AVB (BSS, Biamp, QSC). The issue is now you are stuck with two independent/isolated networks. If the amp has network audio and Data on the same port, you have to load up a computer with multiple NICs to talk to these multiple different networks.
My solution back when I was using iTechHD's with the old Cobranet chip was to simply use the AES3 inputs, and abandon (the amps) networked audio all together. It resulted in a bunch of Dante to AES boxes but i didn't have to deal with multiple networks.
-
Let's continue the conversation here! I'm not interested in joining that forum.
Indeed. Come on in, the water's fine!
-
Hi Scott, thanks for the info. Would you care to talk about what specifically L-Acoustics doesn't like about Dante?
I don't think I would say there are things we dislike about Dante and or the other Ethernet based audio formats. Its just that AVB & Milan represent a better solution for our needs.
The reasons I listed in the my previous post above are the key ones, each of the points has technical details associated with it that are improvements over Ethernet based solutions.
-
On its face, AVB appears to have more promise going forward. That said, the "single cable integration" carrot-on-stick has been held out forever. Lone Wolf?
-
Sorry to derail your post, Andrew...but I am working on a sound system design that will be using Dante console and stage boxes. How do I interface Dante to the AVB speaker DSP? Seems I saw a Meyer Sound write up for a project not too long ago that did such. I know there are some boxes that do this [with lots of other formats included]. This is a pretty simple though, just need Dante [probably 8 streams -not multicast] to AVB DSP.
Anyone have an experience with this?
Thanks.
Kent,
I designed a very large system (1000+ channels) that used Biamp Tesira Server I/O DSP with Dante cards. The design used Dante wall plates spread throughout the facility to transport audio to the Biamp DSP. The Biamp DSP then transported audio with AVB to Tesira (AVB) equipped Lab Gruppen amplifiers.
The design used Extreme Networks switches. The Dante and AVB coexisted on the same switches. That was difficult to achieve but was successful with close coordination between the integrator, Biamp, and Extreme Networks.
-
Once upon a time, Dante was supposed to be fully AVB compliant. Then I think AVB changed their protocol to purposely exclude Dante.
Not quite. The AVB spec originally envisioned having an additional/optional layer-3 version. Audinate said that once that happened, they would add a compatibility mode to their modules so that all customers would be able to connect to such an AVB network.
Since AVB never committed to adding layer-3 routing, Audinate removed it from their roadmap.
AES-67 kind of solved the vendor lock-in issue on layer-3 anyway.
Tom (ex-TASCAM, now Bluetooth Products Engineer)