MOTU goes AVB

[mhschmieder]

I can't say too much on the topic as my company "owns" the authorship of the IEEE-moderated 1722.1 transport layer spec related to delivering AVB on Ethernet, but we were rather thrilled and surprised earlier this week to see MOTU release three new Thunderbolt based interfaces and touting AVB/Ethernet.

Furthermore, there is a nice walkthrough at the MOTU site on AVB:

http://www.motu.com/products/avb/16a/networking.html

I couldn't find my earlier comments on Thunderbolt, Ethernet, AVB (aka IEEE 802.1, which has been around for decades and got rebirthed in the context of the "whole enchilada" that includes the 1722 family of specs that we are helping to develop), IEEE 1722.1, and all the other technologies and specs that intersect at various points (none of those is a subset or superset of the other). I just remember being ridiculed at the time (regarding the long-term viability of Thunderbolt), but I didn't mind because I knew it was the inevitable future (a rare case for me when it comes to hardware technology).

I still have no personal need for Thunderbolt per se, but have been watching discussions across the net and it does seem to be way more of an enabler and integration helper than the scary dead-end proprietary technology many suspected at first.

Of course, Thunderbolt and AVB are not the same thing; though there is some shared DNA.

Anyway, I'm really glad to have seen MOTU do such a good job of presenting AVB on their website; it's one of the better walkthroughs I've seen.

[mikehalloran]

I'm impressed.

Not knowing the protocol, is this something that is software or hardware based. In other words, can legacy switches be upgraded through firmware (should it ever be written, of course) or will this always require dedicated hardware?

I can see the possibilities by linking an AVB switch with the latest 802.11ac wireless from Apple. I am one of those who has tested 802.11ac vs Gigabit Ethernet - 802.11ac is measurably faster.

[mhschmieder]

I forget the minimum Ethernet specs required as we (and every company working on this stuff) are working on making everything work with three generations beyond where consumer net was a couple of years ago (I'm not sure where the consumer net is today; I've been working so hard these past few years that I haven't been able to stay on track with the "real world" vs. the world I see as a developer in the audio industry) -- I think 100G is the current target. MOTU publishes many of the specs on their walkthrough and most of those are probably universal vs. MOTU-specific.

Thunderbird is separate from AVB but now there's a lot of overlap, as I predicted would happen. Thunderbird itself seems to be gaining a lot of ground in terms of hardware converters that act as go-betweens for different formats of connections.

It gets tricky to talk about ALL of the possibilities as most protocols are designed to handle many different types of hardware and vice-versa. There are other broadband protocols in use as well, such as Dante (Focusrite and others). My recollection is that AVB is one of the few, if not the only, to be royalty-free. Also it is designed up-front to be scalable to where things are headed for the next ten years or more, and though I can vouch that the supporting technology is difficult to develop, the goal is to have a plug-and-play experience at the end user end.

When the industry first started talking about this seriously about six years ago or so, one of the things we kept bringing up is that it is very expensive to build new infrastructure, so why not let others do it for us? In other words, Ethernet was chosen because it already exists, is widespread, cheap, scalable, and there's a lot of industries with larger customer bases and more money than the audio industry that will continue to push it forward.

AVB was the primary focus of the AES Show (Audio Engineering Society) in San Francisco in late 2012. Almost half of the talks were on that topic and related topics. The momentum had been growing but it was the first time that consensus was finally reached that it's real, it's happening, and that anyone who doesn't join in will be left behind. The AVnu Alliance signed up several major players right at the conference. Some of the key players include Yamaha, Harman, and Mackie, but more importantly, the computer and chip industry are behind it ($$$$).

Although we live in our own world here, it is to our benefit that one of the primary pushes for this technology is the need for the car industry to improve gas mileage! The best way to do that is to reduce weight, and you'd be surprised how much weight can be saved by eliminating most of the wires in a car and replacing them with Ethernet and WiFi. If you look at the AVnu Alliance's website and/or the AVB site, you'll see that Hyundai and other major car manufacturers are on board.

[mikehalloran]

I spent many years in the auto industry from college to my day job between gigs, 1972-96. I can certainly see the possibilities in weight reduction by replacing the wiring harness.

Dang, 18 years since my last service manager job... Nope, never missed it for a day and am glad to pay others to work on my cars.

[HCMarkus]

Although we live in our own world here, it is to our benefit that one of the primary pushes for this technology is the need for the car industry to improve gas mileage!

Very Interesting!

I so look forward to the day my studio cracks the 100 MPG mark. Or will I again be disappointed by the "your mileage may vary" disclaimer?

[mhschmieder]

Haha, we should put that on the website. Except, I don't think people working on standards appreciate that type of humour (at least in that context).

[TinenTech]

AVB is going to change a LOT of things. And these new MOTU interfaces have many, many uses besides being "interfaces". These really are digital mixers in a single rack space, and by the time you compare the cost of a 24-channel snake to the cost of a Cat 6 cable, they're cheap.

One question...I thought that since Apple has already got support for AVB in OSX, that you wouldn't need Thunderbolt anymore? I thought I'd be able to connect an interface like the 8M to a Mac using a Cat 6 cable instead. But the MOTU diagrams of how AVB networks show a green line connecting any computers to the interface, not a blue one, which I assume means it's not AVB.

[HCMarkus]

One question...I thought that since Apple has already got support for AVB in OSX, that you wouldn't need Thunderbolt anymore? I thought I'd be able to connect an interface like the 8M to a Mac using a Cat 6 cable instead. But the MOTU diagrams of how AVB networks show a green line connecting any computers to the interface, not a blue one, which I assume means it's not AVB.

Yes, MOTU says TB or USB to computer only. Does the AVB spec provide it is possible to connect to PC directly via Cat 6?

[Shooshie]

Umm. if I'm remembering correctly, the CAT6 cable and AVB (ethernet) connectors go between boxes and switches. Thunderbolt goes to the computer, doesn't it? Or can you do it either way?

[HCMarkus]

Umm. if I'm remembering correctly, the CAT6 cable and AVB (ethernet) connectors go between boxes and switches. Thunderbolt goes to the computer, doesn't it? Or can you do it either way?

MOTU says TB/USB to computer only, AVB between audio devices.

I'm wondering if MH can fill us in on what the AVB Spec allows...

[mikehalloran]

I'm wondering if MH can fill us in on what the AVB Spec allows...

Darned good question. There is someone I can ask.

Of course, you are still limited to the speed of GB ethernet - only 20% faster than the theoretical limit of FW800. AVB affords audio a high priority over the carrier but it doesn't make the carrier any faster - as far as I know, that is.

[Shooshie]

I'm wondering if MH can fill us in on what the AVB Spec allows...

Darned good question. There is someone I can ask.

Of course, you are still limited to the speed of GB ethernet - only 20% faster than the theoretical limit of FW800. AVB affords audio a high priority over the carrier but it doesn't make the carrier any faster - as far as I know, that is.

Unless they're able to multiplex somehow on the 8 strands within the cable. That wouldn't be GB Ethernet anymore, though. It'd be their own kind of thing. I always wondered, however, if that's what they did with their PCI-424 boxes. The cables look like Firewire, but I wondered if they utilized the strands better than FW. But those boxes were not daisy-chained; they were star-connected to the PCI-424 card, so the individual load of each box was probably easily handled by the cable. No need for anything fancy. It's their Firewire boxes that daisy-chain and carry quite a load.

Still, the specs for this AVB stuff seem so astronomical at first glance. I wonder what kind of tricks they are using.

Don't mind me. I'm thinking out loud again. Sheesh!

Shoosh!

[TinenTech]

I heard about Apple support for AVB built-in to OSX 10.8? and later at the AES show from an industry insider. I got the impression that at some point in the future you may be able to connect an AVB device directly to a Mac's Ethernet port, but I may have misunderstood.

It seems like the AVB and Ethernet networks can talk to each other, but don't coexist…as MOTU says AVB "goes way beyond Ethernet"…given that it can handle 512 channels of audio with 0.625 ms of latency across 7 switches, and no variability in timing.

I'm really glad that you can hook a Wi-Fi router directly to the Ethernet/AVB port of an 8M and run the internal mixer from an iPad, without having to have a computer connected (like CueMix FX had to have). I guess that's a fair trade for losing Mackie Control of the DSP in the new interfaces

[davedempsey]

On a related theme: this looks like an interesting software product, particularly as it will allow all audio hardware onto the network:

https://www.audinate.com/products/software/dante-via

[mhschmieder]

I'm avoiding answering some of the questions that came up as I expect Mike and others can answer them, whereas it's a lot of effort for me to dig through the internet to make sure it's safe to say things that to me are my everyday life but might still be a bit secret.

As for Dante, it isn't wise for me to say too much since many people know who I work for and it could cause an international incident but I think it's OK to mention that there is a licensing fee (though it's small for typical sound card uses). A lot of vendors (Yamaha in particular) support as many protocols as possible and let the customer decide (sometimes with add-on fees; sometimes all wrapped into the core package) but smaller vendors often have to choose one horse and one jockey due to resources (development time, more than money).

I actually looked into Dante a number of years ago but there are various reasons why it doesn't fit the strategy of some companies. For some people though, it's what's here now (vs. waiting on Next Generation stuff), and thus it has become a bit more widely adopted recently -- even if as a placeholder -- and isn't likely to go away as it seems to bridge well with a lot of stuff. But it isn't an Open Standard (it is proprietary).

http://en.wikipedia.org/wiki/Audio_over_Ethernet

Nevertheless, it is one of the more impressive standards at the moment (certainly an improvement over CobraNet, which I think is probably what it most often replaces).

To be fair, Dante has been around longer, and you'll note that Audionate (who "own" the standard) are also members of the AVnu Alliance:

http://www.avnu.org/about_us/our_members

Plenty of vendors simultaneously promote proprietary and open standards. I have great respect for all of the vendors involved in Audio Over Ethernet, but I'm obviously biased towards AVB since it seems to have the longest timeline for future scalability and flexible topology (i.e. different "shapes" to how network nodes can connect and interact vs. just Ring Topology and Star Topology).

There are secret members of AVnu Alliance and that's where I can't afford to even give the slightest hint -- we're all under the strictest of NDA's. All I can say is that there is a reason for it and it isn't sinister. Some companies don't want it known they're working on AVB until they are confident they have worked through the kinks and know they can announce and deliver product.