Cobranet Boosts Audio Networks

by Michael Karagosian
©1999 MKPE Consulting All rights reserved worldwide
Published in the June 1999 issue of System Contractor News

Cobranet is an exciting technology developed by Peak Audio, Inc. that transmits uncompressed digital audio across Ethernet. Audio over a network implies the availability of control over the network, and yes, the potential is there. Those who once longed for Lone Wolf's MediaLink may find their ears perking up a bit. However, the similarities between the two are thin. Unlike MediaLink, which was highly marketed but didn't have anything to show behind the curtain, Cobranet is not visibly marketed by its creator, but is real and in use today. Also unlike MediaLink, Cobranet does not try to reinvent the network but uses off-the-shelf Ethernet technology. Cobranet has a few limitations will prevent it from being a solution for everyone, but it comes as close as we have ever seen to being a solid, standard platform for digitally interconnecting audio products.

Cobranet 0 was specified by your author for use in a major theme park that opened last year. Cobranet 0 was designed to work over 10Mb Ethernet segments carried by an ATM backbone. Sampled at 44.1KHz, 6 channels of 20-bit audio traversed in one direction and 2 channels in the other. Many Ethernet segments were used, providing a large channel capacity. The significance of Cobranet 0 is that it used unicast packets. In the Ethernet world, unicasting is synonymous with point-to-point connections. In other words, an audio signal applied to the network can have only one destination.

Cobranet I was introduced to the marketplace by QSC Audio Products under the product name RAVE. Unlike Cobranet 0, Cobranet 1 uses the multicast ability of Ethernet to allow "one to many" type of connections. An audio signal applied to a RAVE network can have many destinations. 20-bit audio is sampled at 48KHz, and 8 channels of audio are placed into a packet, or in Cobranet terminology, a network channel. One 100bT network can support up to 8 packets of audo, or network channels, allowing up to 64 audio channels to traverse the network. The use of multicast addressing imposes a problem, however. In the Ethernet world, multicasting audio can clog up an Ethernet switch, rendering it useless. While it is useful for many applications, Cobranet 1 is limited to unsophisticated dedicated networks.

Cobranet II is an attempt to include the best of its predecessors by allowing the use of both unicast and multicast packets. It maintains the 48KHz sampling rate of Cobranet 1, but unlike that version, it allows for 20-bit and 24-bit audio, and a variable number of audio channels within each network channel. It also allows for the virtual addressing of audio channels themselves, not just the addressing of the packet that they reside in. This capability unfortunately prevents it from being backwards compatible with Cobranet 1. But because of Cobranet II's unicast capability, it can traverse Ethernet switches. That allows it to travel over very sophisticated networks, such as 1Gb Ethernet, without the need for virtual LAN technology, or VLANs, per 100Mb Ethernet segment. Without the technical jargon, all of the above means that Cobranet II can be used as an audio router, as well as a solid means of distributing digital audio.

Sounds pretty good, yes? What about control? Even here, there is little surprise. Peak Audio has embraced the SNMP protocol for Cobranet nodes, providing for both monitoring and control of the Cobranet network itself. (This is still in the lab as of this writing, but is expected to be introduced later this year.) SNMP stands for Simple Network Management Protocol. It is not the do-all, end-all of protocols, but it is described in the suite of protocols that are standardized for the Internet. As a result, SNMP is in wide use today for controlling and monitoring networks, both Internet and corporate intranets. We should be in awe: it is a rare company in the audio industry that uses a standard network protocol. SNMP requires IP (internet protocol, network layer 3, for those who have forgotten their networking basics). Cobranet is not IP-based, utilizing only the MAC layer of Ethernet (no, that doesn't mean Apple!), which means the physical and data link layers. Nonetheless, Peak Audio has successfully incorporated IP-based SNMP into Cobranet II, which will give users a standard protocol for monitoring and configuring their Cobranet networks.

Peak Audio boasts an impressive list of licensees for their network technology, including Crest, Crown, Eastern Acoustics Works, Level Control Systems, Peavey Electronics, QSC Audio Products, and Rane. QSC, who is already shipping Cobranet I with their RAVE line, will be shipping Cobranet II products later this year, along with Peavey, who will ship Cobranet II as part of their MediaMatrix product line. Others will soon follow. At NSCA, Rane was privately showing a new Cobranet II-based microphone preamp. Crown was privately showing a new DSP PIP module with Cobranet II inputs.

In the true manner of our industry, each of the companies showing products has implemented networked control of their devices using proprietary protocols. Knowing that one company alone cannot teach an old dog new tricks, Peak Audio provides the means to satisfy whatever communication method their licensees desire. This will result in four categories of control: (1) serial, as in serial port attached to a Cobranet node box to control the remote device or devices; (2) packet-based, but not IP-based, as in proprietary Ethernet protocol; (3) IP-based protocols, other than SNMP, allowing proprietary control protocols in switchable packets; and (4) SNMP, a standard UDP/IP protocol, which can also be switched at the IP level. Of the four methods available, this author prefers the last two. For large systems, anything on IP is a good thing, even if it's not SNMP. IP in itself is a standard, giving the user the advantage of using off-the-shelf IP switchers in their control network. Proprietary protocols need not suffer the stigma they once used to. There are technologies available for incorporating both SNMP and proprietary protocols in off-the-shelf software products, based on ActiveX Controls and JavaBeans. Some companies have already taken advantage of these technologies. That's not to say that any of the other categories will not work for their intended applications. They will all work fine, and each has their strengths for a particular market.

One of the small drawbacks of Cobranet is the latency of the network. All networks introduce latency, which in audio terms, means delay. Some digital distribution methods create delays of only a few sample periods plus the delay imposed by A/D and D/A conversion. (Converter delay can amount to 1 mSec or more.) Cobranet, however, relies upon substantial buffering to insure robust transmission of the signal, which introduces a fixed 5.33 mS delay in the bits. For studio, post production, and stage monitoring applications, this can be a problem. For fixed installations, particularly those that rely significantly upon latency-prone DSP processing, the delay imposed by Cobranet is not a significant issue.

Cobranet II offers a significant step forward in digital audio interconnectivity and routing. Its use of SNMP, a standard control and monitoring protocol for networks, is unprecedented in our industry. The flexibility it offers for control signals allows for forward-looking control methods, which hopefully will be taken advantage of by our industry.