CobraNet is a licensed technology developed by Peak Audio, a division of Cirrus Logic. A CobraNet system is comprised of CobraNet devices and the Ethernet network that connects them. It allows reliable, deterministic (i.e. real-time) transmission of high quality (i.e. uncompressed) digital audio over standard 100Base-T Fast-Ethernet network. Up to 64 digital audio channels (20bit/ 48kHz) are carried over one inexpensive, unshielded twisted pair cable CAT5.
Every device, that uses Cobranet, must have built in the circuit CM-1 of Peak Audio.
Repeater hubs and media converters are used in the CobraNet network. These are components that are used in the number of millions worldwide and are relatively simple network components. Two individual CobraNet devices can be directly connected together.
Audio systems have grown significantly and became more advanced in the last years. Larger systems with a substantial amount of signal processing are becoming more and more common.
These improved systems result also in greater complexity, paging and announcement requirements have increased. Audio sources are located in many locations and audio signals must be distributed to a growing number of destinations. The systems have greater demands on the transmission requirements of the systems.
CobraNet is a a multiple-access network, where each device is assigned with its unique address. Any audio signal on the network can be routed to any other device on the network. One signal can simultaneously be routed to not only one output device, but to any number of output devices in the network at the same time. The network becomes an audio router in itself. This routing can be dynamically assigned.
Because a multiple-access network provides also a communications channel to and from each station, all network devices can be monitored from a central point.
With the signing of many CobraNet licensees, the audio industry has begun to leverage its support behind a common audio networking solution that could emerge as the installation market's answer to convergence.
Each CobraNet device consists of specialized hardware and firmware. The hardware includes a standard 100MB Fast-Ethernet interface along with clock circuitry and custom high-speed DSP. The DSP is used to encode and decode the Ethernet network messages using the proprietary protocol. The clock circuitry is used to accurately decode the system master clock timing needed for high-quality real-time audio delivery.
CobraNet places up to 8 channels of audio per Ethernet packet, and it doesn't use IP. It is important to note that it is not IP-based, because CobraNet will not pass audio through an IP switch unless that switch also switches to MAC addresses.
All audio channels are packaged into groups called Bundles (formerly called Network Channels) for transmission over the Ethernet network. The usual assignment is 8 audio channels of 20 bits or 7 audio channels of 24 bits into one Bundle.
CobraNet has maximum network span or diameter of up to 2560 bit periods (with Fast Ethernet, 1 bit period = 10 nanoseconds), or 25.6 microseconds.
The maximum CAT-5 UTP cable length between two CobraNet network devices is 100 meters. It is possible to cover longer distances by using optical fiber:
A fiber optic run of typical 62.5 Ám multimode fiber can be up to 2 kilometers. Single-mode fiber cable with a much better data transmission still exhibits the same amount of delay as multimode fiber. Fiber optics in a CobraNet network should not be longer than 2000 meters.
In a designed system, the maximum number of 2560 bit periods must always be taken in consideration.
Latency - the big issue -
The first generations of Cobranet were created to ship audio around places like theme parks and stadiums, where latency is not a big problem. But today more and more Cobranet is used for applications were latency is important.
The latency time or propagation delay for an audio signal from the input at one device to the output on another device is defined with 5.33 msec (four cycles).
Lately 'low latency modes' were developed and are supposed to allow a latency time of 2.66 msec and even 1.33 msec.
|Latency||Sample Rate||CM-1 Channel Capacity|
|5-1/3ms||96KHz||48 (4 channel bundles)|
Lower latency is achieved by transmitting audio packets at a higher rate. Since the amount of data remains constant, this implies that each packet is smaller for the low latency case than for the high latency case. A restriction on the number of audio channels allowed in a bundle is due to a restriction on the maximum sized Ethernet packet.
Low latency modes will also put additional demands on network performance. Specifically, in order to achieve the desired latency, forwarding delay across the network need to be reduced by approximately the same factor that that audio latency is being reduced.