Need a question answered? Drop us a line at
For information on the CM-1 CobraNet module read the CM-1 page or visit the products section.
In the past, the CobraNet Technology has only been available to Peak’s CobraNet Licensees. Licensees have the rights to the underlying patents and core technology so that they may totally integrate CobraNet into their products. With the concurrence of our Licensees, Peak has developed the CM-1 module and the new Peak Audio Affiliate Program. In this way, the CobraNet Technology may be readily implemented in a modular fashion by Peak Audio Affiliate Subscribers without the necessity of a technology license.
Each node may transmit no more than single packet via the asynchronous service per isochronous cycle. Assuming a 1-1/3ms isochronous cycle (see How long is an isochronous cycle?) and maximum length Ethernet packets (1500 byte payload) gives you a theoretical maximum of 9Mbit/second.
On a repeater network, the network wide asynchronous service bandwidth is controlled by the conductor. Two metrics are published in the Beat Packet header: Asynchronous Packet Count and Asynchronous Accessibility. Asynchronous Packet count defines the number of control channel transmissions per isochronous cycle the conductor can adjust this metric to balance the relative audio and control channel bandwidths. Asynchronous Accessibility essentially defines how often an individual node is allowed to transmit a control packet. Transmissions may be as frequent as every isochronous cycle to as infrequently as every 255th isochronous cycle. The conductor uses Asynchronous Accessibility to keep collisions to a reasonable level.
On a switched network, no network wide bandwidth limitations on the asynchronous are imposed.
The DSP’s host port supports data rates in excess of 10Mbyte/second. The host port can usually be wired directly to the address and data busses of a microcontroller. The host port may be operated in polled, interrupt or DMA modes.
Host Management Interface (HMI) must service an interrupt for each 24bit word of data transferred. This interrupt service can take as long as 100us limiting usable bandwidth to a worst-case figure of 10Kwords/second (30Kbytes/second). Nominal performance can be over 10 times greater but it is complicated to quantify this as a software host port throttling mechanism slows response time of the host port when the node detects it may not otherwise have enough processing cycles to perform critical audio processing.
In addition to transferring data, you need to tell the the HMI were to transfer to or from. This is accomplished by performing a Set Address transaction. A Set Address transaction can take a full isochronous cycle period or more to complete. The performance of data transfer activities which do not operate on contiguous areas of memory are typically constrained by the time required to perform the Set Address transaction.
The bandwidth of the packet bridge is constrained by the bandwidth of the Host Management Interface (see What is the bandwidth of the host interface?), by pipeline delays in receiving and transmitting and receiving Ethernet packets and by the fact that packet bridge data is single buffered.
The following sequence allows a packet to be transmitted via the packet bridge
The following sequence allows a packet to be transmitted via the packet bridge
Our experience has determined that the turnaround time to receive a packet, formulate a response and transmit the response using 200 bytes Ethernet packets is between 20ms (Pentium type host) and 40ms (8051 type host). Calculating a turnaround time from the absolute maximums described above shows lower performance.
24 significant bits of data is present on the synchronous serial audio interface. The resolution of the audio on the network is determined by the transmitter. If you select to transmit fewer than 24-bits onto the network, the data from the audio interface is truncated to the selected resolution before reaching the network. Receivers automatically adjust their behavior based on the configuration of the corresponding transmitter. If you are receiving a 16 or 20-bit audio channel from the network, the remaining 8 or 4 least significant of the bits are 0 filled to generate the 24-bit data that is sent out the audio interface.
The CobraNet interface can transmit up to 4 bundles onto the network. Each of the 4 bundles may contain up to 8 audio channels. The CobraNet interface is thus capable of transmitting up to 32 channels onto the network.
The CobraNet interface can receive up to 4 bundles from the network. Each of the 4 bundles may contain up to 8 audio channels. The CobraNet interface is thus capable of receiving up to 32 channels from the network.
Audio I/O between the CobraNet interface and the host hardware is accomplished through 4 bi-directional synchronous serial interfaces. The data rate of the serial ports is selectable to provide 2, 4 or 8 audio channels in each direction on each link. All serial interfaces must run at the same data rate. The CobraNet interface thus supports a maximum of 64 audio channels (32 in and 32 out).
The isochronous cycle is currently fixed at 1-1/3ms or 64 samples at 48Ksample/second. In order to support shorter delays through the network (see Are there plans to support lower delay through the network?), the isochronous cycle time will need to be shortened. Different cycle times may be deployed in support of other audio sample rates (see Are there plans to support sample rates other than 48K?) and video (see Are there plans to support video?).
Shorter delay is required for certain live performance applications. Lower delay can be achieved at the expense of channel capacity by shortening the isochronous cycle time. The largest issue in supporting shorter delays is in maintaining interoperability. We envision a network configuration process which would allow the operator to make the delay vs. channel count tradeoff. All devices on the network, including those plugged in after the configuration was performed, would then need to comply with the selected operating mode.
CobraNet now supports 3 different delay settings: Standard is 5 1/3ms, followed by 2 2/3ms and 1 1/3ms.
CobraNet networks can support multiple sample rates. CobraNet now supports both 48K and 96K sample rates.
CobraNet is capable of supporting the data transport requirements of compressed video. We are looking for a licensee interested in developing video over CobraNet products.
Switched network support allows CobraNet data onto both of these advanced networks. We are considering plans to build CobraNet nodes with Gigabit ports. In the context of a switched network, a Gigabit CobraNet device would support 640 audio channels. We are starting to see a growing demand for such capacity in a single a device.
Although the Ethernet controller specified in the reference design will support 10Mbit Ethernet, there are currently no plans to add support for this in software.
To do SNMP queries, you’ll need to first get a PC connected to your CobraNet Network. If you are using the CobraNet Evaluation Kit, the kit consists of two EV-1 or EV-2 boards connected through an Ethernet crossover cable. To get a third device, the PC connected into this setup, you will need to obtain a Ethernet hub or switch. Most Ethernet equipment that you buy at a local electronics or computer store will work with CobraNet.
To do the SNMP queries, you’ll need to download and install a copy of CobraNet Discovery from our web site. Install this on your PC. The purpose of this application is to get IP addresses assigned to the CobraNet devices. Run CobraNet Discovery and select the Configuration tab, select the appropriate Network Interface Card (NIC) and check the Enable Auto Assignment option. Flipping back to the Discovery tab, you should now see entries in the list and the entries should have IP addresses assigned to them.
Download install the MG-SOFT MIB browser. You will use this application to retrieve and values of SNMP variables. Download the CobraNet MIB file CobraNet.mib.
Run the MG-Soft MIB Compiler application that is installed with the MIB browser. Choose file/compile and locate CobraNet.mib on your computer. The CobraNet MIB will be compiled, then select "save all" and save to the default location. Close the MG-Soft MIB Compiler application.
Start the MG-soft MIB Browser application. Select the MIB tab in the main window and move the CobraNet MIB entry from bottom window to the top window.
Back in the first tab window of the MIB browser, you should be able to enter an IP address for a CobraNet device in the upper left and view and modify SNMP variables in the main window. CobraNet specific variables are probably what you are most interested in. These are located in the ...private.enterprises.peakAudio.CobraNet.coreManager branch. You can right click on variables to read a description of them.
