CobraNet CNTraining2

Audio Networks An Overview
Peak Audio, a division of Cirrus Logic, Inc.
Audio Networks
Page 2
Table of Contents:
DETERMINING WHEN AN AUDIO NETWORK IS THE RIGHT SOLUTION ........................... 6 INTRODUCTION ......................................................................................................................................... 6 FACILITY OVERVIEW ................................................................................................................................ 8 DISTANCE AND CHANNEL COUNT ............................................................................................................ 9 OTHER NETWORK REQUIREMENTS AND EXISTING INFRASTRUCTURE .................................................... 10 FLEXIBILITY REQUIREMENTS ................................................................................................................. 11 CONTROL AND MONITORING .................................................................................................................. 12 REDUNDANCY AND RELIABILITY REQUIREMENTS .................................................................................. 13 FUTURE EXPANSION ............................................................................................................................... 14 MATERIAL AND LABOR COSTS ............................................................................................................... 15 STADIUM EXAMPLE:............................................................................................................................... 16 CONVENTION CENTER EXAMPLE:........................................................................................................... 17 ETHERNET: HISTORY AND BACKGROUND................................................................................ 19 INTRODUCTION ....................................................................................................................................... 19 THE FIRST ETHERNET............................................................................................................................. 19 CSMA/CD.......................................................................................................................................... 19 ETHERNET PACKET STRUCTURE ............................................................................................................. 20 THE ETHERNET STANDARD .................................................................................................................... 21 10BASE-T .......................................................................................................................................... 22 100BASE ETHERNET ............................................................................................................................. 23 Fast Ethernet Variations ................................................................................................................... 23 GIGABIT ETHERNET................................................................................................................................ 24 Gigabit Ethernet Variations .............................................................................................................. 24 THE FUTURE........................................................................................................................................... 25 SELECTING THE CORRECT CABLE FOR A PROJECT ............................................................. 26 INTRODUCTION ....................................................................................................................................... 26 THE CABLE PLANT ................................................................................................................................. 26 CAT5..................................................................................................................................................... 26 Distance Limitations ......................................................................................................................... 27 Connectors ........................................................................................................................................ 27 Twists and Terminations ................................................................................................................... 28 Straight-Through vs. Crossover Cables ............................................................................................ 29 Crossover Cables and Uplink Ports.................................................................................................. 30 FIBER OPTIC CABLE ............................................................................................................................... 30 Connectors ........................................................................................................................................ 31 CABLING AND NETWORK PERFORMANCE ............................................................................................... 31 CAT5 and Cable Ties ........................................................................................................................ 32 Pull Force and Bend Radius ............................................................................................................. 32 SOURCES FOR MORE INFORMATION ....................................................................................................... 34 SELECTING THE CORRECT HARDWARE FOR A PROJECT................................................... 35 INTRODUCTION ....................................................................................................................................... 35 DATA TERMINAL EQUIPMENT ................................................................................................................ 35
2001 Cirrus Logic, Inc. Table of Contents
Audio Networks
Page 3
DATA COMMUNICATIONS EQUIPMENT AND ETHERNET HARDWARE ...................................................... 35 Network Interface Card..................................................................................................................... 35 Repeater Hubs ................................................................................................................................... 36 Switching Hubs.................................................................................................................................. 38 Hubs vs. Switches .............................................................................................................................. 41 Bridges and Routers .......................................................................................................................... 41 ESTABLISHING LINK ............................................................................................................................... 42 10Mbps networks............................................................................................................................... 42 100Mbps Devices .............................................................................................................................. 43 Auto-negotiating Devices .................................................................................................................. 43 Auto-negotiation Examples ............................................................................................................... 44 Media Converters.............................................................................................................................. 44 NETWORK ADDRESSING ......................................................................................................................... 45 IP Address ......................................................................................................................................... 46 Subnets and Subnet Masks ................................................................................................................ 47 Classless Addressing ......................................................................................................................... 48 IP Addresses and Audio Networks .................................................................................................... 48 DESIGNING AND BUILDING A NETWORK ................................................................................................ 48 Assign IP Addresses to PCs .............................................................................................................. 49 Configuring the Switch...................................................................................................................... 49 Test Connectivity ............................................................................................................................... 50 RECOMMENDED READING ...................................................................................................................... 51 ROUTING AUDIO ON A NETWORK ................................................................................................ 52 INTRODUCTION ....................................................................................................................................... 52 FILE TRANSFER THEN PLAY ................................................................................................................... 52 Advantages ........................................................................................................................................ 53 Disadvantages ................................................................................................................................... 53 Applications....................................................................................................................................... 54 PLAY OVER NETWORK - PULL MODEL................................................................................................... 54 Advantages ........................................................................................................................................ 55 Disadvantages ................................................................................................................................... 55 Applications....................................................................................................................................... 55 PLAY OVER NETWORKS - PUSH MODEL................................................................................................. 55 Advantages ........................................................................................................................................ 56 Disadvantages ................................................................................................................................... 56 SYNCHRONOUS STREAMING ................................................................................................................... 57 Advantages ........................................................................................................................................ 57 Disadvantages ................................................................................................................................... 58 ISOCHRONOUS STREAMING .................................................................................................................... 58 Isochronous Transmitters and Receivers .......................................................................................... 59 Advantages ........................................................................................................................................ 59 Disadvantages ................................................................................................................................... 60 COBRANET............................................................................................................................................. 60 COBRANET TERMINOLOGY .................................................................................................................... 61 COBRANET INTERFACE .......................................................................................................................... 62 THE ISOCHRONOUS CYCLE ..................................................................................................................... 63
Audio Networks
Page 4
CONDUCTOR........................................................................................................................................... 63 CONDUCTOR ARBITRATION .................................................................................................................... 63 COBRANET PROTOCOL........................................................................................................................... 64 Beat Packet........................................................................................................................................ 64 Reservation Packet............................................................................................................................ 65 Audio Data Packet ............................................................................................................................ 66 NETWORK MANAGEMENT AND CONTROL ............................................................................... 67 INTRODUCTION ....................................................................................................................................... 67 WHAT IS NETWORK MANAGEMENT?...................................................................................................... 67 TELNET, HTTP AND FTP....................................................................................................................... 68 Telnet................................................................................................................................................. 69 HTTP ................................................................................................................................................. 69 TFTP.................................................................................................................................................. 70 MANAGEMENT USING SNMP ................................................................................................................ 70 AGENTS MANAGERS AND MIBS ............................................................................................................. 71 Agent.................................................................................................................................................. 71 MIB.................................................................................................................................................... 72 Manager ............................................................................................................................................ 72 SNMP Message ................................................................................................................................. 73 MANAGEMENT SOFTWARE ..................................................................................................................... 75 Polling and Network Topology ......................................................................................................... 75 Drill Down Capability....................................................................................................................... 75 Event Logs and Performance ............................................................................................................ 76 RECOMMENDED READING ...................................................................................................................... 77 SNMP MANAGEMENT APPLICATION REFERENCE .................................................................................. 77 DESIGNING A NETWORK TO ACCOMMODATE SPECIAL NEEDS........................................ 78 HIGH AVAILABILITY .............................................................................................................................. 78 High Availability Design................................................................................................................... 78 Quality Components.......................................................................................................................... 78 Cold Spares ....................................................................................................................................... 79 Hot Swap ........................................................................................................................................... 79 Fail-over............................................................................................................................................ 79 Fault Tolerance ................................................................................................................................. 79 ENVIRONMENTAL FACTORS ................................................................................................................... 80 MONITORING .......................................................................................................................................... 80 LIFE SAFETY .......................................................................................................................................... 81 NETWORK AVAILABILITY ...................................................................................................................... 81 Link Aggregation (Trunking) ............................................................................................................ 81 Spanning Tree Protocol .................................................................................................................... 82 Meshing ............................................................................................................................................. 83 AUDIO INTERFACE AVAILABILITY .......................................................................................................... 84 Pairing Interface Devices.................................................................................................................. 84 Buddylink........................................................................................................................................... 84 DuaLink............................................................................................................................................. 85 MIXED USE NETWORKS ......................................................................................................................... 85
Audio Networks
Page 5
NON-BLOCKING SWITCHED NETWORK DESIGN ..................................................................................... 86 VIRTUAL LOCAL AREA NETWORKS ....................................................................................................... 86 QUALITY OF SERVICE ............................................................................................................................. 87 LARGE INSTALLATIONS .......................................................................................................................... 87 CORE SWITCHING ................................................................................................................................... 88 AUDIO VLANS ...................................................................................................................................... 88 LAYER 3 ROUTING ................................................................................................................................. 89 DISTANCE AND SWITCH HOPS ................................................................................................................ 89 Forwarding Delay ............................................................................................................................. 89 Delay Variation ................................................................................................................................. 91 TROUBLESHOOTING AUDIO NETWORKS................................................................................... 92 INTRODUCTION ....................................................................................................................................... 92 CHECK THE OBVIOUS!............................................................................................................................ 92 VERIFY LINK .......................................................................................................................................... 92 MANAGED SWITCH CONFIGURATION ..................................................................................................... 93 ARE LEDS INDICATING ERRORS? .......................................................................................................... 94 CHECK THE AUDIO ................................................................................................................................. 94 Signal Presence ................................................................................................................................. 94 Audio quality ..................................................................................................................................... 94
2001 Cirrus Logic, Inc.
Table of Contents
Audio Networks
Page 6
DETERMINING WHEN AN AUDIO NETWORK IS THE RIGHT SOLUTION Introduction

Networks have had a major impact on technical systems of all types. Most commercial buildings include a network infrastructure. These networks can carry many types of data including lighting, security, point-of sale information, as well as general office computer data. The world has indeed become a networked world. Until recently, the audio system has not been a part of the building data infrastructure. Rather, it was always a separate, typically analog wiring system. Having to install a completely separate wiring infrastructure can result in a great increase in costs of materials, labor and time. In certain types of projects, the savings can be very significant. In addition to the cost factors, there are of course, many technical advantages to distributing audio over a network. Flexibility is greatly increased. In a properly designed system any audio input can be routed to any audio output without having to physically move wires or make use of patchbays. This routing capability allows for different system configurations and for future changes to be accommodated without having to make hardware or wiring changes. Redundancy and fault tolerant designs can be implemented using audio networks. A system can be designed so that if any piece of equipment or wire on the network should fail, an automatic switchover can take place. In a traditional analog system, if a wire is cut someone has to discover this through troubleshooting and then it has to be either repaired, replaced or patched around. Through network management a centrally located computer can monitor all network
When a Network is the Right Solution
Audio Networks
Page 7
products and alert the system operator if a fault should occur anywhere in the system. Networked systems can result in a significant reduction in cable and infrastructure requirements. A single CAT-5 or fiber optic cable can replace many cables. The same cable that transports the audio can carry control and monitoring data as well. In deciding whether an audio network is right for your project, you must expand your thinking beyond the traditional cabling means. Networks provide many new capabilities that were heretofore not possible. You must be willing to learn about these new capabilities and then use your system design skills and knowledge to think of ways where these new possibilities can provide better systems. Just the fact that you can now have any channel appear at any output results in the necessity of thinking in an entirely new way. Those contractors and consultants who do not learn how to take advantage of the world of networking will be left behind. It is also just as necessary to not become so enamored with the new technology that you use it when a simple analog point-to-point connection is the best solution. By the end of this class you should have the knowledge tools necessary to move forward and lead this industry into the networked world. While there are many benefits to using a design that distributes audio over a network, it is not always the right solution. In this section we will examine audio system needs and learn how to determine whether an audio network is of benefit for a particular project.
Audio Networks
Page 8
Facility Overview
There are several criteria that must be judged when determining whether a network is appropriate for a project. These include: Overall scope of the project Equipment locations Routing needs Distances Channel counts Other network infrastructure and existing infrastructure Flexibility requirements Control and monitoring requirements Redundancy and reliability requirements Future expansion requirements Material costs Labor costs The first step in determining if a networked approach is appropriate is to look at the big picture. How many physical locations have audio inputs, audio outputs or audio processing equipment? What locations require system control? How physically large is the system? What are the distances between locations? Is the client technology oriented? If all of the equipment is located in one small equipment closet, a network may not be a good solution as one of the benefits of networks is the ability to transport audio and data over long distances. The first step is to locate all of the equipment locations, input locations, output locations and control locations. Typically, as the number of locations goes up so does the convenience of utilizing an audio network.
Audio Networks
Page 9
Next, determine the audio routing requirements. Does all audio need to matrixed between all locations, or are the needs simply point-to point? Is there a need to change routing depending upon the needs of a particular event? An example of this is a convention center where the needs change on an hourly basis. One day a meeting room may be associated with a particular ballroom and the next day it may be tied into the audio feed from an adjacent large hall. Will the needs change over time? The needs of legislative facilities often change after an election, especially if the majority or chairmanships have changed. Will the actual routing needs be determined after the system design must be complete? An example of this is in a theme park. The infrastructure must be designed well ahead of the show itself. The show designer often does not determine his or her actual needs until the very end of the project. The flexibility of being able to change the routing at the end of the project makes a network a very attractive solution. If the project simply has a control room and a single amplifier closet where routing changes are never needed, an analog solution may be appropriate
Distance and Channel Count

The distances between locations are an important factor in determining when a network is appropriate. Typically longer distances lend themselves to networks, although short distance situations may also benefit from a network when other factors such as routing requirements are considered. A single CAT-5 cable can carry 64 channels in each direction (a total of 128 channels) up to 100 meters using CobraNet technology. This same data can be carried up to 2 kilometers using multimode fiber optic cable. It is possible to carry over a thousand channels over Gigabit switched Ethernet. These design rules will be covered in a later part of the class.
Audio Networks
Page 10
As part of the design process you should determine how many input and output channels are required at each device and location. It is also useful to know how many channels need to be transported between locations.
Other Network Requirements and Existing Infrastructure

It is useful to determine what other data will be transported around a facility. If it is a new facility you should make the effort to learn what the data distribution needs are for other disciplines. You may learn that the M&E engineering firm has designed in an Ethernet network for the lighting and security and that this network has excess capacity. If this bandwidth can be made available to you, there could be a huge cost savings on the sound system. Or perhaps you will find that the IT department is designing a facility network and your needs can be included in this, thus reducing the cost impact on your design, perhaps freeing up funds for other areas of the sound system. Is there another discipline that is designing a data network for their purposes within the same facility? Perhaps you can coordinate efforts with them and share in the infrastructure costs. Remember, if your data is being transmitted over Ethernet, and the network utilizes switching technology, you can share the network and your data can travel down the exact same wires as the data being distributed for other purposes. If the project is being installed into an existing facility you should find out what network infrastructure is already in place. If the infrastructure exists, does it have excess capacity? Does it have drops in locations that are useful to you? You should also determine if there is an infrastructure in place that could prove to be of benefit in a design that uses traditional analog distribution. Perhaps there are a number of empty conduits available for your use that could be utilized. Does a cable tray system exist?
2001 Cirrus Logic, Inc. When a Network is the Right Solution
Audio Networks
Page 11
Flexibility Requirements
Traditional wiring is of the point-to-point variety. This means that the signal always goes from one known location to another, and the connection is hard-wired in place. This is true of both analog and digital wiring. Even though an AES/EBU signal is digital, it still only connects from one specific point to another. This type of hard-wired condition is very limited, as it only operates in one manner. It is not flexible or configurable. The method that is normally used to add some level of flexibility is to incorporate patchbays into the system. Although, this allows you to change the routing of the audio, it is still very limited. Each location that the audio goes to must have a dedicated cable that is home run to the patchbay. Patchbays are themselves limited to the signal paths that are available at the patchbay and are not very flexible. Also, patchbays must be operated by a technically competent person. They can create problems such as hum, loud pops, noise, etc. In a network, all of the audio channels can appear anywhere on the network (depending upon the particular network design). This greatly increases flexibility, as the designer did not have to home run dedicated wires (with a dedicated pair for each channel.) The network can be easily configured on site. Channel routing can be stored into presets allowing the functional wiring to completely change to meet the needs of a particular event. It can also be changed in the future to meet the changing needs that may occur down the road. You can design your network so that you have enough capacity and nodes so that you do not actually design the routing ahead of time, but rather do so in the field as the exact needs unfold. As mentioned earlier, this capability can prove very useful in a theme park where the show designer works with you to fine tune the needs long after the basic design had to be completed. This is highly flexible. Is this type of flexibility and configurability desirable for your project?
Audio Networks
Page 12
Since this opportunity was not previously available, you must think about the potential benefits of having these new capabilities. While designing the system using traditional point-topoint wiring may work, these new capabilities may prove to be beneficial to your project. On the other hand, you may not need any of these capabilities. By attending this class, you are equipping yourself with the knowledge to learn how to make that determination.
Control and Monitoring

When we discuss control and monitoring, we are talking about two types of control and monitoring. One is the control and monitoring of the network itself. This includes all cables, network products such as switches, and the actual products producing and consuming data on the network. As an example, you may have a power amplifier on the network that has a network interface built into the amplifier itself. You may want to monitor the status of the network interface in the amplifier to be certain that it is operating properly. This can be done from a remote location over the network. There are commercially available products that perform these functions. If a fault should occur anywhere on the network you can be alerted to that fact and easily diagnose the problem. This type of monitoring increases the reliability and ability to diagnose a network. These capabilities must be weighed in deciding if an audio network fits your particular project. The other type of control and monitoring, in our amplifier example would be performing functions such as remotely observing the clip lights on the amplifier and adjusting the levels from that same remote location. CobraNet networks can transport many types of control and monitoring data. While the network carries this data, the equipment manufacturer supplies the control data. Examples of this include QSControl, MediaMatrix, IQ, RaneWare, Nexsys, etc. The important thing to consider is the fact that if you are going to be using these
Audio Networks
Page 13
types of control and monitoring software in many cases, when using CobraNet, they can be carried over the same wire or fiber that the audio is carried on. This can greatly reduce the complexity and cost of the cabling infrastructure by eliminating an entirely separate infrastructure. Installation and troubleshooting are also greatly reduced when there is only one cable infrastructure. This can be a very significant factor if you are using these types of control or monitoring systems. If your system does not use these technologies, then this audio network advantage would not apply to your project.
Redundancy and Reliability Requirements

Ethernet networks are used in a very large number of mission critical applications throughout the world in many different disciplines. The need for redundancy and faulttolerance has been around for a long time and there are many different solutions available depending upon the needs of the project. One of the benefits of utilizing well established technologies is that these needs have already been addressed. When an entirely new technology is invented, the problems must first be discovered, then the solutions must be developed and finally they must be tested and refined. This takes a lot of time and being involved in this process at the beginning can often be a time consuming and frustrating experience. As mentioned previously, managed products such as managed switches can be purchased that have the ability to be remotely monitored. Using network management software an operator can monitor the operation of the entire network. This greatly improves reliability.
Audio Networks
Page 14
By taking advantage of redundancy schemes it is possible to design many different levels of redundancy including a network that is fully redundant. This means that every cable and device is duplicated in a redundant, parallel system. If any piece of equipment or cable should fail on the network, the failure will be automatically detected and the faulty item or network will be bypassed and replaced with the redundant option. This can all occur without any human intervention and is quite seamless. Any cable can be cut without affecting the continued audible presence of audio over the system. For projects that need a lesser amount of redundancy or fault tolerance other schemes, typically at a lower cost, can be implemented. With traditional cables, if a wire is cut, or if a connector fails, or a product breaks, it will require a person to troubleshoot the problem and manually correct it. In a well designed system with a patchbay this can possibly be easily accommodated. In other systems, it may take a long time, however there are projects where that is acceptable. One other benefit of digital audio, especially when it is transmitted over fiber optic cable is the low susceptibility to RFI (radio frequency interference). All audio networks are digital and thus enjoy this added benefit.
Future Expansion
Any well thought out system can be designed to accommodate future expansion. A networked system offers the benefit of an infrastructure that is readily expanded. It is fairly easy to add more nodes or products onto an existing network. With an analog system new locations often require additional cable, conduit etc. Of course, if the future expansion simply requires a larger mixing console, or adding some new equipment into a new rack, then a network may be of no benefit at all. If the future expansion involves new locations, more channels being distributed, or the ability to change routing in the future, then audio networks can provide a real benefit.
Audio Networks
Page 15
Material and Labor Costs

The materials required for a networked audio system differ from those required for a traditional analog or digital system. A networked system has the added cost of network specific products such as Ethernet switches and management software (if the network is of the managed variety). But a network it greatly reduces other infrastructure costs. In a typical large facility there are numerous runs of multichannel cables through conduits. The ends of these cables are typically connected via large multi-pin connectors. The material cost for these items can be very high, especially when there are long runs in conduit. Connectorizing these cables can take a lot of time, therefore the labor costs are high. The installation of the conduit is also a very labor intensive process with a high cost. And troubleshooting these systems adds further to the labor costs. Of course in a smaller facility such as a restaurant or bar, there may not be very much wiring and therefore the materials and labor costs are not very high. Over a thousand channels can be carried over a pair of fiber optic cables using Gigabit switched Ethernet. This data cable does not need to be run in conduit and results in a significant reduction in both material and labor costs. Even a standard CobraNet signal over CAT-5 cable can carry a total of 128 channels. The connectors are simple crimp type or fiber optic types. These do not require soldering and are very easy to install. They do not require the specialized skills needed when working with the more typical audio connector varieties. Depending upon the system requirements there can be a huge cost savings incurred when using an audio network in a system. Sample cost and design comparisons are shown below.
Audio Networks
Page 16
Stadium Example:
There is a total of 4600 of cable used in this analog design example, and most of the conduit is 3". With cable and conduit costs estimated at $44,700, and a labor estimate of $1,500, the total infrastructure cost for this analog system if $46,200.
ANALOG PROCESSING EQUIPMENT 2-32 PAIR CABLES
PA-1
2-32 PAIR CABLES
PA-1
1-32 PAIR (SPARE)
PA-2
1-32 PAIR (SPARE)
PA-2
PA-3
PA-3
PA-N
PA-N
EQUIPMENT RM 1
EQUIPMENT RM 3
2-32 PAIR CABLES
PA-1
3 C. (TYP.)
1-32 PAIR (SPARE)
PA-2
PA-3
CENTRAL CONTROL LOCATION EQUIPMENT RM 2
PA-N
Traditional Analog System

While the network design may look more complex, its actually much simpler from an infrastructure perspective, and much less costly. Networks can carry more channels over fewer wires, and using fiber in the existing cable tray reduces the infrastructure cost to under $3,000.
ANALOG PROCESSING EQUIPMENT
COBRANET INTERFACE (TYP) CN CN CN 100BASE -FX SWITCH
PA-1
PA-1
PA-2
CN 100BASE -FX SWITCH
PA-2
CN
PA-3
CN
PA-3
PA-N PA-N
CN CN 100BASE -FX SWITCH
EQUIPMENT RM 1 CABLE RACEWAY
EQUIPMENT RM 3
PA-1
CN CN
3/4 C. FIBER OPTIC CABLE (TYP)
CN 100BASE -FX SWITCH
PA-2
CN
PA-3
CENTRAL CONTROL LOCATION EQUIPMENT RM 2
PA-N
CobraNet System
Audio Networks
Page 17
Convention Center Example:

An 8 pair cable and a 4 pair cable are run to a remote equipment closet from each of 4 meeting rooms in a 100 ft. long 1-1/2" conduit. A 3" conduit is home run from each remote equipment closet to the equipment room. These conduits carry four 8-pair and four 4-pair cables each. In this example, well look at the costs for the Meeting Rooms only. The conduit and cable cost is almost $160,000, while the labor to install the cable is a little over $8,000. This results in a total infrastructure cost of $163,000.
CENTRAL CONTROL LOCATION

REMOTE EQUIP RM 1
REMOTE EQUIP RM 2 TO RMS 3-5
3 C. 4-4 PR 4-8 PR
ELECTRICAL PULL-BOX
1-1/2 C. 1-8 PR CABLE 1-4 PR CABLE MTG RM 3 MTG RM 4 REMOTE EQUIP RM 6
ELECTRICAL PULL-BOX
MTG RM 1 MTG RM 2
MTG 5 MTG 6
MTG 7 MTG 8
REMOTE EQUIP RM 7
TO RMS 8-10
3 C. 4-4 PR 4-8 PR
ELECTRICAL PULL-BOX
ELECTRICAL PULL-BOX
MTG 21 MTG 22
MTG 23 MTG 24
MTG 25 MTG 26
MTG 27 MTG 28 TO HALLS 4/3
2-1/2 C. 1-16 PR 2-4 PR
ELECTRICAL PULL-BOX
PANEL 1 PANEL 2
ELECTRICAL PULL-BOX
PNL 4 PNL 5
1-1/2 C. 1-12 PR 1-4 PR
EXHIBIT HALL 1
PANEL 3
EXHIBIT HALL 2
TO BALLROOMS 1-4
Traditional Analog System
Audio Networks
Page 18
Again, theres a bit more equipment involved, but a much simpler infrastructure. With a CobraNet system, one 1000 ft. long " conduit connects the North remote equipment closets together and another connects the South equipment closets. This is a total of 2000 of " conduit plus 4000 ft. of 1-1/2" conduit to connect the I/O panels to the remote equipment closets. Conduit and cable costs total $30,000. With the addition of labor, the total infrastructure cost is $36,700.
CENTRAL CONTROL LOCATION
REMOTE EQUIP RM 1
REMOTE EQUIP RM 2
CN CN
100BASE -FX SWITCH

3/4C. (TYP) 4 STRAND FIBRE
100BASE -FX SWITCH

1-1/2C. 1-8 PR 1-4 PR MTG 1
CN CN
100BASE -FX SWITCH
CN CN
TO RMS 3-5
MTG 3 MTG 4
MTG 5 MTG 6
MTG 7 MTG 8 TO RMS 9/10
ANALOG AUDIO LINES (TYP)
MTG 2
CN CN
100BASE -FX SWITCH
100BASE -FX SWITCH
CN CN
100BASE -FX SWITCH
CN CN
EQUIP RM 6
EQUIP RM 7 MTG 3 MTG 4
CN CN
100BASE -FX SWITCH
MTG RM 1 MTG RM 2
MTG 5 MTG 6
MTG 7 MTG 8 TO HALLS 3/44
100BASE -FX SWITCH CN CN 100BASE -FX SWITCH
CN CN
P1 P2 P3
100BASE -FX SWITCH
CN CN
P4
EXHIBIT HALL EQUIP RM 1 TO BALLROOMS 1-4
EXHIBIT HALL EQUIP RM 2
P5 2-1/2C. 1-16 PR 1-8 PR
CobraNet System
Audio Networks
Page 19
ETHERNET: HISTORY AND BACKGROUND Introduction

In 1972 Robert Metcalfe and his Xerox Palo Alto Research Center (PARC) colleagues developed the first experimental Ethernet system to interconnect the Xerox Altos. This was initially called the Alto Aloha Network, but in 1973 Metcalfe changed the name to "Ethernet.
The First Ethernet

The first Ethernet functioned as shared media, with the available bandwidth shared amongst all of the stations connected to the network. Although the transmissions from any one station are received by all stations, only one station may transmit at a time. A collision is a situation that occurs when two or more devices on the network attempt to send a signal along the same wire at the same time. The result of a collision is generally a garbled message. Collisions are a fact of life in a shared media Ethernet network, and under most circumstances should not be considered a problem. As a result, the network requires a mechanism to both manage access to the network, and to prevent or recover from collisions. CSMA/CD Carrier Sense, Multiple Access with Collision Detection, or CSMA/CD, is the media access method used by Ethernet. Carrier Sense means that network stations with data to transmit should first listen to determine if another station is sending data. If another station is talking, this station will wait until there is no carrier signal present. Multiple Access means that Ethernet provides a number of stations the opportunity to transmit on the single cable. When a station has completed its transmission it is allowed to immediately make another access to the medium. Collision Detection refers to the process by which stations detect simultaneous transmissions.
2001 Peak Audio, Inc.
Fig. 2.1: The First Ethernet
Ethernet History and Background
Course AUD, Audio Networks
Page 20
node
node
node
node
2km
Fig. 2.2: Network Diameter
If the distance between the two end stations, known as the network diameter, is very large (greater than 2000 meters) there is a possibility that data received at one of those stations may be corrupted. If both devices begin to transmit at the same time, small data packets (the entire packet is already on the transmission line) collide in the middle of the line and get sent on to the other station as corrupted data. The transmitting station will never realize that a collision has occurred because the transmission was completed prior to the collision. As a result of this potential for problems, the minimum frame size in Ethernet is specified such that, based on the speed of propagation of electrical signals in copper media, an Ethernet device is guaranteed to remain in transmit mode, and therefore detecting collisions, long enough for a collision to propagate back to it from the farthest point on the wire from it. Ethernet standards do not allow data packets smaller than 512 bits to be transmitted. (This is standard for both huband switch-based networks, and for both 10BASE and 100BASE networks). Any data packets smaller than 512 bits are automatically padded to equal 512 bits.
Ethernet Packet Structure

Destination Address (6 Bytes) Source Address (6 Bytes) Protocol (2 Bytes) Payload (46-1500 Bytes) FCS (4 Bytes)
Fig. 2.3: Ethernet Packet
The figure to the left shows a standard Ethernet packet. The destination address is the MAC address of the receiving station, while the source address is the MAC address of the transmitting device. (MAC addresses will be discussed in a later section.) The protocol is a publicly registered numeric identifier issued by IEEE that signifies what kind of payload the receiver should expect. For example, the ID for CobraNet is 8819; and the ID for the Internet is 801.
Page 21
The FCS or frame check sequence provides a 99.9% probability that if 1 bit of data is corrupted, the frame check will detect the error This is sometimes referred to as a CRC, or Cyclical Redundancy Check.
The Ethernet Standard

Ethernet was standardized by the IEEE in 1980, as IEEE 802.3. The initial standard was based on 10mm 50 ohm coax, but this was followed quickly by many different variations: 10BASE-5: the original coax cable Ethernet 10BASE-2: used a thinner (5mm) 50-ohm coaxial cable instead of the 10mm used in 10BASE-5. Hence the name Thin Ethernet. The transmission rate is also 10Mbps. Maximum cable segment length is 185 meters. A standard T-type BNC connector (barrel connector) is typically used to attach two cable segments directly to the network interface card (NIC), thus allowing a daisy chain configuration in connecting the stations. A 10BASE-2 network supports a maximum of 30 stations per cable segment. 1BASE-2: the same as 10BASE-2, but running at only 1Mbps 10BROAD-36: a broadband implementation of Ethernet.
Out of all of these incarnations, only 10BASE-5 and 10BASE-2 saw much wide use. Both are uncommon now.
Page 22
node
node
node
node
Fig. 2.4: Backbone Topology
These early Ethernet networks were implemented in what is called a backbone topology. Although it looks simple and distributed, the backbone is the weak point in these networks. The problem is that a missing terminator or a faulty connection between a node and the cable can take down the whole network. This could be described as a distributed single point of failure, a worst case scenario in that not only is the cable a single point of failure, but when it fails, the location of the failure can be very difficult to find. The solution was to collapse the backbone. The resultant collapsed backbone topology is referred to as a star topology. And thus, 10BASE-T was born
node
node
hub
node
node
Fig. 2.5: Star Topology
10BASE-T 10BASE-T was introduced in 1990. It uses durable and inexpensive twisted pair cable and a repeater hub as a substitute for the coax backbone. This incarnation of Ethernet is still a shared media with the same maximum network diameter limitations. The packet structure also remained the same. Two pairs are required for each station: one pair for incoming traffic and one for outgoing. Home runs from each station to a central concentrator carry data to and from each station. With 10BASE-T, the vulnerable shared media portion of the network is now safely hidden in the closet and is not strewn across the office. In a 10BASE-T system, a wiring fault typically takes out network service only to a single station. Using this topology, stars may be connected to other stars to create much larger networks. In the star of stars figure to the left, note that there are no loops in the network. If a loop was introduced in the network, data would endlessly re-circulate, completely bringing down the network.
Fig. 2.6: Star of Stars
Page 23
The CAT3 cable is restricted to a maximum distance of 100m. To grow the network to the maximum 2km size supported, fiber is used. This is known as 10BASE-F.
node
node
hub
node
100BASE Ethernet
100BASE Ethernet supports data transfer rates of up to 100 Mbps (100 Megabits per second). Because it is 10 times faster than Ethernet, it is often referred to as Fast Ethernet. Like Ethernet, Fast Ethernet is based on the CSMA/CD media access method, it has the same packet size, and also the same format as the 10Base-X variety. In moving from 10BASE to Fast Ethernet, only the bit rate was increased. Therefore 10 times the number of bits may now fit on the transmission line. Rather than changing the minimum packet size requirements, the maximum allowed diameter was decreased in order to maintain accurate collision detection. This limits Fast Ethernet repeater networks to a 200 meter network diameter. FAST ETHERNET VARIATIONS Like Ethernet, Fast Ethernet also has a number of variations: 100BASE-T4: Fast Ethernet using all 4 pairs of a Category 3 cable. 100BASE-TX: Fast Ethernet using only 2 pairs in a Category 5 cable. This is the most common Fast Ethernet. 100BASE-FX: Fast Ethernet over multimode fiber.
node
node
node
hub
node
node
Fig. 2.7: 10BASE-F
Page 24
Gigabit Ethernet
The latest incarnation of Ethernet is Gigabit Ethernet or 1000BASE Ethernet. It also has the same packet format as the 10 and 100BASE varieties, but a larger minimum packet size. This allows the network diameter limitation to remain at 200 meters. The mechanism that makes a 200-meter network diameter possible is known as carrier extension. Whenever a gigabit network adapter transmits a frame less than 512 bytes long, the gigabit MAC sends out a special signal called a carrier extension (all the while continuing to monitor for collisions). The Ethernet frame and carrier extension will last for a minimum of 512 bytes, which is equivalent in time to transmitting 64 bytes at 100 Mbps. If the gigabit MAC detects a collision during this period, it reacts just like its conventional counterparts, sending a jam signal and telling the offending stations to back off and try again. GIGABIT ETHERNET VARIATIONS 1000BASE-SX: (S is for short wavelength) defines optical transceivers or physical layer devices for laser fiber cabling. Targeted for multimode fiber only, 1000BASE-SX transceivers are less costly than those found in products implementing the long wavelength specification 1000Base-LX: (L for long wavelength) also defines optical transceivers or physical layer devices for laser fiber cabling. 1000BASE-LX is specified for use on either multimode or single-mode fiber. 1000BASE-LH: (LH for long haul) is a multi-vendor specification where each vendor has a set of transceivers covering different distances. Although it is not an IEEE standard, many vendors are working to
Page 25
interoperate with IEEE 1000BASELX equipment. 1000BASE-T: Gigabit Ethernet over Category 5 cable. It uses all 4 pairs, and can be up to 100 meters long.
The Future
Wireless Ethernet or IEEE 802.11 is gaining popularity and is becoming more widely used. It is also a shared media and has a bandwidth of 11Mbps. 50 Mbps versions are in the works and should be available shortly. Powering Fast Ethernet devices over CAT5 cable has been accomplished by a number of manufacturers. Work is underway to standardize how this is done so that equipment from different manufacturers may interoperate. One application will be telephones which communicate over Ethernet using Voice over Internet Protocol (Voice/IP).
Audio Networks
Page 26
SELECTING THE CORRECT CABLE FOR A PROJECT Introduction

This section will provide information about designing and installing network cable plants. It will present a general overview of cable plant considerations that will help lead to a highquality network installation.
The Cable Plant

A typical Ethernet network cable path is shown in the figure to the right. The items that make up the cable plant include: cabling connecting nodes this can be CAT5 or fiber optic cable Wiring closet patch panels Station cables - the cable that runs from node to wall plate Wall plates - the data or information outlet close to the node
DCE Patch Cord - Store Bought, Stranded CAT5 Main Run - Field Terminated, Solid Core CAT5 Patch Panel or Wall Plate DTE Patch Panel or Wall Plate
It is considered good design practice to include the intermediate patch points as shown. This gives the cable plant operator flexibility in accommodating expansion and configuration changes. There are two main types of cables used in audio networks: CAT5 cable and fiber optic cable. The following sections will describe these cable types, as well as the issues associated with each.
Fig.3.1: Typical Cable Plant
CAT5
CAT5 is inexpensive unshielded twisted pair (UTP) data grade cable. It is very similar to ubiquitous telephone cable but the pairs are more tightly twisted. It should be noted that not all CAT5 cable is UTP. Shielded CAT5 also exists but is rare due to its greater cost and a much shorter distance limitations than UTP CAT5.
2001 Peak Audio, Inc. Selecting the Correct Cable
Page 27
DISTANCE LIMITATIONS On Fast Ethernet systems, CAT5 cable runs are limited to 100 meters due to signal radiation and attenuation considerations. A CAT5 run in excess of 100 meters may be overly sensitive to electromagnetic interference (EMI). CONNECTORS CAT5 cable is terminated with an RJ-45 connector. There are two different types of RJ-45s. There is the "bent tyne" connector, intended for use with solid core CAT5, and then there is the "aligned tyne" connector used with stranded CAT5 cable. Errors can occur when using incorrect cable/connector combinations. The diagrams to the left show an end on view of a single contact in a modular connector. The aligned tyne contact (far left) must be able to pierce through the center of the wire, therefore it can only be used on stranded wire. The bent tyne contact has the 2 or 3 tynes offset from each other to straddle the conductor; therefore, it can be used on solid or stranded wire. Cable openings in modular connectors can be shaped for flat, oval or round cable. CAT5 cable does not usually fit properly into connectors made specifically for flat cable. Cheap modular connectors may not have proper gold plating on the contacts, but instead only have a gold flash. Without proper plating, the connectors may quickly wear and corrode, causing unreliable connections. AMP makes quality modular connectors, but the secondary crimp point is located in a different position from everyone elses connectors. Figure 3.4 shows a standard crimper and an AMP plug. Point A is the primary crimp point, and should fold the primary strain relief tab in the plug down so that it locks against the cable jacket. At the opposite end of the plug, the contacts are pressed down into the individual conductors. The B secondary crimp point secures the
Fig. 3.2: Aligned and Bent Tyne Contacts
Fig. 3.3: Flat and Oval Cable Openings
Fig. 3.4: Standard Crimper and AMP Plug
Page 28
individual conductors so that they do not pull out of the contacts. AMP puts this crimp in a different location from all other manufacturers. If AMP connectors are used in a standard crimper they will either jam, bend or break the crimp die. If standard connectors are used in an AMP crimper, the die will usually break. Once either type of plug is properly crimped onto the wire, they are interchangeable and will work properly in any mating jack. Some plugs are made with inserts which guide the wires. These can make the job of properly assembling the connector easier. Some connectors made with inserts may also provide better performance than CAT5.
Fig. 3.5: Plug with Insert
TWISTS AND TERMINATIONS CAT5 cable consists of 4 twisted pairs of wires. One pair is used to transmit (pins 1 and 2) and another pair is used to receive (pins 3 and 6). The remaining two pairs are terminated but unused. Although only 2 of the 4 twisted pairs are used for Ethernet, it is important that all pairs be terminated, and that the proper wires be twisted together. Standards set forth by EIA/TIA 568A/568B and AT&T 258A define the acceptable wiring and color-coding schemes for CAT5 cables. These are different from the USOC wiring Standards used in telecommunications. When terminating CAT5 UTP cable, it is important that the natural twist of each pair be carried through as close as possible to the point of termination. EIA/TIA standard 568B requires no more than 1/2 inch be left untwisted for Category 5. More than 1/2 inch of untwisted cable will affect performance at high bit rates.
Pin 1 2 3 4 5 6 7 8
Signal
EIA/TIA 568A
EIA/TIA 568B
Ethernet X X X Not Used* Not Used* X Not Used* Not Used*
Transmit White/Green White/Orange + Transmit Green/White Orange/White or Green or Orange Receive White/Orange White/Green + N/A N/A Blue/White or Blue/White or Blue Blue White/Blue White/Blue
Receive Orange/White Green/White or Orange or Green N/A N/A White/Brown White/Brown Brown/White Brown/White or Brown or Brown
*These connections must still be terminated Table 3.1: Wiring/Color Coding for CAT5 Cable
Selecting the Correct Cable
Page 29
STRAIGHT-THROUGH VS. CROSSOVER CABLES Two types of CAT5 cables are typically used in a network: the straight-through cable and the crossover cable. The difference between the two has to do with how the conductors terminate to the RJ-45 connector at each end of the cable. The tables to the right show the RJ-45 connector "pin-outs" for CAT5 crossover and straight-through cables. A straight-through cable is used to connect a network device (or DTE), to a switch. DTEs, switches and other networking devices will be discussed in the next section. The transmit pins on the network device connect directly to the receive pins on the switch and vice versa (i.e., pin 1 to pin 1, pin 2 to pin 2, pin 3 to pin 3, etc. as shown in the graphic). Crossover cables are used to connect switches to other switches. In crossover cables, the pins are "swapped" at one end (i.e., pin 1 to pin 3, pin 2 to pin 6, pin 3 to pin1, and pin 6 to pin 2) to allow the transmit of one switch to connect to the receive of the other. It is very easy to tell the difference between a crossover cable and a straight-through cable by looking at the conductors in the RJ-45 connectors. If the wiring is identical at both ends, you are holding a straight-through cable, if it is different, you most likely have a crossover cable. Some switches employ an auto-select crossover feature. This allows the use of either a straight-through or a crossover cable in any port. The switch automatically senses which cable type is in use and adjusts the electronics to suit the cable.
Crossover Cable RJ-45 Pin RJ-45 Pin 1 RX+ 3 TX+ 2 RX6 TX3 TX+ 1 RX+ 6 TX2 RXPin 1
Straight-through Cable RJ-45 Pin RJ-45 Pin 1 TX+ 1 RX+ 2 TX2 RX3 RX+ 3 TX+ 6 RX6 TX-
Fig. 3.6: RJ-45 Connector Pin-outs
Page 30
62.5m Core
125m Cladding
CROSSOVER CABLES AND UPLINK PORTS Some hubs and switches contain uplink ports. These ports are intended to serve as a connection to another switch or hub. As such, the uplink port is wired to use a straight-through instead of requiring a crossover cable. On some switches and hubs uplink ports share their connection with an adjacent port, so be sure to read the manufacturers instructions for proper use.
Fiber Optic Cable

There are two varieties of fiber: single-mode and multimode, and both may be used in Ethernet network designs. Two fibers are needed to make an Ethernet connection: 1 fiber for transmit, and 1 for receive. Multimode fiber is built of two types of glass arranged in a concentric manner. Multi-mode fiber allows many "modes", or paths, of light to propagate down the fiber optic path. The relatively large core of a multi-mode fiber allows good coupling from inexpensive LEDs light sources, and the use of inexpensive couplers and connectors. Two sizes of multimode fiber are available. 62.5/125m is used primarily in data communications, and 50/100m is used primarily in telecommunications applications. The standard for transmission of 100Mbit Ethernet over 62.5/125m multimode fiber is called 100BASE-FX. 100BASE-FX has a 2 kilometer distance limitation. Single-mode fiber optic cable is built from a single type of glass. The cores ranges from 8-10 m, with 8/125m being the most commonly used. There is only a single path of light through the fiber. While single-mode fiber cable costs approximately the same as a multimode cable, the cost of the optical transmitters and receivers is significantly more for a single-mode installation than multimode. Single-mode fiber has a core diameter that is so small that only a
Fig. 3.7: Multimode Fiber
Fig. 3.8: Light Path Through Multimode Fiber
Fig. 3.9: Light Path Through Single-mode Fiber
Page 31
single mode of light is propagated. This eliminates the main limitation to bandwidth, but makes coupling light into the fiber more difficult. Although multi-mode fiber has a specific distance limitation of 2km, distance limitations of single-mode fiber vary according to the proprietary system in use. All are in excess of 2km. There is currently no Ethernet standard for single-mode fiber. CONNECTORS There are two common types of fiber optic connectors: SC and ST. The ST or "straight tip" connector is the most common connector used with fiber optic cable, although this is no longer the case for use with Ethernet. It is barrel shaped, similar to a BNC connector, and was developed by AT&T. A newer connector, the SC, is becoming more and more popular. It has a squared face and is thought to be easier to connect in a confined space. The SC is the connector type found on most Ethernet switch fiber modules and is the connector of choice for 100Mbit and Gigabit Ethernet. A duplex version of the SC connector is also available, which is keyed to prevent the TX and RX fibers being incorrectly connected. There are two more fiber connectors that we may see more of in the future. These are the MTRJ and MTP. They are both duplex connectors and are approximately the size of an RJ-45 connector.
Duplex SC
Fig. 3.10: Fiber Connector Types
Condition
<2kVA
2-5kVA
>5kVA
Unshielded power lines or electrical 5 in. (12.7 equipment in proximity cm) to open or non-metal pathways Unshielded power lines or electrical 2.5 in. (6.4 equipment in proximity cm) to grounded metal conduit pathway Power lines enclosed in a grounded metal conduit (or equivalent N/A shielding) in proximity to grounded metal conduit pathway Transformers and electric motors Fluorescent lighting
12 in. (30.5 cm) 24 in. (61 cm)
6 in. (15.2 cm)
12 in. (30.5 cm)
Cabling and Network Performance

A number of factors can degrade the performance of your Ethernet network, and among these is a poor cable plant. Cabling problems and a susceptibility to EMI can actually lead to packet loss. The following sections present cabling considerations that will help to ensure a high-quality cable plant installation.
6 in. (15.2 cm)
12 in. (30.5 cm)
40 in. (1.02 40 in. (1.02 m) m)
40 in. (1.02 m)
12 in. (30.5 12 in. (30.5 cm) 12 in. (30.5 cm) cm)
Table 3.2: Proximity Specifications
Page 32
Similar to audio cabling, there are certain proximity specifications to be aware of when designing your network cable routes. The adjoining chart lists some CAT5 proximity guidelines. For fiber optic cable runs, proximity is not a concern due to fiber's inherent immunity to EMI and RFI. CAT5 AND CABLE TIES Another factor that can degrade the installation quality is snug cable ties. Ties should never be pulled tight enough to deform the outer jacket of the UTP cable. Doing so produces a slight change in the cable impedance at the point under the tie, which could lead to poor network performance. If tight ties are used at even intervals down the cable length, the performance degradation is even worse. PULL FORCE AND BEND RADIUS A common myth is that fiber optic cable is fragile. In fact, an optical fiber has greater tensile strength than copper or steel fibers of the same diameter. It is flexible, bends easily and resists most of the corrosive elements that attack copper cable. Some optical cables can withstand pulling forces of more than 150 pounds! The fact is, Category 5 cable may be more fragile than optical cables: tight cable ties, excessive untwisting at the connector, and sharp bends can all degrade the cables performance until it no longer meets Category 5 performance requirements. While fiber may have a reputation for being more fragile than it really is, it still has limitations, and as such, care should be taken when installing both CAT5 and fiber optic cables. Here are some guidelines for CAT5 and fiber optic bend radius and pull force limitations: For CAT5 Cable: All UTP cables have pull force limitations much lower than those tolerated in the audio industry. If more than 25 pounds of force is applied to CAT5 cable during installation, it
Page 33
may no longer meet specification. Like most audio cables, UTP cables also have minimum bend radius limitations. Generic CAT5 allows a minimum bend radius of 4 times the cable diameter or 1" for a 1/4" diameter cable. Unless specified otherwise by the manufacturer, it is fairly safe to use this as a guideline. Note that this is a minimum bend radius and not a minimum bend diameter. For Fiber Optic Cable: The bend radius and pull force limitations of fiber vary greatly based on the type and number of fibers used. If no minimum bend radius is specified, one is usually safe in assuming a minimum radius of 10 times the outside diameter of the cable. For pulling force, limitations begin at around 50 lbs and can exceed a more than 150 pounds. In general, it is recommended that you check with the fiber manufacturer for specifications on the specific cable used in your installation.
Page 34
Sources for More Information

One good way to keep up with current happenings in the field of cabling is to subscribe to Cabling Installation & Maintenance magazine. They are also a good source for training videos and reference books.
cim.pennwellnet.com/home/home.cfm
BiCSi (pronounced bic-see) publishes a Telecommunications Distribution Methods Manual which serves as the reference for their Registered Communications Distribution Designer Certification exam. They also Certify 3 levels of Installers. They offer training courses, videos, and books, and hold an annual convention.
www.bicsi.org/
This paper by Belden shows why a "neat" installation may not be a good idea, and defines some of the important tests done on UTP data cabling. It also shows that neatly bundling cables together (as is considered good practice in the audio industry) may actually degrade the performance of UTP cables.
http://bwcecom.belden.com/college/ Techpprs/ieacectp.htm
While it's not your typical FAQ page, the Data Communications Cabling FAQ provides a good source for technical information on cabling, connectors, standards and testing. It also provides a thorough listing of contact information for manufacturers and standards organizations.
alpha.mhpcc.edu/net_class/net_lan/ faq.html
Audio Networks
Page 35
SELECTING THE CORRECT HARDWARE FOR A PROJECT Introduction

There are 3 classes of items that make up a network. These are: cable plant data terminal equipment (DTE) data communications equipment (DCE) In the graphic to the right, dashed lines indicate the network cable plant, and shading represents DTEs. The DCEs in this case are made up entirely of Ethernet switches. Although this section will introduce data terminal equipment, it will focus primarily on data communications equipment.
Switch Switch
Fig. 4.1: Audio Network
Data Terminal Equipment

Data terminal equipment refers to any devices that produce and/or consume data on a network. DTEs are sometimes referred to as network nodes, and include things like PCs and network printers, just to name a few. CobraNet devices and other networked audio devices are also data terminal equipment. Well discuss audio networking devices in greater detail in a later unit.
Data Communications Equipment and Ethernet Hardware

Typically, data communications equipment is classified as either Workgroup Equipment or Interconnect components. Workgroup equipment includes devices such as hubs and switches, while interconnect components include routers, bridges and core switches. NETWORK INTERFACE CARD A network interface card, or NIC, is used to enable a PC to connect to and access a network. NICs are network physical layer specific, meaning that a PC containing an
2001 Peak Audio, Inc. Selecting the Correct Hardware
Page 36
AppleTalk NIC cannot function on an Ethernet network. REPEATER HUBS A repeater hub is a multi-port Ethernet device that receives a signal, re-clocks it, boosts it and sends it out all ports. A loose audio analogy to a simple repeater hub would be a mix minus system. A data signal arriving at the receive pair of any port is electrically regenerated and reproduced out all other ports on the hub via the transmit pairs. As a result, hub networks may only be wired in a star topology. If a ring topology were used, the data would continue circulating endlessly! Because the data received by a repeater hub is simply transmitted to all devices on the network, repeater networks must share bandwidth. Bandwidth is the amount of data that can be transmitted within a fixed period of time, and is typically expressed in Megabits per second or Mbps. On a Fast Ethernet network, this means that there is 100Mbps of bandwidth available network-wide when using repeaters. Repeater hubs operate in half-duplex mode. They cannot transmit and receive data simultaneously. A repeater hub network is considered to be a collision domain. A collision domain is a half-duplex Ethernet system whose elements are all part of the same timing domain. This means that if two or more devices on a repeater network attempt to transmit at the same time, a collision occurs. When this happens both devices cease transmission, wait a random period of time and then attempt to retransmit the data.
Port 1
Port 2
Port 3
Fig. 4.2: Repeater Hub
Hub
Fig. 4.3: Star Topology
Selecting the Correct Hardware
Page 37
A protocol is required to enable each device in the collision domain to: share the available network bandwidth detect when another device is transmitting detect when a collision has occurred The protocol used is known as CSMA/CD or Carrier Sense Multiple Access with Collision Detect. Because there is no designated control station on an Ethernet network, each device must operate independently. The CSMA/CD protocol equips each device on the network with the same set of rules, allowing all stations connected to the shared media to cooperate. Even though signals travel very quickly on an Ethernet network, they still require a finite amount of time to propagate over the entire network. The longer the cables, the longer the time it takes the signals to travel from one end of the system to another. For CSMA/CD to work properly, maximum cable lengths and overall network diameter limitations must be observed. Network diameter is defined as the longest cable distance between any two DTEs on the network. With allowances for propagation times through up to two hubs and the receiver, Fast Ethernet typically supports a maximum network diameter of just over 200 meters on a repeater hub network. Network designers and administrators are quickly abandoning the use of repeaters. This is due primarily to the dramatic decrease in the cost of Ethernet switching hubs. For this reason, the rest of this course will focus primarily on networks utilizing switching hubs.
Component
Round Trip Propagation Bit Periods
Optical Fiber (single mode and multimode) CAT5 cable Receiver Class I Hub Class II Hub
1.000/meter 1.112/meter >=100 >=140 >=92
Table 4.1: Propagation time for selected network components
Page 38
node
node Collision Domain
hub
node
SWITCHING HUBS Based on Ethernet bridging technology, a switching hub (or switch) is a multi-port device that filters and forwards data packets between devices on a network. Unlike a standard repeater hub, a switch is able to read the destination address of each data packet and then forward the packet to the correct port. This intelligence in the switch means that a given device receives only those packets addressed to it. Another benefit is that switches can transmit and receive data simultaneously. Switches are full-duplex devices, and each port operates independently. For example, if the network operates at 100Mbps, each switch port can handle 100Mbps of data in each direction simultaneously. This provides a total bandwidth of 200Mbps over a single link! While each link between switches still has a bandwidth limit, the total bandwidth available in the network is much larger in a switched network. The overall network bandwidth can be easily increased by simply adding more switches to the network. Because each switch port operates independently in a full-duplex configuration, collisions do not occur on switched networks. Therefore, the CSMA/CD protocol used for repeater hubs is not required. This lifts the network diameter limitation, allowing the construction of larger networks. When used with repeaters, switches enable the creation of separate collision domains. Although the propagation of the collision is interrupted, switches are still able to pass data from one collision domain to another. Switches enable the connection of LANs that operate at different speeds, as well as other networking technologies such as Gigabit Ethernet and ATM.
node
switch
switch
node
node
hub
node Collision Domain
node
Fig. 4.4: Full Duplex
Page 39
How does a switch accomplish all of this? A switch is made up of four main parts: ports a lookup table (LUT) FIFO queues switch fabric The ports on a switch can each operate in fullduplex mode, and multiple ports can be carrying on conversations simultaneously. This not only increases the amount of bandwidth available on the network, but makes a switch much more efficient than a repeater hub. The lookup table (LUT) is what gives a switch the ability to send packets out to selective ports. The LUT contains a list of each switch port and the MAC address of the device(s) attached to each port. The MAC address or media access control address is an IEEE issued hardware address that uniquely identifies each node on a network. As network nodes send packets, the switch reads each incoming packets source MAC address. If the address is not contained in the lookup table, it is added so that the switch will later know where to send data addressed to that device. Over time, the switch learns the MAC addresses of all the devices attached to each of its ports. Switches can also unlearn. If no packets are received from a particular source address after a period of time (typically five minutes), or it is suddenly found on a different port, the source address is deleted from the LUT. This allows the flexibility to move nodes to different segments of the network without the worry that the lookup table may contain inaccurate information.
LUT
Switch Cloud
1 Tx Rx Tx
2 Rx
3 Tx Rx
Fig. 4.5: Switching Hub
Page 40
node
node
switch
node
node
node
Switches also have the ability to distinguish between different types of addresses: If an incoming packets destination address is a unicast address (point-topoint), the packet is only sent out the port to which the destination device is connected. If an incoming packets destination address is a multicast address (one-tomany), the packet is sent out all ports.
node
node
switch
node
Fig. 4.6: Unicast Address
If the destination address is not contained in the LUT, the packet is sent out all ports. This is known as flooding, and guarantees that the packet with the unknown destination address will reach the correct node. The first-in-first-out (FIFO) queue provides short-term buffer memory for each port. Each packet is queued here prior to transmission, which allows the switch to handle the transmission of multiple packets to the same destination device. The switch fabric is the means used to move packet data from the input of one port on the switch to the output of one or more ports. The LUT and switch fabric work together to route packets through the switch. The LUT knows where the packets need to go and the switch fabric gets them there. Managed vs. Unmanaged Switches
node
node
switch
node
node
node
node
switch
node
node
Fig. 4.7: Multicast Address
Switches can be managed or unmanaged. Unmanaged switches are plug and play, and contain few, if any, configuration options. Typically, any configuration on an unmanaged switched is performed through the use of dipswitches. Managed switches offer a great deal of configurability options, as well as several fault tolerance options such as Spanning Tree Protocol, Link Aggregation and Meshing. Managed switches also provide network administrators with information about the operation of the switch itself, including traffic reports, bandwidth utilization and port
Page 41
configuration information that can be used for analysis and troubleshooting. The interface for switch configuration is typically a web browser or a telnet interface. Many of the advanced networking topics we will discuss later, are only available with the use of managed switches. Switched Network Topologies Switched networks may be wired in a star topology similar to repeater networks. Simply replacing the hub with a switch automatically increases the available bandwidth from 100Mbps network wide to 100Mbps per link in each direction. Switches may also be wired in a ring, as long as the switches participating in the ring support protocols that prevent loop conditions. These include Spanning Tree Protocol, aggregation, or Hewlett Packards proprietary Meshing technology, which are only available on managed switches. HUBS VS. SWITCHES Because the cost of Ethernet switching hubs has decreased dramatically over the last few years, there really is no reason not to use switches! Switches allow the construction of large networks with great distances between the nodes, they allow fault tolerance to be incorporated into the network design, can provide great channel capacity and even greater channel capacity, like Gigabit. BRIDGES AND ROUTERS Although this course focuses primarily on the use of switches for audio network design, some background information on interconnect components is worthwhile. Routers connect LANs. Although they are protocol dependent, meaning they can only be used to connect LANs using the same Layer 3 networking protocol, routers are independent of Layer 2 and often contain a combination of physical port types (i.e., Ethernet, token-ring, ISDN, etc.). A router first strips the Layer 2
Switch
Switch
Switch
Switch
Fig. 4.8: Ring Topology
Page 42
headers off of all packets it receives, and examines the Layer 3 header to determine how to route the packet. It then generates a new Layer 2 header as appropriate for the destination physical layer and sends the packet. A bridge is two-port device that is also used to connect LANs. Bridges are protocol independent, and work only at Layer 2. For example, an Ethernet bridge only sees Ethernet packets and doesnt care about the Layer 3 protocol carried within the packet. Bridges do not analyze packets - they simply forward the data, and as such, they perform faster than routers. Ethernet switches, which are based on bridging technology, used to be referred to as multi-port bridges. But because switching technology has advanced tremendously in recent years, bridges arent so common anymore. In fact, bridges are now sometimes referred to as two-port switches.
Establishing Link
Whether it be for sending audio signals over long distances, providing immunity from EMI, or due to a lack of cable space in existing conduits, many network designs require the use of fiber. To produce a successful design, it is important to understand how network devices such as switches, hubs and media converters interact with one another.
Normal Link Pulses 10Mbps Hub 10/100 NIC
Fig. 4.9: Normal Link Pulses
10MBPS NETWORKS When 10Mbps network devices are not transmitting data, the transmission line becomes idle. If data transmissions are not resumed within a certain period of time, the device begins transmitting Normal Link Pulses (NLPs). These pulses allow the presence of a non-transmitting 10Mbit device to be recognized by other devices on the network (i.e., enables link), and also serve to distinguish a 10Mbps device from 100Mbps or 10/100 devices.
Page 43
100MBPS DEVICES Unlike their 10Mbps counterparts, 100Mbps devices exhibit activity even when no data is being transmitted. This activity is called a carrier, and is a distinguishing characteristic of a 100Mbps network device. AUTO-NEGOTIATING DEVICES CobraNet devices and many other 10/100 network devices go through a process called auto-negotiation before establishing link. Auto-negotiation is a low bit rate form of communication during which one device tells another device if it is capable of full- or halfduplex operation and whether to connect at 10Mbps, 100Mbps or Gigabit rates. This information is conveyed using Fast Link Pulses (FLPs), which are simply a sequence of Normal Link Pulses that come together to form a message. If the auto-negotiation process results in a 10Mbit connection, the network devices transmit NLPs when idle. If a 100Mbps connection was negotiated, carrier signals are transmitted. Any device capable of auto-negotiation also implements parallel detection. Parallel detection enables link to be established with a non-negotiating, fixed speed network device prior to the detection of Fast Link Pulses. The state diagram below shows how network connections between devices of differing capabilities are established. Notice that a device can never parallel detect to a fullduplex link. This is important, as autonegotiation does not take place over fiber optic cable. Over fiber, carrier signals are used to establish link, which means that if a fullduplex connection over fiber is required, this must be manually configured.
100Mbps Hub
Carrier
10/100 NIC
Fig. 4.10: Carrier
10/100 Switch
Fast Link Pulses
10/100 NIC
Fig. 4.11: Auto-negotiation
no link Carrier FLP LP
100 HD
10 HD 100 FD
negotiating
10 FD
Parallel Detect Key to Abbreviations: 100: 10: HD: FD: 100Mbit 10Mbit Half Duplex Full Duplex
Fig. 4.12: Link State Diagram
Page 44
AUTO-NEGOTIATION EXAMPLES In the example shown in Figure 4.9, a 10/100 NIC is connected to a 10Mbps repeater hub. Because the NIC is capable of connecting at two different bit rates, it has to determine the speed at which it should connect to the hub. When the devices are connected, the 10Mbps hub begins transmitting Normal Link Pulses. The 10/100 NIC sees these pulses and realizes that it must make a 10Mbps half-duplex connection (remember that hubs are halfduplex devices). Link between the two devices is then established. In Figure 4.10 the NIC must again determine the speed at which it should connect to the hub. When the cable between devices is connected, the 100Mbps hub begins transmitting a carrier. The 10/100 NIC sees the carrier and recognizes that it must connect to the hub at a 100Mbps rate, half-duplex. Link is then established. MEDIA CONVERTERS A media converter is essentially a two-port repeater hub that accepts one form of media and converts it to another. Most commonly, copper to fiber. For the purposes of using media converters in audio network designs, we will categorize media converters as: 1. simple media converters 2. intelligent media converters A simple media converter (SMC) has no intelligence. It simply converts light to an electrical pulse. A simple media converter cannot pass or detect Fast Link Pulses and is therefore incapable of auto-negotiation. As a result, it is incapable of establishing link with an auto-negotiating device on its own.
TX Active Indicator
Fiber Code Converter
Copper
FX Active Indicator
Fiber Code Converter
Copper
Fig. 4.13: Simple Media Converter
TX Active Fiber N e g o t i a t e Copper
Code Converter FX Active
Fiber
Copper
Code Converter
Fig. 4.14: Intelligent Media Converter
Page 45
Unlike a simple media converter, an intelligent media converter (IMC) is capable of generating and interpreting Fast Link Pulses. In addition to the electronics required for the media conversion, an IMC contains a "negotiation" block that allows: auto-negotiated or manuallyconfigured connection speeds full- or half-duplex operation
Network Addressing
Weve already discussed MAC addresses, the unique identifiers assigned to all Ethernet devices. The TCP/IP protocol is a higher-level networking protocol and has its own system of addresses called IP addresses. So if TCP/IP data has its own addressing scheme, how does this relate to Ethernet and MAC addresses? Each Ethernet packet carries a payload that contains the network protocol information. TCP/IP, like other network protocols, has its own addressing scheme and data structure. Thus, contained within the Ethernet payload is another series of fields containing IP source and destination address information, as well as another payload containing TCP data. This nesting within the Ethernet payload continues until the protocol(s) being transported is completely defined. The most important thing to understand is that the networking protocol in use is completely independent of Ethernet. Ethernet is simply the delivery system, and it doesnt know or care about the networking data that its carrying in its payload. This enables Ethernet to carry many different networking protocols on the same network. It is left up to the networking software running on the DTEs to understand the networking protocol information extracted from the Ethernet payload field.
Ethernet Payload
Pre- Dest Src P amble Add Add Dest Src P Add Add IP Payload CRC
Fig. 4.15: Ethernet Packet Carrying IP Data
Page 46
IP Address is: 192.168.0.2 Binary Equivalent: 11000000.10101000.00000000.00000010
IP ADDRESS An IP address is a unique identifier for a device on a TCP/IP network. Within an isolated network, it is possible to assign IP addresses at random as long as each address is unique. However, connecting a private network directly to the Internet requires using registered IP addresses (called Internet addresses) to avoid duplicates. An IP address is a 32-bit numeric address (four sections of 8 bits each) written in dotted decimal notation. The number in each section can range from 0 to 255. The address consists of 2 parts, the network number (or as it is more commonly referred to today, the network prefix) and the host address. The host address is the unique address of the network node, while the network number identifies the network on which the host resides. Each host on a network shares the same network number but must have a unique host address. Likewise, any two hosts can share the same host address as long as they have different network numbers. The InterNIC Registration Service originally assigned Internet addresses from three classes, where each class fixes the boundary between the network number and the host address at a different point within the 32-bit address. This is known as classful addressing. Class A - 16 million hosts on each of 127 networks (1.xxx.xxx.xxx - 126.xxx.xxx.xxx) Class B - 65,000 hosts on each of 16,000 networks (128.0.xxx.xxx - 191.255.xxx.xxx) Class C - 254 hosts on each of 2 million networks (192.0.0.xxx - 223.255.255.xxx) Unfortunately, no one expected the Internet to grow as quickly as it did, and the limitations and inflexibility of this method of addressing were unforeseen. Adding another level to the address hierarchy, the subnet, helped to temporarily overcome many of these shortcomings.
Fig. 4.16: IP Address Example
Network Number
Host Address
Fig. 4.17: IP Address Format
Page 47
SUBNETS AND SUBNET MASKS A subnet is a portion of a network that shares the same prefix. It enables a network administrator to further divide the host part of the address. Previously, another network number had to be requested from the Internet before a new network could be installed at a site. When a network is divided into subnets, a part of the host address is used to identify a particular subnet. Lets look at an example using the IP address shown in Figure 4.16. The network number is: 192.168.0 or11000000.10101000.00000000 in binary and the host address is: 2 or00000010 in binary If the network needs to be divided into 13 subnets, the first 4 bits of the host address (0000) would be reserved for identifying the subnet. This would allow for a total of 16 possible subnets, as subnets can only be created in blocks of powers of two. The first 4 bits of our host address could then be changed to indicate the subnet number: nnnn0010 the network is divided into subnets, the IP nodes must be told where the division between the extended network number and the host address occurs. This is done using a subnet mask. The subnet mask is the network number plus the bits reserved for identifying the subnet. To come up with a subnet mask, the bits for the network number, including the new subnet number, are all set to 1. In our example the subnet mask would be:
Once
Network Number
Host Address
Network Number
Subnet
Host #
Extended Network Prefix
Fig. 4.18: Subnet Address Format
11111111.11111111.11111111.11110000 or255.255.255.240.
2001 Peak Audio, Inc. Selecting the Correct Hardware
Page 48
CLASSLESS ADDRESSING Today, the use of classful addressing on the Internet has been replaced by a method of routing based on subnetting called CIDR or Classless Inter-Domain Routing. CIDR eliminates Class A, B and C addresses and replaces them with the concept of an extended network prefix. The standard syntax for CIDR is to write the network prefix in dotteddecimal notation, followed by a slash and the number of bits in the prefix. Including the prefix length in the notation eliminates the need to specify both an IP address and a subnet mask when configuring a host. It is important to note that not all hosts support classless routing. In fact, Microsoft TCP/IP still uses Classful addressing. When networking is set up on a Windows PC, a subnet mask expressed in dotted decimal notation is required. IP ADDRESSES AND AUDIO NETWORKS The most common IP address settings for audio networks are shown in the table to the left. Once an IP address is assigned, audio network devices can communicate using higher-level protocols including the SNMP management protocol.
Designing and Building a Network

Device Type Audio Devices Ethernet Switches Other Devices IP Address Range 192.168.100.1 through 199 192.168.100.200 through 219 192.168.100.220 through 254
All subnet masks should be 255.255.255.0 Table 4.2 Common IP Addresses for Audio Networks
Before plunging into a design, lets first review the switched network design rules: Dont exceed the available bandwidth Dont exceed distance limitations: o No farther than 100 meters over Category 5 cable o No farther than 2km over multimode fiber No loops without Spanning Tree, aggregation, or Meshing capable switches
Page 49
The first step in network design is identifying the proper components. For audio networks, the best bet is to use Ethernet switches. They provide greater scalability and flexibility than repeater hubs and the unmanaged variety is comparably priced. Next, make the physical connections between devices using CAT5 cable or fiber as necessary. Choosing the proper cable requires that you know the physical distances between the devices such that cable distance limitations are not exceeded.
ASSIGN IP ADDRESSES TO PCS IP addresses are assigned on Windows compatible PCs through the Network Properties. Depending upon which incarnation of Windows is used, the exact setup procedure (i.e., menu choices, window tabs, etc.) differs. CONFIGURING THE SWITCH Switch configuration is usually only necessary if you are using managed switches in your design. In most cases, using the factory defaults will be sufficient, and only an IP address and subnet mask need be assigned. The configuration steps listed below assume that a Windows PC is being used for switch configuration.
Connect a null modem cable between a PC and the switchs COM port and use HyperTerminal to access the switch configuration menu. Safe settings for your HyperTerminal session include: 9600 baud, 8 bits, no parity, 1 stop bit (this is also known as 8N1) Once connected, the settings and configurations options are available, and the IP address and subnet mask can be configured. Most switches also provide the option to enter a name for the switch, a location, and a contact name, although these arent required.
Page 50
TEST CONNECTIVITY Probably the quickest and easiest way to determine that all the devices are connected to the network is to ping the devices. Ping is a utility that is used to determine whether or not a specific IP address is accessible, which makes it very handy for troubleshooting network connections. When the ping command is issued, an ICMP (Internet Control Message Protocol) packet is sent to the specified address and a reply is expected. If a reply is received, this indicates that the remote device is reachable over the network. If no reply is received, this indicates a problem with communication over the network.
To issue a ping command, open the Command Prompt, and type:

ping xxx.xxx.xxx.xxx
where: xxx.xxx.xxx.xxx is IP address of the device you wish to contact
Page 51
Recommended Reading
You dont need to be an authority on Ethernet in order to successfully design audio networks. However, if youd like to learn more, the following book is a good place to start: Charles E. Spurgeon, Ethernet - The Definitive Guide, OReilly, 2000
If youre looking for a quicker read, there are some useful websites that present very good overviews of Ethernet and other related topics:
Techfest.com, TechFest Ethernet Technical Summary, www.techfest.com/networking/lan/etherne t.htm Optimized Engineering Corporation, Optimized Engineering Technical Compendium, 1995-2000, www.optimized.com/COMPENDI/index.h tml Chuck Semeria, Understanding IP Addresses: Everything You Always Wanted to Know, www.3com.com/nsc/501302.html
Audio Networks
Page 52
ROUTING AUDIO ON A NETWORK Introduction

There are a number of methods used for delivering audio over a data network. The method used depends on the capabilities of the network and the application requirements. Primary network considerations include bandwidth and reliability. Primary application considerations include audio capacity, quality and latency.
File Transfer then Play

The file transfer then play method of audio distribution involves an audio server and playback device. In a file transfer then play scenario, all operations are typically initiated by the playback device. At the direction of the player, an audio file stored on the server is transferred to a playback device creating a copy of the file in the playback device's local storage (i.e. hard disk drive). The playback device then plays the file copy locally. The file transfer may be accomplished with any network file transfer protocol such as FTP (File Transfer Protocol), HTTP (Hyper Text Transfer Protocol), NetBIOS or the proprietary protocol used by Napster. The file format used determines the audio quality in this scheme. Raw PCM data such as found in the popular WAV file format may offer the best quality. Formats utilizing lossy compression such as MP3 offered reduced audio quality with reduced bandwidth requirements. Network addresses and a file name determine the routing of audio on the network.
Routing Audio
Page 53
Separate operations of file transfer and playback can be automated into a single process, and the combined process can be scheduled or activated from a user interface or contact closure. The file transfer occurs at a maximum rate sustainable by network. Playback occurs on the local storage device once the file transfer is complete.
ADVANTAGES Using file transfer then play, audio will always play reliably. Data integrity is insured by checks during file transfer; integrity checks are an integral feature of any reasonable file transport protocol - or network protocol for that matter. Network delivery delays may postpone file play but cannot interrupt playback once it has commenced since the file is played from local storage. Not network resources are involved in the playback portion of the process
The audio can be of an arbitrarily good quality regardless of network performance. The network bit rate is decoupled from the audio bit rate. If the audio bit rate is greater than the network bit rate, youll wait a long time before playing but you will play, and you will play at the desired audio bit rate. It will sound nice. Multiple playbacks are also possible. Once the audio data is in local storage, it can be played back numerous times without having to go back to the network - perhaps a disadvantage if you happen to be a copyright holder.
DISADVANTAGES One disadvantage to using transfer then play models is that you must wait for the download to complete before the playback begins. The wait time is proportional to the length of the program and speed of network. Slow network and/or high quality audio require patience!
2001 Peak Audio, Inc. Routing Audio
Page 54
Substantial local storage may be required. You must have enough local storage to store an entire program. This calls for a hard disk drive in each playback device for most applications. There is no streaming capability. File transfer-based audio routing models are not suitable for live programming. You cant transfer a file if you dont know how big the file is.
APPLICATIONS Engineers working in post-production facilities typically download a piece of a project from a server, listen to it, edit it, then upload it again.
Evil media distribution companies like MP3.COM and Napster use the file transfer then play model for media distribution. It is also the model proposed by some for video on demand via broadband to the home. Set top boxes with hard disk for video file storage are currently available. In computer-based training scenarios students can download lectures and demonstrations and learn at their own pace. This method may be used for scheduled announcements in transportation systems (i.e. airport, rail, etc.).
Play Over Network - Pull Model

This pull model is a refinement of transfer then play. Using this pull model, playing can begin before the file is fully transferred. In fact, play can begin once the first block of data is received by the player though many players will buffer multiple blocks before play begins as a hedge against network disruptions. This is referred to as a pull model because the player controls the data flow from the server. The player requests additional blocks of data from the server as it consumes buffered data during playback.
Page 55
ADVANTAGES Play can begin while the file transfer is still in progress.
This method requires only moderate local storage.

DISADVANTAGES The play over network pull model is subject to network disruption. Once playback is started, the network and server must be able to deliver data at the audio bit rate. Delivery delays can result in interruption of playback.
Audio quality is limited by network bandwidth. If network bandwidth is limited to 56Kbits/second you will not be able to play a 128Kbit/second MP3 file using this method.
APPLICATIONS A simple distributed music jukebox can be built from a bunch of MP3s on a shared volume of a server. Playback can be performed at workstations using a playback application such as WinAmp.
If the server and network are of adequate performance, a post-production editor may work directly with the server files saving the time and storage requirements of making local copies. This mode of working amounts to using the pull model.
Play Over Networks - Push Model

In this method, the player opens a connection to a streaming server. The server sends the data at the requested audio playback bit rate over the connection. This is called a push model because the server controls data flow out to the player. The server transmits data at the audio bit rate. The server assumes the player will play the material at the audio bit rate.
Routing Audio
Page 56
The initial request from player to server determines the data routing path, and may also request a specific audio quality. The data can be sent from server to player via a "reliable" transport such as TCP or, more commonly, via an "unreliable" transport such as UDP. Packet loss is normal on the public Internet, and thus, the player may request that the server retransmit lost data. It a reliable transport is utilized, retransmission is handled by the transport. The advantage of using an unreliable transport is that if retransmission is not received or is received too late, the player can mute or disguise the error rather than suspending playback.
ADVANTAGES The push model requires a constant bandwidth, and can adapt to non-ideal network conditions. The server determines the delivery rate. Pre-buffering is typically done at audio bit rate, which increases the time to play, but ensures smooth network performance. The server may adjust the requested bit rate dynamically in response to server or network congestion.
The push method supports live streaming. We are no longer talking about a file transfer in this scheme. The server is sending data to the player. The data can come from a file or it can come live from a soundcard input on the server.
DISADVANTAGES The main disadvantages to this method are lackluster audio performance due to aggressive lossy data compression and high latency due to long buffering. However, these traits are not inherent in the push model delivery mechanism. Both compression and buffering are employed to overcome performance problems on the public Internet where the push model is most widely deployed.
Page 57
Data compression reduces the bandwidth requirements. Bandwidth is limited through the public Internet by two factors: 1. The users connection to the Internet is typically a slow dialup connection. Even broadband connections such as DSL or cable rarely exceed 1Mbit/second. One channel of raw PCM CD quality audio requires approximately 700Kbit/second. 2. Although a servers connection may be fast The server may be serving a large number of players resulting in a limited amount of bandwidth allocated to each player Latency is added purposely because it improves resilience against inconsistent network performance. By holding more data buffered in reserve, playback can withstand delayed delivery or data delivered out of order. With adequate buffering a player will have time to request and receive retransmission in the event of dropped data. Delayed delivery, out of order delivery and dropped packets are common on the public Internet.
Synchronous Streaming
Synchronous streaming is the least complicated means of transporting digital audio. Synchronous streaming is used by common point to point digital interconnect such as AES3 ADAT optical and MADI. All DACs and ADCs feature a synchronous interface. The public telephone network is a synchronous streaming distribution system.
ADVANTAGES Synchronous distribution systems are specialized, purpose-built systems. A purpose built system can be the best solution for a well focused application.
Routing Audio
Page 58
The simplest solution for short haul point-topoint distribution such as connecting an ADC to a DSP within an audio product or connecting a CD player to a DAT for duplication.
DISADVANTAGES Synchronous streaming does not efficiently support a mixture of services. Carrying asynchronous data on a synchronous transport is inefficient, and it does not accommodate bursty traffic patterns.
Because of their inflexibility, synchronous systems tend to be purpose-built. Purposebuilt communications systems are not cost effective. A more general data transport reaches a wider market and benefits from economy of scale. A more general data transport is often simpler and cheaper in that only common denominator features are included in commodity technologies.
Isochronous Streaming
Isochronous operation is similar to the push model but, in isochronous streaming, the network is required to deliver quantifiable performance. This allows for reduction in latency and/or increase in audio quality. Network isochronous performance is quantified in a service agreement on the connection between server and player. The service agreement specifies things such as bandwidth, delivery delay and delay variation (jitter). Although in the push model, the data rate is controlled by the server, this does not necessarily constitute clock delivery to the player. Typically both server and player will run jam-sync to their own local clock reference. Slight rate differences in the clocks are handled by introducing discontinuities in the data. The isochronous streaming method attempts to emulate a synchronous transport and as such includes a common clock between server and player.
Page 59
Synchronous Input
0
Packetizer
1
Isochronous Data Packets
ISOCHRONOUS TRANSMITTERS AND RECEIVERS Data arriving at the synchronous input (from an ADC, for instance) is partitioned into packets. The packets are then time-stamped and are queued for transmission. At the receiver, the packets arrive over the transport and queued for serilization. Packet timestamps are examined to determine which packet to deserialize at what time. This buffered receiver scheme can handle out of order as well as deferred receipt of data packets.
Transport
Fig. 5.1: Isochronous Transmitter
Both the transmitter and receiver have clocks which drive their synchronous interfaces. The clocks must be synchronized by some means. Examples of clock synchronization schemes include:
Transport
1
Isochronous Data Packets
0 1
1 0
0 1 0
Serializer
Synchronous Output
Fig.5.2: Isochronous Receiver
Out of band clock distribution such as a separate wire or GPS receivers. Data queue servo schemes where the receiver clock is disciplined based on data receipt timing or receive queue fullness. In band where a special type of data packet is used to carry clock information. The receiver clock operates with an offset with respect to the transmitter clock. This offset gives the system time to packetize the data and transport it. The minimum offset possible is the time taken to produce a packet of data, plus the time taken to transport the packet.
ADVANTAGES Low latency is possible because network performance is constrained by the isochronous service agreement.
High quality. An isochronous service agreement may specify a high bandwidth capability allowing for full fidelity and/or uncompressed audio.
Page 60
True streaming capability. The clock distribution required by synchronous-toisochronous-to-synchronous conversions provides for true digital audio streaming. Isochronous streaming emulates a synchronous connection.
DISADVANTAGES One disadvantage is that isochronous service is not widely available on data networks. It is closely tied to a quality of service (QoS) agreement. Quality of service is just starting to come into play on data networks. In addition to a service agreement from the underlying network, a framework for data transfer is required. CobraNet provides this framework for Ethernet. For other network technologies such as ATM and IEEE1394 isochronous transport is an integral service.
Another disadvantage is that you are often limited to point-to-point connections. For instance, isochronous service is commonly provided by using a synchronous transport. An example of this is teleconferencing over ISDN connections. ISDN guarantees 64Kbits/second throughput in each direction. This is a simple QoS agreement suitable as a basis for providing isochronous service.
CobraNet
CobraNet is a combination of hardware (the CobraNet interface), network protocol and firmware. CobraNet operates on a switched Ethernet network or on a dedicated Ethernet repeater network. CobraNet provides the following additional communications services for an Ethernet network.

Audio Ethernet
Control and Monitoring
Computer Data
Clock
Isochronous data transport Sample clock distribution Control and monitoring data transport
Fig. 5.3: CobraNet Data Services
Routing Audio
Page 61
The CobraNet interface performs synchronous to isochronous and isochronous to synchronous conversions as well as the data formatting required to transport real time digital audio over the network. The CobraNet interface has provision for carrying and utilizing control and monitoring data such as Simple Network Management Protocol (SNMP) through the same network connection as the audio. Since CobraNet is Ethernet based, data transport capabilities of Ethernet are shown here as computer data. In most cases, data communications and CobraNet applications can coexist on the same physical network.
CobraNet Terminology
CobraNet is a technology that spans the audio and data communications industries. While each industry has its own set of impenetrable terminology, the following terms are associated with CobraNet itself.
CobraNet Interface - The CobraNet interface is the hardware (or hardware design) and software supplied by Peak Audio to CobraNet licensees and affiliates. CobraNet Device - A CobraNet device is any piece of gear containing one or more CobraNet interfaces. Conductor - The conductor is the CobraNet interface elected to provide master clock and transmission arbitration for the network. The role of the conductor and the means for selecting a conductor will be described in more detail shortly. All other CobraNet devices on the network can be said to be operating in a performer role. Audio Channel In professional audio, the meaning of an Audio Channel should be completely unambiguous J. What were referring to in the case of CobraNet is digital audio with a 48KHz sample rate, and user selectable 16, 20 or 24 bit resolution.
Page 62
Bundle - A bundle is the smallest network audio routing envelope. A bundle represents transmission of an Ethernet packet once per isochronous cycle. A bundle may carry 0 to 8 audio channels. Bundles are numbered 1 through 65279. Bundles have a single transmitter. There are two types of bundles: Multicast Bundle - Bundles 1 through 255 are multicast bundles. A transmitter configured to source a multicast will always transmit the audio regardless of whether anyone is receiving. Point-to-multipoint connections are possible. Multicast addressing is used for the data packets associated with a multicast bundle. Unicast Bundle - Bundles 256 through 65279 are unicast bundles. Only point-to-point connections are allowed. A transmitter and receiver must both be dialed-in to a unicast bundle before any data flows.
CobraNet Interface
The CobraNet interface can be thought of as an integrated circuit. The pinout of this integrated circuit defines its interface to the other portions of an audio product in which it resides. The "pinout" for the CobraNet interface contains the following functional groups.

Flash
Synchronous audio I/O connection to DAC, ADC and DSP.

Clock
RAM Sample Clock DSP Audio Serial Host Ethernet MAC OP Timer Ethernet PHY Ethernet
Clock I/O for synchronization of host digital audio circuits. Host Interface for connection to the product host processor for the manipulation of management variables (i.e. bundle assignments) and control data I/O. Asynchronous serial port for the serial bridging feature. Buddylink: an interconnect scheme to allow CobraNet devices to work as redundant pairs. Buddylink inhibits transmission and reception on the secondary unit as long as the primary unit is working properly.
Fig. 5.4: CobraNet Interface
Routing Audio
Page 63
The Isochronous Cycle

CobraNets isochronous service on Ethernet is made possible by imposing an isochronous cycle structure on network audio traffic. A variation of this scheme is used by IEEE1394 and USB in their isochronous service provisions. An isochronous cycle is initiated by the transmission of a beat packet by the conductor. A CobraNet device desiring to transmit audio onto the network will wait to receive the beat packet before transmitting audio data. Once all audio data has been transmitted, other types of data packets such as control and monitoring data and reservation requests can be transmitted.
Beat Packet Audio Data
Beat Packet
Asynchronous Data (Reservation Requests)
Isochronous Cycle Period (1-1/3ms)
Fig. 5.5: Isochronous Cycle
Conductor
The conductor provides master clock for the network. Clock distribution is via beat packet transmission timestamp. Performers adjust their local clock such that it is in sync with received beat packets. The local clock is generated by stable voltage controlled crystal oscillator (VCXO). Adjustments to the local clock are done in a smooth manner as to avoid any jitter artifacts that could affect audio performance. The conductor arbitrates bandwidth and network routing resources. If a CobraNet device wants to transmit onto the network, it puts out a reservation request to the conductor. The conductor approves the request by adding a permission entry to the beat packet. The request may be denied by the conductor if there is insufficient network bandwidth or if the requested bundle is already in use by another transmitter.
Conductor Arbitration
The conductor is elected through a conductor arbitration procedure in which the device with the highest declared priority takes precedence. An established conductor at same priority as a CobraNet device just joining the network takes precedence to prevent unnecessary switches of conductorship. The Ethernet MAC address is
Page 64
used as a tie-breaker when multiple devices of similar conductor priority vie for conductorship simultaneously, as may occur when the acting conductor is abruptly taken off-line. All CobraNet devices contain the resources required to perform the conductor role for a network. A switch in conductors does result in a brief (10ms) audio outage on the network. Conductor priority is a user configurable parameter on many CobraNet products. The system designer may wish designate a higher conductor priority for a centrally controlled UPS powered device in a large CobraNet installation. Some digital synchronization issues can also be addressed through the use of conductor priority.
CobraNet Protocol
The CobraNet protocol operates at the Data Link Layer (OSI Level 2). CobraNet uses three basic packet types. All packets are identified with a unique protocol identifier (0x8819) assigned to Peak Audio.
BEAT PACKET The beat packet is multicast addressed and is therefore received by all devices on the network. By its arrival time and the timestamp, the beat packet distributes the sample clock to all CobraNet devices on the network.
The permission lists in the beat packet indicates which devices are being allowed to transmit on the network. And, in the case of a repeater network, the beat packet also determines the order of transmission such that collisions can be avoided on these networks.
Ethernet Header
(Destination 01:60:2B:FF:FF:00 Protocol 8819)
CobraNet header (Beat)
Network operating parameters and current time
List of units with transmission permission
List of transmission permissions
Ethernet trailer (CRC)
Fig. 5.6: Beat Packet
Routing Audio
Page 65
RESERVATION PACKET Reservation packets are sent out by all CobraNet devices at approximately 1 second interval.
Network performance reporting is used by the conductor in determining whether resources are available when deciding to grant new audio transmission requests. The IP address is published in the reservation request so that an inventory of CobraNet devices and their IP addresses is available for utilities such as Disco and CobraCAD. The forward reservations are requests to the conductor for transmission permission. The reverse reservation requests enumerate the bundles the device wishes to receive. Reverse reservation requests are directed at potential transmitters of the requested bundles. The reservation packet is multicast addressed. There are two reasons for this. All potential conductors, not just the current active conductor, would like to know the current reservation requests. This facilitates a seamless switchover if the current conductor should fail. Also, all transmitters must examine all reverse reservation requests and determine if they should initiate a transmission in response.
Ethernet Header
(Destination 01:60:2B:FF:FF:01, Protocol 8819)
CobraNet header (Reservation)
Network performance reporting
IP List of address forward reservations (request to transmit)
List of Ethernet reverse trailer reservations (CRC) (request to receive)
Fig. 5.7: Reservation Packet
Routing Audio
Page 66
AUDIO DATA PACKET The audio data packet contains digital audio data. The data may be 16, 20 or 24 bit format. All data is uncompressed PCM data.
An audio packet may be multicast or unicast depending on whether it is carrying audio
Ethernet Header
(Protocol 8819)
CobraNet Bundle header number and (Audio Data) current time
Audio channel 1 data format and PCM data
Audio channel 2 Ethernet data format and trailer PCM data (CRC)
Fig. 5.8: Audio Data Packet
for a multicast or a unicast bundle.
Routing Audio
Audio Networks
Page 67
NETWORK MANAGEMENT AND CONTROL Introduction

A key benefit of using an audio network is the ability to remotely control and monitor all of the devices connected to the network from a central location. This section will focus on the basics of network management, as well as control and monitoring using SNMP.
What is Network Management?

Network management is a collection of processes used to maintain the health and efficiency of a network. This includes monitoring for: security to protect the network from being accessed by unauthorized users performance to eliminate bottlenecks reliability - to ensure that the network is available Network management is independent of the network size and use. The same techniques used to manage a computer network comprised of hundreds or thousands of computers and peripherals can also be used to manage an audio network comprised of only a handful of CobraNet devices. Devices on a managed network can be separated into three categories: 1. management nodes sometimes referred to as clients, masters or managers 2. managed nodes sometimes referred to as servers or slaves 3. unmanaged nodes A management node is most commonly a workstation or computer running a network management application. Its sole purpose is to gather information from the managed nodes. A managed node runs a process that services requests for information from a manager. If the manager has permission to access the managed
Network Management and Control
Page 68
node, the managed node will collect the requested information and return it to the manager. Examples of managed nodes are PCs, Ethernet hardware and CobraNet devices. If a node is unmanaged, it may support an incompatible management protocol or it may not support management at all. In some cases, a proxy may exist for this device that can speak on behalf of the unmanaged node. There are many different management and control protocols available. Some are proprietary and work only with specified devices. Most amplifier control schemes operate in this way. Other protocols are computer-specific and provide information about your manufacturerspecific PCs operation. Sometimes these computer-specific protocols can be interfaced to network-wide management protocols. This section will focus on the networkwide management protocol called SNMP, as well as some other basic tools used to control devices over a network.
Telnet, HTTP and FTP

As we learned in Section 4, a managed Ethernet switch can be configured via a console session using HyperTerminal. This type of session is most commonly used to assign an IP address to the switch. Because a console session requires a direct serial cable connection between a computer and the managed device, it is not a very convenient communication method for the long-term management of networks consisting of many devices spread out over a large area. However, once an IP address has been assigned to the device, there are many options for remotely accessing not only Ethernet switches, but also other managed devices over the network.
Page 69
TELNET Telnet is a terminal communications protocol used over TCP/IP networks. The Telnet client runs on a local computer and enables a connection to a remote server located on the network. In this case, the server could be another PC, an Ethernet switch, or any other device running a Telnet server. Any commands entered through the Telnet session on the local PC will be executed as if they were entered directly on the remote device.
Both console sessions and Telnet offer the most comprehensive access to managed switch configuration settings. Unlike a console session, a Telnet session requires a valid username and password before access to the remote device is granted. However, this access control is provided at the application level, and both console and Telnet are insecure.
HTTP Some managed devices also allow access to their configuration settings via a web browser or HTTP interface. Similar to using telnet, HTTP requires that the device is accessible over a TCP/IP network it must have an IP address.
By simply opening a web browser and typing the IP address of a managed switch, one is able to browse the configuration settings as if looking at a web page on the Internet. In fact, the information is provided by a web server residing on the managed device itself. On some managed switches, HTTP access is limited to only certain configuration settings. These web-accessible parameters vary from manufacturer to manufacturer, as do the general layout and navigation structure of the web pages themselves. Most switch manufacturers provide HTTP interfaces containing colorful graphs and charts that are periodically updated to reflect
Page 70
changes in network status. These web browser interfaces are great tools for managing network performance, as constant information about switch status, errors, as well as traffic flow to and from the switch is provided in a simple, easy-to-use interface. This method of direct HTTP management of the device typically falls short when trying to manage large networks.
TFTP Trivial File Transfer Protocol, or TFTP, is a protocol that allows files to be transferred to and from remote computers or devices. In an audio network, and more-specifically in a CobraNet network, TFTP is used to perform firmware upgrades for both audio devices and switches over the network.
Management Using SNMP

Application, Presentation, Session Transport Network Data Link Physical UDP IP Ethernet SNMP TCP
Fig. 6.1: SNMP and the OSI Model
SNMP stands for Simple Network Management Protocol. It is a standard for managing TCP/IP networks and is the most common management protocol in use today. Among its many advantages, SNMP requires very little code to implement. This enables management to be added to a device without placing additional demand on system resources and disrupting the primary function of the device. SNMPs extendibility makes it easy to add management capabilities to existing products. Because SNMP separates the management architecture from the hardware architecture, this broadens the multivendor base. SNMP communication is connectionless and uses the User Datagram Protocol (UDP) transport, which is unreliable. This means that unlike TCP/IP, error-free delivery of each packet cannot be guaranteed. This places the reliability burden on the SNMP management application to keep track of all of the messages it sends, and to resubmit in the event a message is lost.
Ethernet
IP Packet
UDP
SNMP PDU
CRC
Fig. 6.2: Ethernet Packet Carrying an SNMP Message
Page 71
SNMP is implemented by software residing in the application layer, and thus it operates on Layer 7 of the OSI reference model. It requires the Presentation Layer to put data into a form that can be transported across the network. In order to carry SNMP data over Ethernet, PDUs (Protocol Data Units) are used. These are blocks that pass data from one OSI layer to another. Each layer has its own PDU, and for information to pass from one layer to another, the data must get packed into the PDU of each layer it will traverse. Sending SNMP messages over Ethernet is similar to the mail analogy used in Section Four to describe the sending of TCP/IP data over Ethernet.
Agents Managers and MIBs

AGENT An SNMP agent is an interface to a managed node. It resides on the managed node itself and is responsible for storing and returning data.
It is also responsible for the unsolicited signaling of events called traps. Traps enable managed nodes to immediately notify a manager of a failure, or a predefined exceeded threshold (to name just a few examples) without having to be prompted for that information by the manager. An SNMP agent can also serve as a proxy for unmanaged nodes. As a proxy, the agent acts on behalf of the unmanaged node to perform the management operations required by the manager. A proxy can be viewed as a type of translator. It receives an SNMP command, translates it to a language that the unmanaged node can understand, and then issues the command. The converse of this example is also true, such that the proxy can return SNMP data to the manager. Variables are defined in each managed devices MIB and are specific to the device. These are sometimes referred to as managed
2001 Peak Audio, Inc. Network Management and Control
Page 72
Objects, and are the leaves of the MIB tree (see MIB section below). These MIB variables include information about the devices hardware, its configuration parameters, and even performance statistics. SNMP agents can only retrieve information as it is defined in the managed devices MIB. MIB MIB stands for Management Information Base. It is a set of definitions that managers and agents use to exchange information about the managed devices hardware and software. The MIB is a standardized format that allows any SNMP tool to monitor any device defined in a MIB.
The objects in a MIB specify a structure and format for the properties of the managed object. Its representation is a hierarchical tree structure with each branch having a unique name and object identifier or OID. The intermediate branches group related MIB objects, while the leaves are the actual uniquely identified MIB variables. The full OID for a MIB object is the complete path from the top of the tree to the leaf containing the desired MIB object. Periods separate each interim branch number. For example, using the example MIB tree to the right, the OID for the mib-2 branch would be: 1.3.6.1.2.1 OIDs are assigned by the Internet Assigned Numbers Authority, and are unique. Once an OID is received, a manufacturer may extend the MIB as desired with no limit to the number of branches off of a single node.
MANAGER The manager is the console, or computer, performing the management functions. It runs an SNMP management application that can query SNMP agents, receive responses from the agents and also acknowledge events from agents. Managers can also perform
2001 Peak Audio, Inc. Network Management and Control
Page 73
some amount of control by performing SNMP sets. Most of the processing power and data storage resides on the manager, which lessens the burden on the managed node.
SNMP MESSAGE There are four SNMP operations used to manipulate the SNMP agents managed objects. These include:
1. 2. 3. 4.
GetRequest SetRequest GetNextRequest Trap
Each of these can be considered as a template that an SNMP management system or agent fills in with specific information and then sends on its way. The GetRequest message is used to retrieve values from the managed objects maintained by an agent. For instance, a get may be issued to a CobraNet device to get information about the bundle number its transmitting. The SetRequest is used to set values for managed objects. An SNMP set could be used to change the Bundle number on a CobraNet device. The SetRequest is the only mechanism available to a manager that enables management data to be modified. A GetNextRequest is the same as a GetRequest, but is used to walk through tables. A GetResponse is the response message returned by an agent as a reply to a request. The GetResponse would be used by the CobraNet device to return the requested information about the bundle it is transmitting. Responses are not issued unless a request has first been received.
Page 74
Version
Community
PDU Type
Request ID
Error Status
Error Index
The last operation, called a trap, is an unsolicited response sent by an agent without having fist received a request. A trap is simply an event report, indicating the occurrence of something unexpected. The usefulness of traps can be illustrated in an example. When an audio component fails, the problem is typically not noticed until its too late the system is turned on for use during an event, and the gear doesnt work! Using a managed audio network, the ability of a device to automatically signal an equipment fault prior to system use, can enable a fix long before the event even begins. Although the CobraNet agent does not support traps, some management applications will enable the monitoring of specific variables within the CobraNet MIB, and issue a trap in the event a pre-defined threshold is exceeded. This ability of a management application to monitor MIB variables is applicable to any managed device. Each SNMP message consists of 3 major components: 1. version number 2. community name 3. SNMP PDU The version number is simply the version of the SNMP protocol in use. The community name is the management community name of the host that is to receive the information. A community name is typically indicative of a group of nodes that perform a common function, are in a common location or serve a common user. The PDU contains the MIB object identifier of a managed object and the actual operation to be performed on that object. Multiple MIB OIDs and SNMP operations can be contained within a single PDU.
OID
Value
OID
Value
...
OID
Value
Fig. 6.3: SNMP Message
Page 75
Management Software
There are many different management software applications available today, and most Ethernet equipment vendors make available their own applications. Some software is shipped free with the purchase of managed switches, while other enterprise management applications must be purchased separately, and can be costly. The enterprise versions are much more feature-rich and powerful, allowing not only monitoring of network devices, but also providing an interface for custom scripting to extend the control aspects of the software.
POLLING AND NETWORK TOPOLOGY At its most basic, the status of network devices is established by issuing simple ping commands. If a device responds to the ping, it is considered to be up and communicating over the network. Additional information about device status can be gained by querying specific MIB variables.
Most enterprise management applications provide a very detailed picture of how the network segments interconnect. Device status is usually depicted using color schemes, with red icons indicating potential problems and green indicating normal operation. This allows quick visual detection of network faults.
DRILL DOWN CAPABILITY As networks get larger and larger, it may not be possible to show the entire network on a single screen. When this occurs, the network is typically represented as segments, each containing a number of individual devices or additional segments. A graphical hierarchy is maintained, such that a red problem segment from a low level is represented all the way through to the top level.
Page 76
EVENT LOGS AND PERFORMANCE Most management programs, including the bundled applications, provide the ability to log events. These events include things like traps, downed network segments, nodes down or unavailable, and exceeded thresholds. Events can sometimes be categorized to provide a visual identification of severity.
Custom scripts can enhance the benefits of SNMP traps. For instance, a script may be written that would automatically send an e-mail or a pager message upon receipt of a device down trap. This greatly enhances troubleshooting, by not only automatically notifying maintenance personnel about a potential fault, but also naming the specific failed device. By using a managed audio network, maintenance personnel already know where the problem is, before even visiting the site! The more powerful enterprise applications also include charting capabilities that allow the user to graphically represent network trends such as bandwidth utilization. They also allow customized charting based on user-specified criteria such as particular MIB variables.
Page 77
Recommended Reading
James D. Murray, Windows NT SNMP, OReilly, 1998 Yoram Cohen, SNMP Simple Network Management Protocol, http://www.rad.com/networks/1995/snmp/snmp.htm
SNMP Management Application Reference

Bundled with Managed Switches:
HP Top Tools 3COM Transcend Network Supervisor

Larger More Powerful Management Suites:
HP OpenView 3COM Transcend Tivoli TME10 NetView Cabletron Spectrum Computer Associates Unicenter TNG
Audio Networks
Page 78
DESIGNING A NETWORK TO ACCOMMODATE SPECIAL NEEDS High Availability

A system can deliver high availability (HA) in a number of dimensions. Means of measuring system availability include the following: Percent uptime Mean time between failures (MTBF) Mean time to repair (MTTR)
HIGH AVAILABILITY DESIGN Availability can be improved for any system design by specifying high quality components, by using hot swappable components and/or by adding redundant components to a design.
High availability systems can be classified into systems which should never be allowed to fail and systems which must recover quickly from failure. Different approaches may be required for the two types of requirements. The following strategies may be used to increase availability of systems. It is necessary to mix and match these techniques to achieve the appropriate balance for each installation.
QUALITY COMPONENTS It may be an exaggeration to say a system is only as good as its weakest link. It is true that poor reliability of any component in a system generally affects reliability of the system as a whole. The problem in selecting quality components is burrowing through the marketing fluff and identifying the quality stuff. There is no simple advice to give here.
Accommodating Special Needs
Page 79
Selection of quality components is made more feasible by the use of off-the-shelf technologies such as Ethernet and CobraNet where the choices are many.
COLD SPARES Availability can be improved by simply having spare components close at hand. Spare components are only useful if you also have a trained troubleshooting staff.
Remember to include a spares allocation in specifications and to include provisions for staff training.
HOT SWAP The ability to change out failed portions of a system while the rest of the system operates normally (or at reduced capacity) creates a more available system.
All Ethernet connections are hot-pluggable. Some network gear (especially larger core switches) feature hot-pluggable elements such as power supplies and interface cards.
FAIL-OVER Redundant components can be waiting in reserve as a hot spare. The hot spare is brought into play in the event of a failure either automatically or manually. This is a simple and effective approach to reducing MTTR. Contingent on ones definition of failure, a fail-over availability solution may or may not affect a systems MTBF. FAULT TOLERANCE Redundant components can be connected and configured to be in continuous operation sharing system load. In an audio system this could be accomplished by using more and louder speakers than technically demanded by the application such that the system can still operate acceptably with one or two failed channels.
A major advantage of this approach is that all portions of the system are being continuously exercised. You wont find
2001 Peak Audio, Inc. Accommodating Special Needs
Page 80
yourself in a situation where a backup unit had silently failed several weeks ago and you don't know you're operating without a safety net.
Environmental Factors
Some availability problems are caused by external factors. Special attention should be paid to surge protection and grounding and isolation in lightning prone installations, for instance. Note that fiber optics provide excellent EMI and lightening isolation and that all CAT-5 100BASE-TX Ethernet interfaces include 1,000V transformer isolation. The reliability of AC power should be assessed and specification of uninterruptible power supplies (UPS) specified if necessary. UPSs can greatly improve MTTR on system components such as PCs that take a long time to boot.
Monitoring
Automated system monitoring and availability are closely tied. The MTTR for a system is improved if you're alerted to a fault immediately when it occurs through a red light and/or automated call or page. It should be obvious that you can fix a problem more quickly if you are alerted to it as soon as possible. Even better if you can be alerted before the failure occurs (i.e. when a fan fails but before the smoke gets out or when a hard drive's error recovery circuits detect marginal data). These may seem farfetched concepts to the audio world but they are in everyday practical use in the data communications industry. Many failures in conventional analog audio system are not easily detected until the system is active use. This is due to the fact that distribution of audio in these systems is unidirectional. Audio goes out to the amplifiers and speakers on analog wiring and often there is no return path acknowledging that the audio reached its destination.
Page 81
A network provides a bi-directional (even multidirectional) communications medium. Continuous acknowledgment even in the absence of program material can immediately notify you of any communications or audio link failures. The detailed nature of acknowledgment in a bidirectional distribution system can help you identify and correct problems quickly. Contrast "Bad impedance reading in HF driver, speaker enclosure W23" to "sounds a bit muddy over near the end-zone."
Life Safety
Life safety systems require a different definition of availability. These systems are expected to continue operating under unusual circumstances (i.e. while the building is burning). Life safety systems often require a monitoring component. In many cases these systems are used infrequently and there must be an integral means of knowing that the system will operate when called upon. The bi-directional communications of a networked audio distribution system can provide that assurance.
Network Availability
Ethernet is used in mission critical applications the world over. There are many off the shelf and standardized techniques for achieving high availability in an Ethernet based network. This is a major advantage of using such a widely deployed technology for audio distribution. A survey of Ethernet availability technologies follows.
LINK AGGREGATION (TRUNKING) In any non-trivial network you will want to interconnect two or more Ethernet switches. A single link between switches may carry many audio channels. That link may be strained from a bandwidth perspective. It may also be considered strained from an availability perspective in that loss of that link may take down an unacceptable portion
Page 82
of system functionality. Link aggregation addressed both issues. Link aggregation is a means of allowing two or more parallel links between switches to behave as a single link. Normally if multiple links are connected in parallel between switches you will create a loop condition. Link aggregation prevents the loop condition from occurring. All connections remain active and traffic is balanced across both links. This increases the bandwidth between switches. Also should any link fail, the remaining links immediately pick up the slack. Since all links are always active there is essentially no fail-over delay on an aggregated link. An IEEE standard for link aggregation (802.3ad) has recently been ratified. Prior to standardization a number of manufacturer proprietary schemes proliferated and these non-interoperable schemes are still entrenched in the industry. 3Com's implementation is called Port Trunking. Cisco's proprietary implementation is called Fast EtherChannel. In order to use link aggregation the switches involved need to be properly configured. Ports need to be manually designated as members of an aggregation group.
SPANNING TREE PROTOCOL Spanning tree protocol (STP) is designed to prevent loops from occurring in an Ethernet network. Loops can be an availability problem on their own you cannot discount human error or sabotage/vandalism as a legitimate source of potential failure. In solving the loop problem, STP allows you to intentionally include loops in your network design. Loops amount to redundant paths and so a looped topology can be made to be fault tolerant.
w/o Aggregation - loops
loop
loop
w/ Aggregation - no loops
Fig 7.1 Link aggregation allows multiple links (two in this example) to behave as a single higher capacity link. The loop condition that usually results from parallel connections between switches is avoided.
Page 83
node switch
node
switch
switch
node
The spanning tree protocol works by sending out feeler packets to determine the way the network is wired. The switches themselves generate these packets. Under STP a port begins in a disabled state, once the algorithm determines that the connection does not form a loop in the network, the connection is enabled. Any connection made or broken causes the spanning tree algorithm to re-evaluate the network. As such, a failure of a link or even failure of a switch can be routed around by the enabling previously disabled links. STP is a long-standing IEEE standard (802.1D). With 15 years of service, it is an interoperable, general solution. Spanning tree configuration is often as simple as globally enabling the feature on each switch. And most Cisco switches are shipped with STP enabled as factory default. STP is a safe but slow algorithm. At no time is any link enabled which would, even momentarily, cause a loop condition. It does typically take STP 30 seconds to enable a new connection or enable a disabled connection in the event of a failure elsewhere on the network.
MESHING Meshing is a feature currently only available in the HP ProCurve family of switches. This feature is a fusion of spanning tree and link aggregation. With meshing enabled, switches can be interconnected in any way you desire - a ring, parallel connections, or a scrambled mesh. All links remain active and load is shared amongst links as in aggregation. The proprietary meshing algorithm magically avoids the dreaded loop condition, not by wholesale disabling of links but by refusing to forward packets blindly and endlessly in circles.
node
switch
switch
node
switch
disables a link to prevent the ring condition
node
Fig. 7.2 Spanning Tree Protocol: Disabled Link
node switch
node
switch
switch
node
X
node switch
Broken link
switch
node
switch
re-enables the link to route around the failure
node
Fig. 7.3 Spanning Tree Protocol: Re-enables Link
Meshing gives the quick fail-over and increased bandwidth of link aggregation combined with the wiring flexibility of STP.
Page 84
And like link aggregation, meshing require that ports participating in the mesh be manually identified when the switch is configured.
Audio Interface Availability

CobraNet provides several means of achieving high availability at the audio interface to the network.
PAIRING INTERFACE DEVICES Multiple CobraNet devices can be assigned to transmit on the same bundle. The conductor will only grant transmission permission to one device. If that device fails, permission will transfer to another device. If all devices are fed the same audio sources, youve achieved a redundancy with this simple setup.
Multiple CobraNet devices can be set up to receive the same multicast bundle. They will all receive the same audio. Multiple receivers can be assigned to the same unicast bundle. Only one receiver will receive the audio. If that receiver fails, the receipt will transfer to another unit. In either scheme, audio outputs of the two devices can be combined in the analog domain to produce a robust signal.
BUDDYLINK Buddy link is a refinement for redundant paring setups as described above. Buddylink is a clock connection from primary to secondary device. A clock signal is present at the input to the secondary device whenever the primary device is working properly. Presence of this signal inhibits the secondary device from participating on the network. Should the primary device fail, the clock signal is lost and the secondary takes over immediately.
Page 85
DUALINK In many system designs, an Ethernet switch can constitute a single point of failure for an essential portion of the system. Without special consideration, loss of a switch means loss of connectivity for everything connected to it. This risk can be mitigated by making two separate network connections for critical pieces of the system. The approach is routinely employed in data communications networks. Servers are populated with multiple NICs that connect to different portions of the network for fault tolerance.
The newest version of CobraNet interface, the CM-1, sports two Ethernet connections in an arrangement known as DuaLink. If connectivity is lost on the primary port, communications are switched over to the secondary.
Mixed Use Networks

CobraNet and data communications traffic can coexist on the same switched Ethernet. Data communications traffic, however, can be highly variable, and difficult to quantify. Data communications bandwidth requirements can go from almost nothing for an idle computer to full 100Mbit during a large file transfer operation. CobraNet traffic, on the other hand, is constant at approximately 1Mbit per audio channel. You may do a simple design for a mixed use network and everything will work fine. But then there might come a day when a bootleg Britney Spears video gets copied to a server on the network and the data communications traffic goes through the roof. If this event causes any portion of your network to become overcommitted and packets will be lost indiscriminately. Some lost data will be CobraNet audio and some may be Britney's bellybutton. What do you do about this? You may decide that for a paging system, an infrequent audio dropout is not of great concern. You may decide
Page 86
to control the types of applications running on computers connected to your network. Internet browsing and most system control and monitoring functions are of limited bandwidth. You may decide to install a physically separate network infrastructure for CobraNet. Or you may consider the following network remedies.
CobraNet
Non-Blocking Switched Network Design

Networks can be designed with adequate bandwidth to handle any conceivable traffic pattern. Previously the additional expense of such a design was prohibitive. The advent of gigabit Ethernet makes this a more feasible undertaking especially for small to medium sized installations. Examples of non-blocking switched network designs are shown in Figures 7.4 and 7.5.
CobraNet
Fig. 7.4 The high internal bandwidth of modern Ethernet switches insures that a simple network such as this is inherently non-blocking.
Virtual Local Area Networks

One means of limiting interaction between CobraNet and data communications applications on the same networks is through the use of virtual local area network (VLAN) features. This allows a switch to be partitioned into noninteracting network domains. Note that although this achieves the goal of preventing data communications from affecting CobraNet audio, you have also lost the ability to communicate between the computers and CobraNet devices. You have lost the ability to monitor and manage the CobraNet devices from the data network. We will discuss solutions to this problem later on. Use of VLANs requires a switch supporting VLANs (most managed switches have VLAN support) and manual configuration of the switch to assign specific ports to specific virtual networks. Figure 7.6 demonstrates use of VLANs to isolate data communications from CobraNet traffic on the same physical network.
CobraNet
CobraNet
Fig. 7.5 The link between the two switches in this example network would normally be a potential bottleneck for this network. In this case, however, we know that most of the network traffic will be from computer to computer and from CobraNet device to CobraNet device. The link between switches is only being asked to carry low bandwidth control and monitoring data.
Page 87
CobraNet CobraNet CobraNet
Taking the virtual concept one step further, through the use of VLAN tagging, we can multiplex data for both VLANs onto a single uplink. By doing this we increase flexibility and reduce wiring complexity for the network but in the shared link we have created a potential interaction point between the data and audio networks.
Quality of Service
Fig 7.6 Example use of VLANs to isolate data communications from CobraNet traffic on the same physical network. Note that separate uplinks for each VLAN are used in this simple example.
VLAN 1
VLAN 2
Quality of service (QoS) and class of service are sophisticated features that are beginning to come into bloom in cost effective network equipment. QoS allows you to specify the relative priority of network traffic. Means of identifying important traffic include by network address or port number, by VLAN membership or by protocol type. QoS features require careful configuration of the switches that comprise the network. QoS comes into play when a port or link is overcommitted. Traffic identified as less important will be dropped in favor of the more important data. QoS is most commonly used to resolve contention on potentially over-committed uplink ports between switches. This allows use of shared uplinks for multiple VLANs without risking undesirable interactions between competing traffic streams.
FIFO
FIFO
Fig. 7.7 Switch supporting VLAN tagging and QoS. Packets sent and received via the shared middle port are tagged to indicate VLAN membership. Dual outbound queues and priority selection logic on this port implements the QoS feature.
FIFO
FIFO
DeMux
FIFO
FIFO
Priority
Large Installations
In large scale projects networking information technology (IT) as well as audio experts are typically called in to do design and system commissioning. In the new digital world order, audio and IT professionals must work closely together to build a system that meets current and future needs of the facility. As is true in the audio profession, there are competent and not so competent IT professionals. In this section we strive to give you an introduction to some of the networking issues and solutions for large scale installations.
Page 88
Core Switching
Ethernet networks prefer to be wired in a star configuration. Smaller networks can be done with a single star. Medium sized installations can be done with several interconnected stars. For the larger installations a star of stars is the most appropriate configuration. At the center of this configuration is a high capacity switch known as a core switch. Core switches, being in the center of the network, are in a vulnerable single point of failure position. Thus, most core switches are the size of a small refrigerator and contain numerous redundancy features such as hot-swappable power supplies and interface cards. Many of these critical large networks are wired with a redundant pair of core switches as shown in Figure 7.8.
Attraction Equipment Room Central Equipment Rooms Core Switch Core Switch Edge Switch Attraction Equipment Room Edge Switch
Rack Switch
Rack Switch
Attraction Equipment Room Audio I/O Audio I/O Audio I/O Audio I/O Edge Switch
Attraction Signal Processor
Fig 7.8 Example theme park network with redundant core switching
Audio VLANs
CobraNet itself has some scalability limits. The first (and only) limit one is likely to encounter is a limit on multicast bundle traffic. Multicast traffic is forwarded to all stations on a network and as such only a limited amount of this traffic is acceptable network wide. Detailed guidelines for managing multicast traffic on a network are given in our Bundle Assignments in CobraNet Systems included as an addendum to these materials. The most recent version is available on the Peak Audio web site at http://www.peakaudio.com/cobranet/papers/Bun dle_Assignments.html. Upon reaching the multicast limit one has two options, convert some multicast traffic to unicast or divide the network into separate VLANs. The use of audio VLANs in a large installation may make sense from an administrative point of view. These larger systems are often viewed as an interconnected collection of smaller subsystems. VLANs can be used to define the boundaries of the sub-systems and assure there are no undesired interactions between the subsystems.
Page 89
Central Control and Monitoring Control and Monitoring VLAN IP Router Attraction A Signal Processing Parkwide Signal Processing
Layer 3 Routing
A large network is typically subdivided into multiple domains through the use of VLANs. When this is done, it becomes impossible for the different domains to communicate with one another. While it is desirable to maintain the isolation between domains with respect to audio traffic, it is probably desirable to allow control and monitoring traffic to flow freely between domains. Layer 3 routing can accomplish this. Layer 3 refers to the OSI networking model. Routing is used on the public Internet. As a network grows larger and larger, use of such constructs becomes more valuable. Figure 7.9 demonstrates the utility of routing in a large theme park application.
Parkwide Audio VLAN
Audio I/O
Audio I/O
Attraction A Audio VLAN Spare Signal Processing
Audio I/O Audio I/O
Attraction B Audio VLAN
Fig 7.9 Example theme park network with layer 3 routing. The diamond represents the layer 3 routing element of the network. This routing element is, in some cases, a separate piece of hardware but more commonly it is a software feature of the core switch.
Distance and Switch Hops

Larger systems can involve long cable or fiber runs and numerous pieces of network gear through which data must pass before reaching its destination. CobraNet, being a real-time system, imposes performance requirements on the network for audio delivery to work reliably. The performance requirements are detailed in Protocol Description and Timing Practices included as an addendum to these materials. The most recent version is always available on the Peak Audio web site at: http://www.peakaudio.com/cobranet/developer/ protocol.html. The core of these requirements is a soft requirement on forwarding delay and a hard requirement on delay variation.
FORWARDING DELAY CobraNet guarantees that all audio is delivered in exactly 256 sample periods (51/3ms). This, however, is premised on the network meeting the forwarding delay specification. On a network that exceeds the forwarding delay specification, receivers may automatically add additional latency in 64 sample period (1-1/3ms) on a per bundle basis. Violation of the forwarding delay specification does not guarantee additional latency; this is dependent on variables such
Page 90
as the path taken by data through the network and the location of the conductor on the network. Receivers can be queried via SNMP to see how much, if any, additional delay has been added to the bundles they are receiving. Limiting the number of switches data must pass through before reaching its destination is the primary means of controlling forwarding delay. Each 100Mbit switch contributes approximately 150us of forwarding delay. Increasing link speed also cuts forwarding delay. Gigabit Ethernet operates ten times faster than 100Mbit fast Ethernet and forwarding delays are one tenth those found in 100Mbit equipment - about 15us. Overhauling your network topology is the most effective means of reducing switch hops in a design. Star topologies offer the fewest hops. Due to the universal speed limit of light, cabling introduces some forwarding delay. Normally these delays are negligible but for truly large systems they can be significant. A 2km fiber run adds approximately 20us of forwarding delay. An 80km run nets 800us. Bouncing your data off a satellite in geosynchronous orbit will cost you about 750ms!
Page 91
DELAY VARIATION The delay variation specification is a hard specification. CobraNet cannot operate reliably on a network in violation of the delay variation specification. Sample clock distribution is accomplished by observing beat packet arrival times and adjusting a local clock to match those arrival times. If beat packet arrival times become too unpredictable due to delay variation introduced by the network, a device will not be able to reliably synchronize its sample clock to the network. Without a valid sample clock, audio transmission and reception are suspended.
Delay variation is generated as data is forwarded through switches. Forwarding time through a switch is not necessarily constant especially when the switch is more heavily loaded and a packet being forwarded may need to wait its turn before being allowed to pass through. In systems where forwarding delay is dominated by delays through switches, delay variation tends to be proportional to forwarding delay. We suggest as a general rule no more than 7 switches in series in a network design. This guideline endeavors to keep delay variation (but not necessarily forwarding delay) within the specification. Spanning tree protocol and meshing also require that a 7 switch diameter limit be observed.
Audio Networks
Page 92
TROUBLESHOOTING AUDIO NETWORKS Introduction

Understanding how to troubleshoot an audio network is just as important as understanding how to troubleshoot an analog audio system. If things dont appear to work right away, the following troubleshooting tips can help you get to the root of the problem more quickly and efficiently.
Check the Obvious!

Sometimes the problem may be so obvious that you dont even consider checking for it! The most common problem under this category being, is there power to all devices? Always make sure all of the devices have powered supplied and are turned on.
Verify Link
Link status lights are typically found on the front or rear panels of most networking devices. Link is required for the network to function properly. If there is no link: Ensure that cables are connected and properly seated. Sometimes not pushing a cable all the way into an RJ-45 jack can prevent link from being established. If reseating the cable doesnt work, its possible that your may have a bad cable somewhere in the path. The easiest check for a bad cable is to replace the suspect cable(s) with a known good cable. If the link indicator isnt lit on a particular switch port, it is possible that the switch port faulty. Try moving the cable from the suspect port to another port on the switch.
Troubleshooting Audio Networks
Page 93
If none of the above tests seem to have solved the problem, the switch may be improperly configured. In networks where no advanced features (Spanning Tree Protocol, Meshing, Trunking, VLANs, etc.) are used, try resetting the switch to its factory defaults.
Managed Switch Configuration

If youre using unmanaged switches in your design, there are probably no configuration options to check. However, if youve verified that link is established and your network uses managed switches, double-check your switch configuration. CobraNet devices use auto-negotiation to determine whether they are attached to a switched or repeater network. This means that all TX ports should be configured for autonegotiation. When auto-negotiation is disabled (and the switch port is configured manually) a CobraNet device will assume it is connected to a repeater hub and establish a half-duplex connection. Even if the switch port is configured for half-duplex operation, the CobraNet device thinking it is connected to a repeater network, will have activated its collision avoidance mechanism. This mode of operation is inappropriate on a switched network and may result in intermittent audio transmission problems. Because no auto-negotiation takes place over fiber, all fiber ports must be manually configured for 100Mbps, full-duplex operation. Mismatched or inappropriate operating modes result in sub-optimal performance and/or packet loss. If you have rings in your network design, remember to make sure that Spanning Tree Protocol or Meshing is enabled on all switches in the ring(s).
Page 94
Are LEDs Indicating Errors?

If so, its possible that there is a bad cable somewhere in the system. Cable problems can be due to manufacturing flaws, incorrectly twisted pairs, or the wrong cable type. Regular phone cable or CAT3 cable will produce adverse results. It is also possible that someone has installed the incorrect RJ-45 connector type. Use of the wrong connector results in intermittent or high impedance connections.
Check the Audio

In general, never be afraid to simplify the system. If possible remove the intermediate components and try to send the audio directly from the source to a receiver.
SIGNAL PRESENCE Check any indicators available on the audio devices for signal presence. Is audio present at the inputs? If so, it may be a problem at the outputs. If not, check the obvious is the audio network device powered up, is your program source turned on and playing, etc.
Is audio signal present at the outputs? If so, maybe the problem resides downstream. Is the bundle number correct? Make sure that the CobraNet bundle assignments have been assigned properly. If a device is transmitting on channel 301 and no device has been configured to listen on that channel, no audio will be transmitted.
AUDIO QUALITY Sometimes your best troubleshooting tools are your ears! Listen for the audio signal. Does it sound distorted?
If you hear distortion, it is important to try to categorize the quality of the distortion. In other words, is the distortion equal at all sound levels, or only present at higher levels? Different distortion types can have different causes.
2001 Peak Audio, Inc. Troubleshooting Audio Networks
Page 95
If it is just at higher levels like analog overload distortion, then use your best analog audio system skills to troubleshoot the problem. Consider things like; your gain structure, are you clipping, is the device configured for mic level audio instead of line level, etc. If it sounds distorted at all levels, the network may be experiencing dropouts. Dropouts can be caused by over-committed links, a bad cable, or an improper cable. Digital audio tends to be an all or nothing type of operation. Usually if you have audio, all is well. However, just as with analog audio there can be intermittent or marginal conditions. The way these show up in the audio is different, due to the way audio networks packetize the audio. Lost packets cause small holes in the audio, but since the holes are so short (around 1 ms) they often sound to the ear like distortion rather than dropouts. If you hear what sounds like distortion that is not just at high signal levels, most likely it is actually packet loss. Finally, check to ensure that all of the channels are being received. Confirm that bundle numbers are set correctly, and once again, check the obvious!

CobraNet CNTraining2

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

CobraNet CNTraining2

Încărcat de

Drepturi de autor:

Formate disponibile

Audio Networks An Overview

Peak Audio, a division of Cirrus Logic, Inc.

2001 Cirrus Logic, Inc.

DETERMINING WHEN AN AUDIO NETWORK IS THE RIGHT SOLUTION Introduction

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

Distance and Channel Count

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

Other Network Requirements and Existing Infrastructure

Control and Monitoring

Redundancy and Reliability Requirements

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

Material and Labor Costs

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

2-32 PAIR CABLES

1-32 PAIR (SPARE)

1-32 PAIR (SPARE)

2-32 PAIR CABLES

1-32 PAIR (SPARE)

CENTRAL CONTROL LOCATION EQUIPMENT RM 2

Traditional Analog System

COBRANET INTERFACE (TYP) CN CN CN 100BASE -FX SWITCH

CN 100BASE -FX SWITCH

CN CN 100BASE -FX SWITCH

EQUIPMENT RM 1 CABLE RACEWAY

3/4 C. FIBER OPTIC CABLE (TYP)

CN 100BASE -FX SWITCH

CENTRAL CONTROL LOCATION EQUIPMENT RM 2

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

Convention Center Example:

CENTRAL CONTROL LOCATION

REMOTE EQUIP RM 2 TO RMS 3-5

1-1/2 C. 1-8 PR CABLE 1-4 PR CABLE MTG RM 3 MTG RM 4 REMOTE EQUIP RM 6

MTG 27 MTG 28 TO HALLS 4/3

2-1/2 C. 1-16 PR 2-4 PR

1-1/2 C. 1-12 PR 1-4 PR

Traditional Analog System

2001 Cirrus Logic, Inc.

When a Network is the Right Solution

100BASE -FX SWITCH

100BASE -FX SWITCH

100BASE -FX SWITCH

MTG 7 MTG 8 TO RMS 9/10

ANALOG AUDIO LINES (TYP)

100BASE -FX SWITCH

100BASE -FX SWITCH

100BASE -FX SWITCH

EQUIP RM 7 MTG 3 MTG 4

100BASE -FX SWITCH

MTG 7 MTG 8 TO HALLS 3/44

100BASE -FX SWITCH CN CN 100BASE -FX SWITCH

100BASE -FX SWITCH

EXHIBIT HALL EQUIP RM 1 TO BALLROOMS 1-4

EXHIBIT HALL EQUIP RM 2