Intro to livewire
Introduction to Livewire
IP-Audio System Design Reference & Primer
Version 2.0 — April, 2007
About this manual
This manual is your introduction to Livewire. We explain the ideas that motivated
it and how you can use and benefit from it, as well as nitty-gritty details about wiring,
connectors, switches, and the like. Since Livewire is built on standard IP networks, we
also help you to understand general network engineering so that you have the full
background for Livewire’s fundamentals. After reading, you will know the essentials of
Livewire along with many of the details of the gear that vendors and the network guys
that are often hanging around in radio stations will be using.
This document covers topics common to all Livewire equipment. It is only a part of
your full documentation package. You will also have manuals for each specific piece of
equipment that are used to build your system. From this document, for example, you will
not learn how to install or operate an Element, but you will understand the nature of the
network they plug into.
It was early in 2004 when Livewire was first introduced and came to market being
new and fresh. Today, Broadcasters and Engineers have grasped this standards based
technology to the point where you now find Livewire incorporated into many other
vendors’ equipment. Not to mention making Axia, the fastest growing console company
in the world!
There will continue to be updates to this document because new ideas for using standard Ethernet components will be explored and tested in our labs. This will most likely
result in new equipment being released. This document will include the general descriptions on how this equipment will work with a Livewire network. Also, as we assist with
your installations, we’ll find new and better ways to explain things. So check our web site
or contact our support department for the latest version.
As always, we welcome your suggestions for improvement. Please feel free to contact
Axia Audio with your comments:
Axia Audio, a Telos company
2101 Superior Avenue
Cleveland Ohio 44114 USA
Phone: +1.216.241.7225
Web: www.AxiaAudio.com
eMail: [email protected]
ii
Intro to Livewire
Contents
1: Livewire for beginners............................ 2
Why Ethernet?...................................................................................2
Compared to AES.............................................................................3
Audio Routing...................................................................................4
The Livewire Advertising System ..............................................4
Control..................................................................................................4
Livewire and PCs...............................................................................5
Support for Surround......................................................................6
Audio Quality.....................................................................................7
The Pac-Man of Protocols: Internet Standards................... 1 0
Converged Networks................................................................... 1 1
2: What can you do with it?..................... 13
Make A Snake.................................................................................. 1 3
A High-Performance Sound Card Replacement................ 1 3
5: Designing & building your Livewire
Ethernet system.......................................... 51
CABLING........................................................................................... 5 1
Architecture Options.................................................................... 5 7
Fiber................................................................................................... 6 1
Designing For Security................................................................ 6 2
6: The Ethernet switch.............................. 6 4
Livewire Ethernet Switch Requirements............................... 6 4
Some Switches We Like............................................................... 6 5
Switch Configuration................................................................... 6 6
7: Testing, 1-2-3….......................................... 67
General Ethernet Troubleshooting......................................... 6 7
Diagnosing Problems using Livewire Components......... 7 1
Build An Audio Router................................................................. 1 4
8: Network engineering for audio
engineers...................................................... 7 3
Build A State-Of-The-Art Broadcast Studio.......................... 1 5
Ethernet / IP Networks................................................................. 7 4
Make a Flexible Two-Way Multi-Channel STL...................... 1 6
IP and Ethernet Addresses......................................................... 7 7
Create A Facility-Wide Audio Network That Includes Integrated Studio Consoles............................................................... 1 7
Ethernet Switching....................................................................... 8 1
Create An Integrated National/Local Radio Network....... 1 7
3: The Axia Livewire components............ 18
Livewire Hardware Nodes.......................................................... 1 8
Axia IP-Audio Driver for Windows®......................................... 2 0
iPlay (PC Router Selector)........................................................... 2 1
Element Broadcast Console & StudioEngine....................... 2 2
PathfinderPC Routing Control Software: ............................. 2 4
iProbe Network Management Console................................. 2 8
iProFIler Automated Program Archiving............................... 2 8
4: Nuts & Bolts: Making Livewire play... 2 9
Livewire’s Channel and Name system.................................... 2 9
Livewire Networks......................................................................... 9 2
A Note about Protocol Design.................................................. 9 8
9: FREQUENTLY ASKED QUESTIONS......................................101
General Questions................................... 101
PCs and Livewire........................................ 105
Building Livewire Facilities.......................................................105
Ethernet Media.............................................................................106
The Internet and Livewire........................................................107
The StudioEngine and Consoles............................................108
Analog Audio & AES On RJs And Cat 5 ...............................109
Livewire, Standards and Other Vendors..............................111
Examples.......................................................................................... 3 2
Hardware Node & Configuration Access............................... 3 8
Plugs & Cables................................................................................ 4 1
Intro to Livewire
iii
were simple, used only by budget stations
to reduce operating expenses. But soon the
applications evolved and were embraced by
larger stations. Slowly, the PC was taking center stage in the radio studio. Like many, I was
captivated by the PC. Stations retired carts,
phonographs, open-reel decks, cassettes
— even more modern digital equipment such
as DAT and CD players, replacing all with PC
apps. Client/server systems emerged and entire facilities began using PCs to provide most
– or all – of their recorded audio. Yet consoles
continued to treat PCs as nothing more than
A note from the
president of axia
N
designers were going to have to rethink our
designs to deal with computer-centric stu-
early 20 years ago, I designed my first
dios. During this time, traditional broadcast
broadcast console for PR&E. I look
console companies began producing digital
back on that time with great fondness; we
“Once you’ve experienced
the pleasures of networked
audio, you’ll never go back.”
iv Intro to Livewire
audio peripherals. I knew that we console
versions. But early digital consoles were nearly
were building bullet-
identical in form and function to their analog
proof boards for the
predecessors. It took a fresh look from a Euro-
world’s most presti-
pean company outside broadcasting to merge
gious broadcasters,
two products – audio routing switchers and
making each new
broadcast consoles – into a central processing
console design bigger and fancier to accom-
engine and attached control surface. Eventu-
modate a wider variety of source equipment
ally nearly every console and routing switcher
and programming styles. The console was the
company followed suit, and a wide variety of
core of the studio; all other equipment was on
digital “engines” and control surfaces flooded
the periphery. Then things changed: the PC
the market.But, advanced as these integrated
found its way into broadcast audio delivery
systems were, they still handled computer-
and production. At first, PC audio applications
based audio sources like their analog ances-
tors. Sure, the router and console engine were
Telos. Axia is all about delivering innovative
now integrated, but the most important stu-
networked audio products to future-minded
dio element – the PC – was stuck in the past,
broadcasters. On behalf of our entire team, I
interfaced with 100-year-old analog technolo-
welcome you as a charter client. Axia is the
gy. The PC and console couldn’t communicate
culmination of nearly 40 man-years of some of
in a meaningful way – strange, considering
the most ambitious R&D ever applied to the
that PCs everywhere were being networked,
radio industry. And this is only the beginning.
fast becoming the world’s most popular and
We have more products, innovations, and
powerful communication tool. Then a group
partnerships in the pipeline. You already know
of Telos engineers developed a method of
your Axia system is unlike anything else. So it
using Ethernet to interconnect audio devices,
shouldn’t be surprising that your new system
allowing computers and consoles, control-
is loaded with new thinking, new approaches,
lers and peripherals to interact smoothly and
and new ideas in virtually every conceivable
intelligently. Powerful, flexible networks had
area. Some concepts will challenge your
finally come to our studios. As with the transi-
traditional ideas of studio audio systems, but
tion from carts to computers, the benefits
we’re certain that once you have experienced
are many and impressive. A few networked
the pleasures of the networked studio, you’ll
components can replace routing switch-
never want to go back. And now, for some-
ers, consoles, processing peripherals, sound
thing completely different...
cards, distribution amps, selector switches
and myriad related devices. This deceptively
simple networked system costs a fraction of
Michael “Catfish” Dosch
other approaches, yet has capabilities surpassing anything else. The system is modular
and can be used to perform discrete functions
in a traditional environment. Concurrently, it
easily scales to serve both the humblest and
the very largest of facilities. Console, router,
and computer work in harmony. So, equipped
with this new technology and countless
ideas, we launch Axia, the newest division of
Intro to Livewire
A Note from the
president of Telos
I
t’s been a tradition since Telos’ very first
product, the Telos 10 digital phone system,
that I share a few words with you at the beginning of each manual. So here goes. In radio
broadcast studios we’re still picking up the
pieces that have fallen out from the digital audio revolution. We’re not using cart machines
anymore because PCs are so clearly a better
way to store and play audio. We’re replacing
our analog mixing consoles with digital ones
and routing audio digitally. But we’re still
using decades-old analog or primitive digital
methods to connect our gear. Livewire has
been developed by Telos to provide a modern PC and computer network-oriented way
to connect and distribute professional audio
around a broadcast studio facility. Your question may be, “Why Telos? Don’t you guys make
phone stuff?” Yes, we certainly do. But we’ve
always been attracted to new and better ways
to make things happen in radio facilities. And
we’ve always looked for opportunities to
make networks of all kinds work for broadcasters. When DSP was first possible, we used
it to fix the ages-old phone hybrid problem. It
was the first use of DSP in radio broadcasting.
When ISDN and MP3 first happened, we saw
the possibility to make a truly useful codec.
vi Intro to Livewire
We were the first to license and use MP3 and
the first to incorporate ISDN into a codec.
We were active in the early days of internet
audio, and the first to use MP3 on the internet.
Inventing and adapting new technologies for
broadcast is what we’ve always been about.
And we’ve always been marrying audio with
networks. It’s been our passion right from the
start. In our genes, if you will. As a pioneer in
broadcast digital audio and DSP, we’ve grown
an R&D team with a lot of creative guys who
are open-eyed to new ideas. So it’s actually
quite natural that we would be playing marriage broker to computer networks and studio
audio. What you
get from this is
nearly as hot as
a couple on their
wedding night:
On one RJ-45,
two-way multiple
audio channels,
sophisticated
control and data
capability, and
built-in computer compatibility. You can
use Livewire as
a simple soundcard replacement
– an audio interface connecting to a PC with
there’s little question that you are going to be
an RJ-45 cable. But add an Ethernet switch
seeing a lot of it in the coming years. The 20th
and more interfaces to build a system with
century was remarkable for its tremendous
as many inputs and outputs as you want.
innovation in machines of all kinds: power
Audio may be routed directly from interface
generators, heating and air conditioning,
to interface or to other PCs, so you now have
cars, airplanes, factory automation, radio, TV,
an audio routing system that does everything
computers. At the dawn of the 21st, it’s clear
a traditional “mainframe” audio router does
that the ongoing digitization and network-
– but at a lot lower cost and with a lot more
ing of text, audio, and images will be a main
capability. Add real-time mixing/processing
technology story for decades to come, and an
engines and control surfaces and you have a
exciting ride for those of us fortunate to be in
modern studio facility with many advantages
the thick of it. Speaking of years, it has been
over the old ways of doing things. Ok, maybe
a lot of them since I wrote the Zephyr manual
this is not as thrilling as a wedding night
intro, and even more since the Telos 10 – al-
– perhaps kissing your first lover is a better
most 20 years now. Amazing thing is, with all
“The ongoing networking
of audio will be an exciting
ride for those in the thick.”
analogy. (By the way,
the change around us, I’m still here and Telos
and way off-topic,
is still growing in new ways. As, no doubt, are
did you know that
you and your stations.
the person you were
kissing was 72.8% water?) While were on the
subject of history… you’ve probably been
Steve Church
soldering XLRs for a long time, so you feel
a bit, shall we say, “attached” to them. We
understand. But no problem – you’ll be needing them for microphones for a long while, so
your withdrawal symptoms won’t be serious.
But your facility already has plenty of Ethernet
and plenty of computers, so you probably
already know your way around an RJ-45 as
well. It’s really not that strange to imagine live
audio flowing over computer networks, and
Intro to Livewire vii
1: Livewire for beginners
Livewire offers a revolutionary change in how studios can be built. But at the same time,
it’s a natural continuation of general trends and what you already know. This section
explains the basics and puts audio over Ethernet into context.
ithin the next few years, it is certain that the transition to digital now hap-
W
pening in our studios will be complete, with all audio storage, mixing, pro-
cessing and routing being digital. This transition is now in full swing. But the sheer
number of competing methods for integrating computers, legacy hardware and
state-of-the-art audio devices is (and long has been) bewildering. What we need
is a connection method that gets the interconnection job done easily, effectively,
flexibly, and cheaply. So why not look to the computer and telephone worlds to
find the technology? We can then take advantage of the huge manufacturing scale
in those industries and can piggyback on the billions of dollars (and Euros, Yen,
Yuan…) of R&D going on in those industries.
Why Ethernet?
Ethernet makes overwhelming sense. Today’s computers are near universally
linked via Ethernet – and telephony is gradually moving that way as well, with VoIP
rapidly gaining market share. Even remote controlled stage lighting is transitioning
from the XLR-based DMX protocol to Ethernet. Ethernet cables, plugs,
cards, and chips are produced in the hundreds of millions so we get tremendous economy of scale. We get patch bays and cords, testers, and
all kinds of “structured wiring” components ready-made. Plugs are easy
to install and jacks are efficiently small.
But much more important is that Ethernet allows us to combine
many channels of digital audio with whatever data transmission we
might need on a single cable. This data could be as simple as a start command for
an audio player or could be anything that computers and Ethernet do, such as file
transfer, e-mail, web communication, etc.
Further, we are in the line of future development. Since its invention over 30
years ago, Ethernet has been constantly evolving. It started as a 2Mbps shared bus
over coaxial cable and has grown to today’s modern (and very common) 1 Gigabit
Intro to Livewire
star and switched system. 10 Gigabit is already widely available on many Ethernet
devices and is following the usual curve to low cost as volumes increase. While copper is the most common Ethernet connection, fiber is popular as well and media
converters allow the two to be interconnected. Ethernet switches cost $6000 for
8 ports a half-decade ago; now high-end 24-port switches cost $500. And they
include many advanced features that were unheard of only a few years back.
There are radio links in many varieties, from WiFi for short-range to sophisticated long-range systems like the PTP Series from Motorola. There are satellite links.
And LASER links. Ethernet opens the door to a world of options.
Ethernet has proven to be the PC of networking: Initially released with only basic capability – low speed and bussed – it has been expanded to today’s fast, flexible, switched architectures.
The combination of huge R&D expenditures, open standards, massive economies of scale, technological evolution, and flexible multi-service packet design is
hard to beat. Not to mention the surprisingly appropriate name.
Compared to AES/EBU
For digital audio transport, AES3 is the main alternative to an Ethernet based
system. Invented in the days of 300-baud modems, it was the first practical answer
to connecting digital audio signals. But it’s now over 15 years old and is showing
its age. Compared to Livewire’s computer-friendly, two-way, multi-channel + highspeed data capability, AES3 looks pretty feeble with its 2-channel and one-way
only limitation. Not to mention 50-year old soldered XLR connectors and lack of
significant data capacity. AES3 is a low-volume backwater, with no computer or
telephone industry R&D driving costs down and technology forward. Your 300baud modem has been long retired; it’s well time to progress to the modern world
for studio audio connections as well.
That said, AES and Livewire may comfortably co-exist in your facility. You can
use Axia interface nodes to connect from one to the other. If you are using a house
sync system for AES, Livewire may be synced to that system also.
Ethernet was named
by its inventor, Robert
Metcalf. He had
been involved in a
radio data network in
Hawaii called ALOHA.
The first Ethernet was
a bussed coax that
carried data packets
similar to the way
ALOHA had sent them
over the “ether.”As to
the origin of ether…
for many years after
James Clerk Maxwell’s
discovery that a
wave equation could
describe electromagnetic radiation,
the aluminiferous
ether was thought to
be an omnipresent
substance capable of
carrying electromagnetic waves. In 1887
scientists Albert Michelson and Edward
Morley disproved its
existence. The ingenious experiment that
did so was performed
at Case Western Reserve University, just
down the street from
Telos/Axia main office
in Cleveland.
Intro to Livewire
Audio Routing
Low-cost mass-market Ethernet switches offer us something very interesting:
Since their function is to direct packets from port-to-port, we can use them to
move our audio signals from whatever source to whatever destinations we want.
This means we get a simple, flexible, facility-wide audio routing system, almost for
free. Say goodbye to racks of distribution amps or expensive proprietary mainframe audio routers.
An audio source entered into the system from any point becomes available for
any number of receiving destinations.
The Livewire Advertising System
Livewire has an audio advertising system. Every source has a text name and numeric ID. These are transmitted from source devices to the network. Receivers can
build lists of all available sources from which users can select.
Configuration is made simple with hardware nodes; you enter the names, numbers, and other configuration information through a web browser via an attached
PC. With PC nodes, it is even easier, requiring only the opening of a configuration
window to make all necessary source parameter changes.
Control
Most audio these days needs associated control. A delivery system needs a
start input at minimum, but could well benefit from a richer control dialogue such
as text identifying what is playing that can be sent to the studio mixer and to the
HD Radio and RDS encoders. Satellite receivers have control outputs. Telephone
systems need dialing, line status, hold, transfer, etc. Even a simple CD player needs
ready indication out and start in. Even the simplest source, a microphone, needs to
convey on/off status for the air lights. Most conventional controls have been done
with primitive GPIO parallel “contact closures.”
As a first step, Ethernet can transport GPIO data, reducing and simplifying
cabling, and Livewire offers this basic capability to replicate traditional start/stop
control. But it continues from there; Livewire also supports sophisticated remote
operation of studio equipment over Ethernet which allows the network to trans
Intro to Livewire
port much more advanced information than just simple start commands.
For instance, we can send the song title from a delivery system to a display on
a mixing console’s fader channel. Control of telephone systems and codecs can
follow fader assignment and be accessible from any location. With a high-bandwidth network linking everything and a flexible communication protocol, the door
is open to many interesting possibilities. Why couldn’t the satellite receiver identify
its content with “metadata” tags? Then an automatic system could store a program along with the information about it for later play. An on-air audio processor
might respond to program type information to adjust its parameters. Microphones
switched-on could activate a logger. There are many possibilities yet to be explored.
Livewire and PCs
One of the advantages of a Livewire system is that PC-based audio may be
directly connected to the network without soundcards. This means radio station
delivery systems can use the Ethernet connection they already have to send and
receive audio. Soundcard problems such as noise and multiple conversions are
avoided – the audio remains in digital form from the PC’s files to the network with
no alteration or degradation. Received audio may have originated from another PC
or from a hardware audio node. Audio sent from a PC may be received by other PCs
or hardware nodes.
With so much audio in radio stations being either played from computers or
recorded into computers, isn’t this a tremendous advantage? Not only do you save
the soundcard, but also the port that it needs at the other end to connect to your
console or router. And you can pass control and other information over the same
connection.
Support for Surround
The introduction of surround into radio broadcast was made possible by recent advances in multi-channel codec technology making surround a possibility
over European DAB channels and the USA HD Radio system. Experimental DAB
surround broadcasts are already underway in the UK and Sweden, and the US’ first
full-time surround station debuted in Boston early in 2007. Surround is big news in
Intro to Livewire
the home entertainment industry. Audio showrooms and computer shops are full
of 5.1 channel home theater systems. DVD-Audio and Super Audio CD disks offer a
surround reproduction format to serious audiophiles today. Some high-end cars already offer surround audio systems, such as the DVD-Audio player from Panasonic
pictured in the Acura TL.
AES 3 would be an impractical and expensive way to handle multichannel audio. The 5.1
system needs 6 channels: 2 front, 2 surround, 1
center, and 1 subwoofer. It might also be required to keep a separate stereo-mixed version
independently, so there could be 8 total audio channels. Using a traditional approach, that’s a lot of plugs, cables, router cards, and rack space!
On the other hand, Ethernet has plenty of bandwidth to carry the multiple
channels surround broadcasting will require. All eight channels plus associated
control could easily be conveyed on one convenient Ethernet cable.
Expanding a traditional console or audio router from stereo to eight channels
would be either impossible or very expensive. Not so with Ethernet and Livewire. In
fact, there is no additional cost for the core Ethernet switch because the one you
need for stereo would also be fine for surround, and audio from PCs can be multichannel.
We have designed Livewire with the future well in mind. It is already today providing the infrastructure for over 600 modern radio studios. These facilities are also
outfitted with surround capability – ready to implement simply and with low cost.
Audio Quality
We’re always asked, “Is Livewire like audio on the internet?” Yes and no. While
Livewire uses internet transport standards, it is intended to operate only over
switched Local Area Networks (LANs). Without the limitations of the public internet and with 100% control over all parts of the system, we are able to achieve full
studio quality.
So now the question would be, “Will Livewire, audio on the Internet and Audio
Intro to Livewire
over IP be reliable?” The answer is that Axia uses the same technology that underlies VoIP telephony.
Did you know that over half of the Fortune 100 companies now use VoIP? Or
that VoIP PBX systems now outsell the old kind by a wide margin? With these
systems, telephones plug into a standard Ethernet/IP network. Contrast this with
traditional PBX phone gear — proprietary devices which required you to purchase
phone sets and parts exclusively from the company that built the mainframe. You
were locked into a single vendor, because the technology that ran the mainframe
was owned by the company that made the gear (kind of like the TDM router companies).
IP is growing as a universal transport for almost any kind of signal. You see it
now in television studios, business teleconferencing, government communications,
banking, etc. And it’s hardly unproven, even for applications specific to radio studio
infrastructure. There are plenty of people successfully using it – now.
Fidelity
Internet streams are usually compressed for transmission over public links with
limited, variable bandwidth and low reliability. Livewire audio is not compressed
– we use studio-grade 48kHz/24-bit PCM encoding. Axia audio interface nodes have
more than 100dB dynamic range, < 0.005% THD, and headroom to +24dBu. LANs
offer a safe, controlled environment where there is no risk of audio drop-outs from
network problems and plenty of bandwidth for many channels of high-quality
audio without compression.
Indeed, we often hear from Livewire users that they notice an improvement in
fidelity when they transition from other systems. This is probably due to the direct
connection of PCs, the 64-bit accumulator in the Axia mixing engine, and to the
careful design of the audio stages of our Livewire Nodes, rather than the network
itself. But, in any event, the network takes nothing away from audio performance.
Delay
In packet-based systems, delay is an important issue and certainly has an effect
on your talent’s perception of “quality.” Packetizing audio for network transmission
Intro to Livewire
necessarily causes delay, and careful design of the system is required to reduce
this to acceptable levels. Internet audio delay is often multiple seconds because
the receiving PCs need long buffers to ride out network problems and the delays
inherent in multiple-hop router paths. However, with fast Ethernet switching on a
local network, it is possible to achieve very low delay. To do this, we must have a
synchronization system throughout the network. This also avoids sample or packet
slips that cause audio dropouts. Internet streaming does not use this technique, so
even if it were to have guaranteed reliable bandwidth, you still couldn’t achieve the
very low delay we need for professional studio applications.
For Livewire, we generate a system-wide synchronization clock that is used by
all nodes. Within each node, a carefully-designed PLL system recovers the synchronization reliably, even in the case of network congestion. Hardware nodes provide
this clock and in each system, there is one master node which sends the clock signal to the network. If it should be disconnected, or stop sending the clock for any
reason, another node automatically and seamlessly takes over.
In broadcast studios we care very much about audio delay in the microphoneto-headphones path for live announcers. Maximum delay must be held to around
10ms or announcers will start to complain of comb-filter or echo problems. We
need to consider that this is a total “delay budget” and that multiple links and some
processing will often be in the path. So we’ve decided to have a link delay around
1ms end-to-end for anything in this path, allowing us a few links or maybe a couple
of links and a processor before we get into links: one from the mic node to the mix
engine and one from the engine to the headphones out node. Thus, 2ms total.
Delay
Effect
1-3 ms
Undetectable
3-10 ms
Audible shift in voice character (comb filter effect)
10-30 ms
A slight echo turning to obvious slap at 25-30ms
30-50 ms
Disturbing echo, disorienting the announcer
> 50 ms
Too much delay for live monitoring
Here are the air-talent reactions to delay in a test conducted by Jeff Goode at WFMS in Indianapolis
In our experience, delays to around 10ms are not a problem. From 10-25ms announcers are annoyed but can work live; anything above 25-30ms is unacceptable.
Intro to Livewire
Another way to think about delay: Audio traveling 1 foot (0.3 meters) in air
takes about 1ms to go this distance.
And another data point: A common professional A-to-D or D-to-A converter has
about .75ms delay.
But, as is universally the case in engineering, there is a tradeoff – otherwise
known as the “if you want the rainbow, you gotta put up with the rain” principle. To
have low delay in a packet network, we need to send streams with small packets,
each containing only a few accumulated samples, and send them at a rapid rate.
Bigger packets would be more efficient because there would be fewer of them and
they would come at slower rate. But they would require longer buffers and thus
impose more delay. Big packets would also have the advantage that the necessary
packet header overhead would be applied to more samples, which would more effectively use network bandwidth.
With Livewire, we enjoy our rainbow and avoid the rain by having different
stream types: Livestreams use small and fast packets, while Standard Streams
have bigger and slower packets.
Livestreams require dedicated hardware and achieve the required very low
delay for microphone-to-headphone paths. PCs are not able to handle these small
packets flying by so quickly, therefore they use the Standard Streams. These are
compatible with Internet standards and can be directly received into the network
from PCs running delivery software. The network delay in this case is around 5ms
and the PC’s latency is likely to add perhaps 50-100ms more. Since PCs are playing
the files and they are not in live paths needing Livestreams, this is not a problem.
Our only concern is how long it takes audio to start after pressing the On button,
and delays in this range are acceptable.
Standard Streams can also be sent from the network to PCs for listening and
recording. Again, this small delay is not an issue – especially given that PC media
players have multiple seconds of buffering. However, off-the-shelf PC hardware
with a special operating system and software optimized for real time audio is able
to handle the fast streams. Indeed, we use this approach for our studio mixing and
processing engine.
Intro to Livewire
All Livewire hardware devices transmit both stream types and can receive
both stream types. There is no inefficiency from having both available because all
streams stop at the Ethernet switch and take no system network bandwidth unless they are subscribed to by a receiver or node. Each receiver takes only the one
it needs, taking the low-delay version if available, or the higher-delay version if not.
The selection happens transparently with no user action needed; users just select
the channel they want and audio is delivered by whichever method is appropriate
for the equipment they are using.
Livewire’s low-delay streams are also fixed-delay. The delay is constant, regardless of the system size or anything else. In fact, a source being received at multiple Nodes will have a differential delay of less than 5µs - less than ¼ sample at
Livewire’s 48kHz rate.
The Pac-Man of Protocols: Internet Standards
We use the internet’s IP standard for streaming media called RTP/IP for Standard Streams. RTP stands for Real-Time Protocol. It is the internet’s standard way
to transport streaming audio and video, just as TCP/IP is the standard for general
data. Both use the same underlying IP packet structure, but each has a header and
transmission method appropriate to the content.
Since we adhere to internet standards, your audio may be played by PC players such as Windows Media and Real that support standard protocols and uncompressed PCM audio.
Converged Networks
The headline at right taken from the Wall Street Journal nicely
captures what is happening in the telephone and networking
worlds: IP has become the “Pac-Man” of protocols, eating up everything in sight.
Major networking companies like Cisco, 3Com, and HP are dedicated to the
idea that a facility needs only one network for data, telephones, and media. Many
companies have jumped on this idea supporting the products that these companies are building today.
10 Intro to Livewire
Meanwhile, PBX companies like Lucent, Nortel, Mitel, Alcatel, and Siemens have
plunged into IP transport for their telephone products. This is bringing converged
networks, serving all needs from PBX companies too.
Ethernet might just as well be said to be the Pac-Man of local networks. It has
nearly a 100% share of new LAN installations and is the network that all VoIP phone
systems we know about use for connection to the desktop.
An Ethernet network being used for Livewire audio may be shared with any
other data transmissions such as file transfers, web browsing, and the like. An Ethernet system with a switch at the center may have a mix of audio nodes and normal
servers, PCs, et cetera. The Ethernet switch directs traffic only to where it is needed.
Even on a single link, traffic can be mixed because we use modern Ethernet’s priority mechanism to be sure audio packets have first call on the link’s bandwidth. A
studio audio delivery system could use this capability, for example, to download an
audio file from a server while simultaneously playing another, live.
Livewire adds to the convergence possibilities in a broadcast facility. We foresee
that you will eventually will have your computer data, telephone, audio, and control on a single network and that this will use computer/telephone industry standard wiring. Many of our customers are using this advantage today.
Until a few years ago,
there was skepticism
that Ethernet would
handle convergence
with services like
telephone and live
media being assured
reliable bandwidth
while sharing the network with computers.
A network technology called ATM was
proposed as a better
solution. But it was
expensive, difficult
to administer, and
would have required
a “fork-lift” upgrade
to existing systems.
So it never caught on
and has pretty much
faded from sight
for local networks,
although it has a
role at the core of
some Telco networks.
Ethernet’s switching,
priority mechanisms,
and increasingly fast
speed has put most
concerns to rest, and
all the vendors who
offer VoIP telephones
connect them over
Ethernet, not ATM.
Intro to Livewire 11
2: What can you do with LIVEWIRE?
Imagine everything that you can do with a PC connected to a network: Share files, send
and receive emails, chat, surf the web, listen to audio, etc., etc.. PCs and networks are
designed to be general-purpose enablers. You have a similarly wide range of possibilities for audio applications using Livewire. Here are examples, starting with the most
simple, and continuing to the most interesting.
Make A Snake
Concert sound guys need
to get a lot of audio from the
stage to their mixing consoles
in the center of the house.
They call the multi-conductor
cables they traditionally use for this function a “snake”. LW lets you put such a snake
on a diet! A single Ethernet cable connects multiple audio channels. Add a switch
at each end and you can have as many nodes as you want. Use Gigabit Ethernet
and you can have hundreds of channels. Add fiber optic media converters and
cable to extend the distance between units to many kilometers. Maybe you need to
get something from here to there?
A High-Performance Sound Card Replacement
Livewire can talk directly to PCs, making the network
look like a soundcard to delivery systems, editors, etc. Axia
LW nodes have excellent audio performance: Balanced I/O
with more than 100dB dynamic range, < 0.005% distortion,
headroom to +24dBu, etc. They make excellent multi-channel “soundcards” for professional applications. You can position the node at a distance from the PC, and you get balanced audio on connectors
that are a lot more reliable than mini phone jacks.
With the addition of an Ethernet switch you can feed your audio to multiple
12 Intro to Livewire
computers and/or have multiple I/O boxes – which take us to the next application…
Build An Audio Router
A system with Livewire nodes, one or more Ethernet switches, and PC-based routing controller
software make an excellent facility-wide audio router. PCs send and receive audio directly to the network without soundcards or audio ports, thus lowering costs and eliminating conversion steps. Telos
and Omnia telephone, codec, and processing equipment is now able to connect directly, as are numerous products from other broadcast manufacturers such as International Datacasting, AudioScience and
Radio Systems, to name a few
(for a complete list, see ­
www.AxiaAudio.com/partners/) .
To interface conventional analog and AES signals,
Livewire interface nodes come
in a number of versions. One
node operates like a traditional audio router X-Y control
panel. But with a difference: audio in and out is available on the same box.
A PC-based router control package, called PathfinderPC, is available that makes your whole system
look like a single entity. You can control which outputs are connected to which inputs just as if the system were a single-location box.
Since there is no requirement for a mainframe, the base cost is low – you can make a small system at
very reasonable cost and expand it over time. Indeed, the total cost of a large system will be much lower
then older approaches due to the use of commodity switches at the core. Just as using standard PCs to
play audio makes much more sense than any proprietary approach, building routers from common computer industry parts makes similar sense. Indeed, this approach gives you a true “audio network” quite
unlike other approaches.
Intro to Livewire 13
Build A State-Of-The-Art Broadcast Studio
Plug an audio processing engine and a control surface into the network and
you have a modern radio studio with many advantages over the old way:
•
Simplified and unified cabling for audio, control, general data, and telephone.
•
No multiple conversions. With most studio audio coming from or going
to PCs, audio is kept in the networked digital domain. Audio may be monitored on any PC with a player such as Windows Media, Real Audio, etc.
•
Integrated data means you are ready for synchronized text and metadata, which is needed for HD-Radio in the USA. It will also be possible for
audio processor parameters to be controlled depending upon source characteristics.
•
Tighter integration with delivery systems means that mixing, scheduling,
and playing can work together. For example, song titles can appear on the
mixer surface, start and other control functions may be conveyed over the
network, and logging can confirm that an audio piece was really played on
the air.
•
Troubleshooting and repair are transformed. Extensive diagnostics are
available over the same network that connects the audio. A suspect surface
or engine may be swapped by re-plugging only one Ethernet cable.
•
Low-cost power. Computers replaced cart machines because they are a lot
more powerful, convenient, reliable, and cheap. The technical side of radio
broadcasting is tiny compared the computer and networking industries. We
get tremendous value by plugging into the massive R&D and production
scale offered by the computer world. Leveraging low-cost mass-produced
computer components makes the same sense for studio mixing and audio
distribution as it did for cart machine replacement.
•
Surround-ready. As one would expect from its flexible computer technology-based origins, Livewire readily adapts to future technologies such as 5.1
surround.
•
14 Intro to Livewire
In the example below, a Livewire-based system is being used as a studio console. Sources such as microphones and CD players are interfaced to the network
with a node in the studio, while sources such as network feeds interface with a
node in an equipment room.
Certain peripheral equipment connects
directly to the network. Audio from the delivery PC goes to the network via an Ethernet
connection and control is also over the network. Telos Zephyrs or Telos hybrids such as
the Nx12 can connect directly to the switch
to make the audio and control connections
much simpler, only one RJ-45. The network also supports file transfers to the delivery system from a server. The studio operator surface controls a rack-mount mix
engine, which has a single Ethernet connection for both control and audio.
Make a Flexible Two-Way Multi-Channel STL
Studio and transmitter sites may be linked with “Ethernet STLs”. LW nodes provide
audio interface to Ethernet point-to-point radios. These are off-the-shelf today
from such companies as Motorola, Aeras Networks, Dragonwave, MikroTik and many more. In addition to the audio, anything that can be
carried over Ethernet can be conveyed over the radio link, such as VoIP
telephones, email, file transfers, and transmitter remote control.
Ethernet radio systems are available that can connect at speeds of
more then 100Mbps. This data rate would support more then a dozen
stereo uncompressed audio channels in each direction, with capacity
remaining for VoIP telephone and facilities control. Seems these radios
would make an interesting two-way RPU also. For co-owned stations
that are not co-located, these could be an effective way to link studio
facilities.
NOTE: Many of these systems are optimized for speed vs low error rate and
Intro to Livewire 15
therefore may not work. Axia has evaluated several units and we can offer guidance if you are interested in pursuing this option.
Create A Facility-Wide Audio Network That
Includes Integrated Studio Consoles
Combine all of the above for maximum power, convenience, and flexibility. You
get facility-wide audio routing, state-of-the-art studio mixing, a single wiring infrastructure for audio, computer data, control, and telephone.
Audio processors with Livewire ports may easily have multi-channel outputs,
such as for simultaneous analog FM, HD Radio, and low-delay monitoring feeds. A
single Ethernet would serve for all needed inputs and outputs. With a data capability alongside the audio, it would be possible to control processing parameters
depending upon which audio source is active.
Create An Integrated National/Local Radio
Network
Imagine a satellite transmitting IP packets. Now live audio, audio to be stored
for later play, and identifying data can be delivered. Wouldn’t this transform radio
networks into something much more interesting, useful, and powerful? Including
an Internet-based return path which adds a whole new dimension.
16 Intro to Livewire
3: The components of Axia Livewire
Livewire is not only a technology. It is a solution, made for broadcast. Here are the essential pieces that put Livewire to work for you.
A Livewire system usually has a mix of hardware nodes and PCs with driver software that lets them send and receive Livewire audio streams. There will also be one
or more Ethernet switches, unless you are making only a very simple 2-box snake or
a PC soundcard replacement. This section gives an overview the current available.
Switches are covered in another section.
Mor and more manufacturers are offering equipment with onboard Livewire
jacks and we expect many more in the future. We are also now including these
ports on new products, meaning you can now get Telos telephone hybrids and systems, Zephyr ISDN codecs, and Omnia audio processors with Livewire connectivity.
Livewire Hardware Nodes
These interface analog and AES/EBU audio to the Livewire network. Some are
used to interface GPIO such as for starting CD players or lighting on-air signs. Configuration and monitoring is via a networked web browser.
Analog 8x8 Node: Eight balanced inputs and outputs with more than 100dB
dynamic range, < 0.005% distortion, headroom to +24dBu. Software controlled
gain lets you trim adjust to accommodate different levels. Front panel LED audio
level metering.
AES/EBU 8x8 Node: Eight AES3 inputs and outputs. An input can be used to
sync your Livewire network to your house AES clock, if desired.
Intro to Livewire 17
Mic + Line Node: Eight microphone inputs with very high-grade pre-amps,
Phantom power, and eight balanced line outputs. Intended mainly for on-air and talk
studios.
Router Selector Node: Emulates the function of traditional x-y audio router
controllers, but includes a single on-board input and output in both analog and AES3
digital forms. The LCD presents a list of active audio channels, which are selected with
the adjacent knob. Programmable “radio buttons” offer immediate access to oftenused channels. The unit could be used for equipment room monitoring and production studio or newsroom audio interface. Also useful as a test instrument to check and
generate audio streams or change feeds to codecs.
General Purpose Input/Output Node: This GPIO interface for parallel closures
has eight DB-15 connectors, each with five inputs and five outputs. Interfaces control
to CD players, delivery systems, on-air lights, etc. that need simple parallel control.
The Element power supply also offers identical GPIO functionality.
Axia IP-Audio Driver for Windows®
This is the software interface between your PC audio applications and the
Livewire network. Components are included that provide various interface capabilities.
16-in/16-out Driver :This is a driver that interfaces sixteen inputs and sixteen outputs and is available through one of our many delivery system partners. It provides
18 Intro to Livewire
these functions:
•
Interface for audio sent to Livewire from audio applications such as delivery
systems and other audio players.
•
Interface to receive audio from Livewire into applications such as audio editors.
•
A GPIO function to convey “button-press’ data from the network to applications,
such as from a control surface fader start button to an audio player.
1-in/1-out Driver: This is a one input and one output version of the above and
is available for use with any delivery system and editor that support standard Windows audio. It provides these functions:
•
The Axia Windows® Driver connects PC audio directly to the network via Ethernet and without sound cards.
•
Provides one stereo input and one stereo output via the Livewire network.
•
Audio applications see the Livewire network as if it were one or more standard
sound cards. A sample rate converter and clock generation functions are included.
Intro to Livewire 19
iPlay (PC Router Selector)
The second application in the LW Windows Suite is an interface to display and
select Livewire streams – essentially a software version of the Router Selector. The
selected audio is sent to any audio application that works with standard Windows
sound cards. The Preview function lets you listen directly without another application.
Sources are listed for selection with a mouse click. They may be filtered by category.
There is a capability similar to the radio buttons on the hardware Router Selector. Dragging a listed source to one of the buttons allows it to be used to quickly
select a desired source.
Media Player Interface: Streams can be adapted for listening by standard
internet audio players such as Microsoft Windows Media and Real players. The list
of Livewire streams is presented within the player’s usual interface as if they were
standard internet streams.
20 Intro to Livewire
Element Broadcast Console & StudioEngine
With all audio sources in your facility available on a single Ethernet jack, the
door is open to new ways of mixing and processing audio signals. We are now able
to build a low-cost, but very powerful mixing/processing engine that subscribes to
networked audio streams, modifies them and presents the resulting streams back
to the network on that same jack.
StudioEngine: The Axia StudioEngine is a powerful processor designed to add
console functions to a Livewire audio system. The Studio Engine performs all the
mixing and signal processing functions that would have been performed in the
past by an audio console. Of course, a Livewire-based routing system may be used
with any traditional console, but integration brings many advantages.
Each engine can perform all the mixing and processing functions needed by
even the largest console, with per-channel mix-minus feeds, multiple outputs and
monitor feeds, EQ, mic processing, et cetera. There’s plenty of headroom to support
future features. One StudioEngine is required for each radio studio.
The front panel display on the Studio Engine provides confidence feedback. The
selector knob allows you to easily perform basic configuration. As with all Livewire
components, web-based interaction is used for more advanced configuration.
Of course, operators still need to have control interfaces. The Axia Element
shown next is a StudioEngine-compatible control surface. It, too, connects with a
single Ethernet plug.
Element Broadcast Console: Designed for the needs of live programming,
Element provides your on-air staff with a familiar and comfortable set of controls
in an uncluttered and intuitive format. With a lot of broadcast experience under
our belts, we worked carefully to keep the basic functions simple and trouble free,
but still include all the sophisticated functions of large traditional consoles sup
Intro to Livewire 21
ported in a deeper layer. Metering, time, timer, and essential status info are clearly
presented on an adjacent SVGA LCD monitor of your choosing. Pressing the Option
button on monitor or fader channels brings up all the fancy stuff. All sources in
your Livewire system are listed and available for selection, and there is pan, EQ, L/R
select, send bus access, mic processing, headphone EQ, and more.
But it goes further. As you would expect from Telos and Axia, our control surfaces have a smart approach to mix-minus for phones and codecs. Every channel has
the ability to provide a mix-minus output automatically. Operators simply select a
phone or codec source and the backfeed is automatically generated based
on preferences established when the user profile was configured. There
is a single button that selects a Phone Record mode when users need to
record phones off-air for later play.
Clear, bright LED text labels show the active source for
each fader, and icons indicate status when needed.
Element can save profiles for each user, allowing different preferences, layouts and
defaults for a variety of shows and talent.
In addition to console functions,
­Element provides controls and displays that interact with phone systems, codecs, editors, PC delivery systems, and other broadcast gear.
Together with the StudioEngine, Element was designed to meet all the
console/control needs of the most demanding live and live-assist radio operations.
PathfinderPC Routing Control Software
Distributed Livewire systems can be controlled just as if they were traditional
centralized audio routers. In this case, you will need a way to control the multiple
nodes as if they were a single device.
We offer a PC software package called PathfinderPC, developed in cooperation
with a partner, Software Authority, which specializes in router controls. It is a client22 Intro to Livewire
server system that serves as a front end for X-Y style router switching. The server
communicates will all of the Livewire nodes in your system, and offers a common
point of control to clients. Multiple clients can connect to the server to provide any
number of control points. Each client may be optimized for a particular style of
operation or control features. For example, a master control client will probably be
very different from a controller within a studio.
Scenes (presets) can be created and recalled to allow changes to the local
studio or to the global network. A “virtual patch bay” function provides an intuitive
way to manage routes. The server and clients run on Windows PCs.
Because Livewire nodes put audio level information onto the network, PathfinderPC clients are able to display level metering. This is indicated with on-screen
cross point icons – green dots indicate the presence of audio. Users may also select
accurate multi-segment meters for audio sources they want to check carefully.
You can use PathfinderPC to make “virtual routers,” which can be subsets of the
real routers. For example, if a
Livewire system has 128 different sources and destinations on
the network, but you may only
wish to use a small number of
these points in a particular studio area. You can create a virtual
bay that specifically includes
only the sources and destinations required by this studio. This
virtual router can have its own
set of “snapshots” (scene changes). The virtual router also allows you to map multiple points to a single virtual
point. For example, you can make a virtual source and destination that contains
both the audio inputs and outputs for a particular device, and also the GPIO points.
Thus when the route is made, both audio and GPIO is routed simultaneously.
PathfinderPC supports non-Livewire routers including video routers and ma
Intro to Livewire 23
chine control routers. Thus, you can make routing points in the virtual bay which
will simultaneously route audio, video, GPIO, and Machine Control. This makes the
software ideal as a master centralized router control package. Software Authority
continues to expand our list of supported products, and the software is designed
to allow us to add support for additional protocols and routers quickly.
PathfinderPC supports the use of tie lines or gateways between routers. For
example if a system has both an analog video router and an SDI video router, one
or several tie lines can be wired through Analog to SDI converters between the two
routers. PathfinderPC will then combine the routing tables and automatically use
the tie lines when necessary to get analog sources to the SDI router. The complexity
is hidden from the end user. This capability allows Livewire terminals to extend an
older and already filled router.
Provisions for Redundancy and Back-up
PathfinderPC includes a silence detector that allows you to place a “watch” on
a particular Livewire destination. If the audio level falls below a specified threshold
for longer than a specified period of time, the system will switch to a backup audio
source. This lets you build automatic redundancy into a signal path. If the primary
and backup sources and destinations in the silence detector are assigned to different Livewire units and these units are wired to different AC power sources, the
signal path can be maintained even in the event of a failure of a terminal or power
source.
In addition, multiple Pathfinder servers can be simultaneously monitoring
the Livewire network, building redundancy into the control system as well. The
Livewire system is an ideal tool for building a redundant audio chain. Since every
audio unit is an independent device, the server can automatically switch audio to
a different unit if the usual one fails. With careful planning, you can arrange your
system so that the primary and backup audio units are connected to different LAN
switches which are chained together using the inherent Ethernet redundancy protocols. Thus audio is continuously and reliably passed, even in the event of a LAN
switch failure.
24 Intro to Livewire
Timed Events
PathfinderPC has a simple timed-event system built into the server with which
you can program events to happen at specified times. Individual routes or snapshots (scenes) can be triggered at a particular time and date or on a rotating schedule on certain days and times of the week. Events can also be created which will
monitor a GPIO and initiate a snapshot (scene change) or route whenever a GPIO
condition changes.
For more sophisticated timed operation, external automation systems can access and manipulate the routing tables provided by the Pathfinder server using the
protocol translator. Multiple protocols may be simultaneously translated and connection may be on either IP/Ethernet or serial ports.
Livewire Audio Router Control Protocol
We provide a documented protocol for those who want to develop their own
controllers.
Applications designed for controlling traditional audio routers can implement
Livewire Audio Router Protocol directly or use a software gateway between this
protocol and their native protocol. The first solution may be preferable, as it enables applications to fully control every Livewire unit and is free from potential
problems with the gateway reliability. To avoid multiple TCP/IP connections, the
gateway solution may be used. In this case, there must be gateway/translator software developed for each protocol that has to be supported.
Livewire Routing protocol assumes multiple audio input/output nodes. Every
node has its unique IP address, N input ports and M output ports.
We offer a software interface that emulates a traditional router and does the
mapping and translation. Input to this module can be either serial port or TCP/IP
over a network. Network configuration of Livewire devices can be communicated
to this program using command line or a text configuration file. There is a TCP/IP
server waiting on every Livewire node. The client simply writes text commands to
the appropriate device.
Intro to Livewire 25
iProbe Network Management Console
Axia iProbe is an intelligent network maintenance and diagnostics tool that makes
managing, updating, and remote controlling a Livewire system very easy. There’s
an auto-documentation
feature that generates
configuration docs for
every device. The organizer lets you group Audio
Nodes into logical groups for
easy management, upload software to single or multiple devices, make device configuration
backups and more. The status of every devices connected to the Livewire network
is immediately available from a single point of control.
iProFIler Automated Program Archiving
iProFiler logging software simultaneously captures up to eight stereo
audio channels to timestamped MP3 audio logs
directly from the Livewire
network — no audio cards
required. iProFiler monitors
GPIO channels you select and begins recording when
those channels activate. Record mode can be set for
logging, skimming, or a combination of both. Logged audio can be stored locally or
remotely, and may be auditioned via LAN, WAN, or Internet, if you choose.
26 Intro to Livewire
Networked Audio Processing
With the addition of Livewire connections to the latestaudio processing gear, it is
possible to route program streams directly from control room to processor without
any interim conversions and in fact, without leaving the network at all. At this writing, several new Omnia audio processors with Livewire capability have begun shipping, with plans for more in the works.
Omnia ONE is a single-rack-space, full-featured broadcast audio processor for FM and HD Multicast. Designed
to provide main station audio processing, its features include 24-bit processing, four-band AGC plus wideband
AGC, and a four-band limiter. Omnia ONE also contains a fully distortion-controlled
final limiter/clipper, an integrated digital stereo generator with advanced peak control, and GPI remote control ports, are of which are accessable and configurable via
Web browser. Analog and AES/EBU I/O is provided for compatibility with legacy
audio chains, but with Livewire I/O, it is possible to build an audio chain that remains
completely within the networked digital domain.
Omnia.8X features eight discrete 3-band stereo Omnia
processors in a single box. Imagine what can be done
with this sort of high-density processing power: a portion of the processor’s resources could be dedicated to
DAB, Multicast or Web audio channels, while the rest is reserved for on-demand processing of in-studio performances, remote broadcasts, or any other audio channel
you might want to give some “sweetening” to. All of the processor’s I/O and control
travels on a single-cable Gigabit connection to the Livewire network, making connection, setup and configuration simple.
Networked Codecs
Broadcast audio codecs can also become a native part of Livewire networks, of
Intro to Livewire 27
fering the same networked control, consolidated I/O and service density described
above. Here are some examples now available from Telos:
Zephyr iPort MPEG Gateway enables broadcasters to transport eight bi-directional channels of stereo audio across any network with guaranteed QoS, such as T1
and T3 connections, MPLS networks, et cetera. Each iPort contains eight stereo MPEG-AAC codecs, and connects to Livewire
networks using a single CAT-6 cable for all I/O. The Zephyr iPort
makes use of Livewire’s advertising system in such a way that
when two remote facilities are connected via its gateway, each
facility’s audio streams are visible to the users on the other end of the link. Operators
can then select desired streams by name, and consume and control them just as if
those streams were local audio sources.
Zephyr Xstream ISDN Codec, the most widely-used broadcast codec in the
world, now has Livewire capabilities. It has all the capabilities
familiar to Zephyr users, with the additional functionality of a
Livewire interface. With this addition, the codec can become
a truly shared resource within the broadcast facility, ebanling
users in-studio to take control of an idle codec and make use of it on-demand, from
anywhere inside the studio complex.
Zephyr/IP (Z/IP for short) is the natural product of the work
on codecs, streaming, and Audio-over-IP that Telos has been
conducting for years, making generic IP networks as reliable
and easy to use as ISDN. Z/IP features lots of new tech and the
latest Fraunhofer algorithms to deliver broadcast-quality audio
over IP connections from 16kBit to 256kBit via UDP/RTP and
TCP/IP, with network edge traversal (NAT) and built-in directory
assistance. With Z/IP as a shared resource of a Livewire network,
broadcast remotes can remain IP-based from begining to end.
28 Intro to Livewire
Networked Telephone Systems
Besides the consolidated I/O, one of the advantages of
equipment that connects via Livewire is the ability to bring
control directly into the console. Phone systems can natrually
benefit from this type of distributed control. One of the most
concentration-breaking operations for air talent has alsways
been phone operation; removing the operator’s hands and
eyes from the console to operate the phone control next to it
can easily disturb the flow of on-air productions. Telos phone
systems such as the Nx12, Series 2101 and TWOx12 make use of
a console module specifically designed for on-console control
of the phone system, enabling operators to keep their eyes on the board whil simultaneously taking callers, leading to smoother shows and fewer errors.
Also, phone systems with Livewire can now be used as shared devices. The Nx12,
Telos’ first phone system a native Livewire connection, includes 4 hybrids and two
Program-on-Hold inputs; this mix of features means that Nx12 could easily handle
telephone needs for two
stations simultaneously or,
thanks to its Livewire connectivity, could be treated
as a shared device usable
from more than one studio at a time.
Intro to Livewire 29
4: Nuts & Bolts: Making Livewire play
Now we move to making audio happen. Time to take the gear out of the shipping
carton and make it play. This section gives you practical information. Details about the
underlying tech are reserved for later.
Livewire’s Channel and Name system
An advantage of having a data network carrying our audio streams is that we
can send identifying information on the same cable and system. Receivers can
build tables of available audio, and testers can identify specific streams on a cable.
In Livewire, we have both a numeric and a text ID for each audio source.
Hardware LW devices are configured either using a networked PC’s web browser, or with local pushbuttons and displays. PC LW nodes will have a configuration
window that opens when you click on the application icon. Details for each are in
the manual specific to the product, but the general approach is the same for all
audio and GPIO.
Channels
Channel numbers may range from 1 to 32767. You assign these to audio sources as you wish.
New units are pre-configured from the factory to start with channel 1, thus an
8-channel node will come assigned to channels 1-8. Two new units can be connected to each other with a “cross cable” (described later) for immediate out-of-thebox testing. For your network, you should reserve channels 1-8 for testing and not
assign them for routine use. Then, if you plug a new unit into the network before
you configure the channels, there will be no problem with conflicts.
In a large system, you will probably want to have a people-friendly naming and
numbering system that reflects studio use or location and to help prevent accidental duplication of channel assignments (a big no no by the way). You have plenty
of numbers to use, so you don’t have to conserve them. For example, the channels
associated with Studio 1 could start with 100, Studio 2 with 200, etc. There is no
requirement that channels be assigned in order or contiguously from a multi-channel device.
30 Intro to Livewire
Devices such as telephone hybrids and codecs need audio in both directions,
so when appropriate, a single channel contains a “to device” audio stream as well
as the usual “from device” audio. In this case, you can think of the channel as something like a telephone number that connects a call with audio in both directions.
The advantage of this “bundling” of the two audio directions is that the association
is naturally maintained when studio mixers are in the picture. Mixers generate the
feed to devices (usually mix-minus, but not always) and automatically assign it to
the source channel number, and this association is kept regardless of which fader is
being used, etc.
Text Name
The text name may be up to 24 characters and you choose it as you wish. This is
what will appear on the Router Selector’s LCD, studio mixing surface source select
lists, etc. Most devices are not able to display all 24 characters, so will truncate to
show what they can. The Router Selector, for example, can display 16 characters.
You may wish to include in the name the rack number or room name of where the
Node is located, to help orient yourself in the case of a future emergency.
A typical name might be ST1CD2 for Studio 1, CD Player 2.
Our studio mixing systems (Element or SmartSurface, for e.g.) automatically
generate return feeds to devices that need them, creating the text name for these
in the form “To: name”. For example, if you have a source called “Hybrid 1”, the mixer
will generate an audio stream named “To: Hybrid 1” and advertise it to receivers.
GPIO
There are also GPIO (General Purpose Input/Output) channels and text names.
These work in a fashion very similar to the audio source channels and names.
GPIO channels often share the same channel number as an audio source. A
typical situation would be when you have a CD player that needs start control from
an audio mixing console. The mixer automatically generates the start command
and puts it on the channel number you assigned to the audio source. To cause a
particular hardware GPIO to output this command as a contact-closure pulse, you
configure the GPIO device to listen to this channel. As with the back audio, control
Intro to Livewire 31
follows the audio source to whichever fader is being used.
But GPIOs may also be independent of audio sources. In this case, the Livewire
system provides a pass-through function where outputs follow inputs – sort of like
a GPIO distribution amplifier.
Sources vs Destinations
We’ve always struggled with terminology when referring to audio input/output
from devices such as codecs and hybrids where there’s local audio I/O as well as a
combined network I/O port. We will try to be consistent within the Livewire realm
by using the following terms:
•
Source – this is an audio input to a hardware Node and therefore available on
the Livewire network as an audio stream that can be accessed by other Livewire
nodes. Of course a StudioEngine can generate new audio sources and in this
case there is no associated hardware audio input.
•
Destination – this is an audio output from a hardware Node and therefore represents playback of some stream from the network. Of course a StudioEngine or
Livewire capable audio device may access a Livewire stream and in these cases
there would be no associated hardware audio output.
So, to reiterate, sources represent the feeding end of the audio stream equation
whereas destinations are just that, one or more destinations where that stream is
used.
Examples
Following Gauss’ dictum that “an example is worth two books,” let us now turn
to some to show you how Livewire’s
channel and name identification
work.
On the next few pages we’ll show
some of the web configuration pages
for the 8x8 Analog Node. The first
example is the home page that is
displayed once you have logged into
32 Intro to Livewire
the node. It simply lets you navigate to the other configuration screens.
The Configure Sources page shown above permits you to configure locally
generated sources. It allows you to assign names and channels to the sources that
will be generated by this node, and to configure the audio inputs associated with
those sources.
•
The Name entry at the top is where you put the text ID for the node. There are
8 Source Name entries, one for each audio channel. This is the text name for the
individual audio source.
•
Source Channel is where you enter the channel number for each source.
•
Mode allows you to decide whether each stream should be a Livestream, Standard Stream or Surround Stream. Why would you use one versus the others? A
satellite feed will never be in a mic-to-headphones path, so only the Standard
Stream would be required. A microphone source is live audio, and so would be
a Livestream. Surround streams are only used if yours is a 5.1 Surround broadcast facility.
•
Our example node has selectable gain for inputs. You can choose the appropriate value with Input Gain (the range of values will depend on the node to be
configured). This can also be set on the Meter screen, in case you desire to set
gain “by eye”.
Intro to Livewire 33
The sample Destination Screen, pictured here, is used to configure the local
units’ outputs, or destinations. This is where you configure the output channels,
with the menu options described below.
Hot Tip! You will only
have a complete list
of audio sources if
you have already
configured all your
source nodes and
have them connected
and operating so that
they are advertising
to the network. You
can always just enter
the channel number
here if you don’t yet
have your source
nodes working, but it
will be more convenient if you prepare
all your sources
before moving on to
destinations.
•
As with sources, you can enter a text Name to be associated with the destination
associated with each output port.
•
Channel is where you tell the unit which audio stream is to be output from each
audio output. As previously discussed, each Livewire stream is identified by
both a text name and with audio channel number. You can enter the channel
number directly, or you can use the button to the right of the channel entry
to open a page that gives you a list of all the active audio channels, and can
choose from among the channels listed within it. Usually the list contains text
names of audio streams, but if no text name has been assigned, you will see the
device type and IP number instead.
•
Type lets you choose whether the Destination feed is one-way (outbound to a
monitor or headphone feed, or a record input) or two-way (supplying a backfeed for a phone system or codec).
•
Load and Gain appear only in the Analog Node Destination screens to allow
you tailor specific outputs to feed High-Z (professional) or 600-Ohm (prosumer)
gear.
•
34 Intro to Livewire
The Meters page for our example audio
node (shown at left) lets you monitor the levels
for each input and output channel. This is also
an alternative to the source page for setting
input gain.
The System and QoS pages shown below
permit checking the IP address and related
settings. The QOS page is an advanced feature
page, and its lets you set some other values
such as the IP address for the unit, stream
characteristics, and clocking mode. These are
described in detail in the unit’s manual, but to appreciate the context, you will also
need to understand more of the Livewire and networking basics described later in this document.
Our samples so far have been from the 8x8 Analog Node. Next let’s look at how
sources and destinations are handled by the IP-Audio Driver used on Windows™
computers.
Soundcard Emulation. The IP-Audio Driver looks like a standard sound card to
Windows™. Each of the Drivers eight sources (e.g. streams originated by this computer) shows up as a sound card, called Axia Wavexx. You can define one of these
Intro to Livewire 35
Livewire sources as Windows’ Preferred Sound Playback Device from the
Windows™ Sounds and Multimedia Properties Control Panel as shown here.
Driver Configuration. The Axia IP-Audio driver is configured for
sources and destinations much like the Axia audio nodes. The driver is
configured from the window shown below; the various settings are described as well.
•
Sources and Destinations – You can see that the node and source nam-
ing and channel idea is the same as for the hardware nodes. Any audio
channels you want to receive are entered into the Destinations boxes. If
you don’t know the ID number, you can choose from text lists instead, by clicking on the Browse button.
•
Livewire Network Card – A PC running this driver may have two network cards, one for general data
and another for audio streams. The Livewire Network Card entry let’s you associate Livewire audio
with the appropriate card.
•
Advanced – Clicking this brings up a screen that lets you set stream characteristic values. This is covered in greater detail in the Axia IP-Audio Driver manual.
•
Statistics – This button brings up a screen with lots of information useful for debugging network
problems.
When you have finished configura-
tion, the Livewire network looks like a
sound card to any Windows application that uses the standard wavin/out
audio interface. In Windows applications where you normally select the
soundcard you want to use, you will
select a Livewire channel instead. In
the example, an audio player that has
selected Axia Wave01 will put its audio
into Livewire stream channel 11111 and
be available to all LW devices on the network.
36 Intro to Livewire
The GPIO node is a hardware box with 8 DB-15 connectors, one for each port.
Each has 5 inputs and 5 outputs.
GPIO channels may be
associated with audio channels or may be independent.
If they are independent, they
must not use the same number as any audio channel
– they share the same “channel space”.
You can monitor the status
of each with the indicators at
the top of the page.
Hardware Node & Configuration Access
As shown in our examples above, you’ll need to configure various parameters in the hardware nodes.
Some very basic parameters such as the name and IP address can be configured from the front panel. In
fact, for the basic audio snake application we need not even access the nodes’ web pages. However in
most cases we will need to do so, but we will need to assign an IP address first.
Front Panel Node Configuration
A number of items can be programmed from the front panel of most of the hardware nodes. This is
covered in more detail in the individual manuals. However we will cover setting the IP address here since
you’ll need to assign and IP address to enter via the web browser, and we’ll cover that next. Here’s how
to assign an IP address to a typical hardware node, such as the 8x8 Audio Terminals (the GPIO node must
have it’s IP address configured using a BootP server, see the GPIO Users Manual for details).
Configuring Node IP address
Each Livewire™ node must have a unique IP address. The only exception is when two nodes are connected in the point-to-point configuration.
Intro to Livewire 37
To program the node’s IP address follow these steps:
1. Starting from the metering screen, press the <SELECT> button once. The default IP address is “0.0.0.0”, so unless the unit has previously been programmed,
the screen will show “000.000.000.000”.
2. Press and hold the <ID> button for 4 seconds. A blinking cursor will appear
below the first digit. Use <SELECT> to change the digit indicated by the cursor
(each press of this number will increment the displayed digit by one).
3. Press the <ID> button to jump to next digit. Use <SELECT> to change the digit
indicated by the cursor. Continue until all digits of the IP address have been
entered.
4. Once the changes are complete, press the <ID> button repeatedly until no cursor is shown then press <SELECT> to exit. If you do not wish to save your changes
do not press <SELECT> after reaching the last digit; after approximately 10
seconds the meter screen will return and the old settings will be restored.
The node’s IP address can also be remotely assigned over the network using a
program included with the your node called BootP (with some nodes this is required). To do so follow these steps:
1. Start bootps.exe on any Windows 2000/XP PC. You will see the following screen:
2. Hit ID button on
GPIO front panel. You
will be prompted for
new IP address :
3. Enter new IP address and press ENTER.
38 Intro to Livewire
Make note of the IP address you have entered so that you can access the Node
using a Web browser (see below). You can now continue to assign additional Node
IP addresses, or shut down the Bootp program.
Accessing a Node via a web browser
To access the built in web server from a computer, the computer and node
must be connected to the same LAN (or the computer and node can be connected
using a “crossover 10/100 Base-T” Ethernet cable). To connect enter the following in
your browser:
http://123.456.789.101
where “123.456.789.101” is the IP address of the node to be configured.
Note that:
•
The IP range (e.g. the first three numbers of the four numbers of the IP address
of the computer and the node must match, or additional configuration will be
required.
•
Microsoft Internet Explorer version 5 and later, and Mozilla Firefox have been
tested with Livewire Nodes. Other browsers may work, however they have not
been tested.
Your browser should now display the login window to allow
you to access the node. Enter a valid user name and password
and click on “OK” to log in.
•
The default user name for all Axia nodes is: user
•
The default password for all Axia nodes is: <enter>
Once you have logged in you will see the Axia node home
page as shown earlier.
Intro to Livewire 39
Plugs & Cables
Livewire systems use primarily copper cables, but you can add fiber where it
makes sense. We’ll start here with copper.
An important goal in the design of Livewire was to simplify installations. One
of the ways we do this is to let you standardize on a single cable type, plugs, patchfields, etc. This is consistent with the modern way of thinking about cabling in
office buildings where a common type can serve different applications. You can
use the same connectors and cables for everything in your plant. And for big new
installations, outside contractors can install and test the wiring infrastructure for
everything. In fact this is one reason why we suggest that Broadcast Engineers
should become familiar with the relevant standards such as EIA/TIA-568-A & B.
The 100BASE-T Ethernet we need for most Livewire devices requires RJ-45 8-pin
modular plugs and jacks. So, we’ve standardized on RJ-45s for balanced high-level
analog and AES connections as well. There are a lot of connectors being used for
analog audio these days, so why did we go this route? The reasons are cost, density,
compatibility, and convenience. RJ-45 sockets and plugs are a lot cheaper than
other choices, both for us at the manufacturing level and for you at the time of
installation. Density is an important advantage: We can get only a few XLRs across
the rear panel of a 1U rack box and we need two of them for each stereo connection. Our basic nodes would have to be 2U to have the same channel capacity as
we have now with 1U nodes. A single RJ can do both channels on one jack and we
can fit dozens of them on a 1U box. XLRs and DBs need to be soldered, and shells
assembled, etc. Molexes need a separate crimp for each wire and are not standard.
RJ crimping is convenient procedure compared to the others. And you will already
have the plugs, cables, and tools at hand.
The following tables summarize the cable types that could be used in a LW
system and their applications.
40 Intro to Livewire
Non-Ethernet Cabling Relevant to Livewire Systems
Description
Cable
Analog Audio, balanced, high-level
Usually shielded Cat. 5, but unshielded with care
AES3 Digital Audio
Usually shielded or unshielded Cat. 5 or 5e
Ethernet Cabling Relevant to Livewire Systems
Name
Description
Max
Length
Where Used
10BASE-T
10Mbps on 2 pairs Cat. 3 copper. Obsolete
for new installations.
100 m.
Not used.
100BASE-TX
100Mbps on 2-pairs Cat 5 copper (Cat. 6
recommended for LW to add a safety margin). Most common Ethernet media.
100 m.
Livewire nodes, PCs.
100BASE-FX
100Mbps on fiber
•
multi-mode
•
single-mode
1000BASE-T
1Gigabit on 4 pairs Cat 5e copper (Cat.
6 recommended for LW to add a safety
margin).
100 m.
1000BASE-SX
1 Gigabit on short wavelength fiber, multimode
220-550m Switch-to-switch
1000BASE-LX
1 Gigabit on long wavelength fiber, singlemode
5 km.
2 km.
20 km.
LW nodes with ext.
media converters
Studio Engine to
switch, PCs, switch-toswitch
Switch-to-switch
CAT-5 for Audio?
Using CAT-5 “digital cable” for audio may seem strange at first, but it does make
sense. The low capacitance and tight twisting requirements necessary for highspeed networks are good for analog and AES audio as well. Because you have a
single cable and connector type for everything in your facility, as your studios
evolve, the same cable that was once used for analog may be used for AES, Livewire
digital audio, general Ethernet data, or whatever else might come along.
Does this work in the real-world? Sure it does – as demonstrated in the many
installations that have been done using the Radio Systems “StudioHub+” product
family. We use the StudioHub 2 format for our connectors by the way, to allow convenient use of their system by Livewire users. More on this below.
Ethernet 100BASE-TX
Livewire uses 100BASE-TX copper wiring with RJ-45 style plugs and jacks for
connections from audio nodes to switches.
You must use Category 5e cable and accessories or better. If you’re interested in
Intro to Livewire 41
Interesting note:
100BASE-T with the
final X being dropped
is oftentimes used
as shorthand for
100BASE-TX. The
100BASE-T designation officially refers
to both copper and
fiber formats at
100Mbps rate, with
TX the specific designation for copper.
The abbreviation in
popular use arises
from the fact that
the copper formats
on either side are
called 10BASE-T and
1000BASE-T. And that
the -T is supposed
to stand for “twisted
pair” – except here for
some reason. Leave it
to standards bodies
to be non-standard.
“future-proofing” your new facilities , you might also consider CAT-6 in order to be
ready for 1000BASE-T. CAT-6 is not much more expensive that 5e and it does perform better, particularly when a run has lots of bends that could disturb the pair relationships within the cable jacket, or has many punch blocks and/or patch cables.
Pin numbering, jacks, cables, and color codes
Modular wall jacks are normally installed so that the pins are at the top of the
cavity. This helps to protect the contacts from water when baseboards are mopped
and from dust. When the jack is oriented in this position, looking into the jack with
the contact pins at the top, the pins are numbered 1 to 8 from left to right. Some
jacks may not have all pin positions populated, but the numbering would still
begin with the first position. For instance, the “RJ-11” style jack is a 6-position 4-pin
jack. Therefore it has pins 2-3-4-5 and pins 1 & 6 are usually absent.
You should take care not to plug an RJ-11 into an RJ-45 jack. It will work to connect the pairs that are supported in the plug, but the plastic part on both sides will
push the outer pins on the jack up, and they may not make good connection when
the jack is again used for an RJ-45 plug.
Ethernet uses 8-position 8-pin modular connectors. TIA/EIA specifies two standards for wiring RJ-45 style cables. The T568A color code is “preferred” by TIA/EIA
but is not so usual in the USA for business installations.
The TIA/EIA T568B color code cable
specification has the same electrical
connections, but has the green and orange pairs swapped (as shown at right).
This is also known as the AT&T 258A
wiring sequence and has been widely
used in the USA. It is used by the Radio
Systems StudioHub+ system for analog
and AES connections, so we recommend it for all new installations.
Either sequence will work just fine if you have it on both ends. In either case,
you have a cable with 4 pairs wired straight through, both ends wired identically.
42 Intro to Livewire
Depending on the cable manufacturer, the color conductor of each pair may or
may not have a white stripe. The other half of the pair is usually white with a colored stripe, but sometimes can be only white. Both formats are shown in table form
below:
TIA/EIA-568-A T568 Wiring Standard (preferred for Livewire)
Pin
Function
Color
Shield
Protective Ground
1
Transmit +
White/Green
2
Transmit -
Green
3
Receive +
White/Orange
4
N/C
Blue
5
N/C
White/Blue
6
Receive -
Orange
7
N/C
White/Brown
8
N/C
Brown
TIA/EIA-568-B T568 Wiring Standard (Optional and acceptable for Livewire)
Pin
Function
Color
Shield
Protective Ground
1
Transmit +
White/Orange
2
Transmit -
Orange
3
Receive +
White/Green
4
N/C
Blue
5
N/C
White/Blue
6
Receive -
Green
7
N/C
White/Brown
8
N/C
Brown
Something to watch out for: The old telephone USOC wiring code has the pairs
in the wrong place, with the wiring in simple one-pair-after-the-other sequence.
You’ll have a split-pair if you have this sequence – and a lot of crosstalk and interference problems. You need to be sure that the pairs correspond to Ethernet’s requirements.
Why does Ethernet have such a strange wiring sequence? Because the center
two pins, 4 & 5, are where telephone voice circuits are wired. The designers of the
standard thought that some people would want to use a single cable for voice and
data, so they kept Ethernet clear of the telephone pins. There is also this: if a user
plugs his PC’s network connection into the phone jack, he doesn’t blast the net
Intro to Livewire 43
work card with ringing voltage.
Even though you have two unused pairs in the standard CAT-5 4-pair cable,
you should not share the cable with any other service since 100BASE-TX was not
designed to withstand additional signals in the cable. The reason for the extra pairs
is that you might want to upgrade to 1000BASE-T or some other yet-to-be-introduced service later.
Finally, on this topic, something really nuts… The overall cabling specifications
standard and document from TIA/EIA was called TIA/EIA-568-A Commercial Building Telecommunications Standard. Within this were the T568A and T568B pinout
standards. Note the dashes and lack of same. Now there is a new TIA/EIA-568-B
overall standard, which has the same two pinout standards within. Couldn’t these
guys have been a bit less confusing?
Crossover 100BASE-T Ethernet Cable
Sometimes you want to connect two LW nodes directly together without a
switch, such as for testing or when you want to make a snake. Or you might want
to connect a node directly to a PC for initial configuration or as a sound card. In this
case, the Transmit of one device must be connected to the Receive of the other.
For this, you’ll need the special crossover cable wired as shown below.
Pin
Color
Pin
1
White/Green
3
2
Green
6
3
White/Orange
1
4
Blue
Not Used
5
White/Blue
Not Used
6
Orange
2
7
White/Brown
Not Used
8
Brown
Not Used
Many modern Ethernet switches have ports that automatically sense the need
for a crossover function and configure their ports appropriately. So when you are
connecting ports from two switches, you probably will not have to use a crossover
cable.
44 Intro to Livewire
1000BASE-T Gigabit Copper
We use 1000BASE-T to connect studio processing engines to switches. If your
Livewire network consists of multiple switches, you will also want to use it to link
switches to each other.
1000BASE-T works with CAT-5e, but again we recommend CAT-6, because
1000BASE-T uses the same RJ-45s as 100BASE-TX, but needs all four pairs. Either
the T568A or T568B wiring sequence will work, but you will have to be sure all four
pairs are wired through and working. Again here, the advantage of choosing one
scheme and using it for everything (e.g. T568B on CAT-6) is obvious. 1000Base-T
Signal Designations are as follows:
Pin
Color
Function
1
White/Orange
BI_DA+
2
Orange
BI_DA-
3
White/Green
BI_DB+
4
Blue
BI_DC+
5
White/Blue
BI_DC-
6
Green
BI_DB-
7
White/Brown
BL_DD+
8
Brown
BL_DD-
There are no separate send and receive pairs for 1000BASE-T. Each pair both
sends and receives with a hybrid at the ends to separate the two signal directions.
Thus, there are effectively four paths each way. The signaling rate for 1000BASE-T is
the same as for 100BASE-T – which is why it can be run over the same cable.
Nevertheless, 1000BASE-T is more sensitive to certain performance issues owing to the hybrids and twice the number of signals in a 4-pair cable. That’s why
CAT-5e or CAT-6 is recommended. And you should always use high-quality factorymade patch cables.
You shouldn’t ever need a 1000BASE-T crossover cable, but who knows? Anyway, a universal crossover cable can be made (or better, purchased) that works for
both 100 and 1000BASE-T; the following table shows the configuration for a Universal 1000Base-T/100Base-T Crossover Cable.
Intro to Livewire 45
Pin
Color
Pin
1
White/Green
3
2
Green
6
3
White/Orange
1
4
Blue
7
5
White/Blue
8
6
Orange
2
7
White/Brown
4
8
Brown
5
100Base-T Crossover Cable
Audio connections
We use the pin-outs established by the Radio Systems StudioHub+ wiring system, which has become a de-facto standard. Since we follow this standard, Studio
Hub wiring components may be used for the analog and AES part of Livewire installations. Radio Systems offers an extensive line of single “dongle” and multi-pair
harness cables pre-wired to connect to a variety of popular studio gear. They also
make balanced-to-unbalanced, AES to S/PDIF, and AES to TOSLINK adapters, headphone amps, etc.
We do stay with traditional XLRs for microphone inputs, however. We don’t
think RJs would be sufficiently reliable for such low signal levels. And we sometimes have parallel XLRs for your convenience when panel space allows us to do it,
such as with the Livewire Router Selector node.
Router Selector Node Rear Panel – Both XLR and RJ-45 connectors are present for inputs and outputs.
While unbalanced connections can be used be very short runs with unshared
and shielded cables, balanced connections are essential for anything over a few
feet in length. The input stage of any attached analog equipment needs to have
good CMRR (Common Mode Rejection Ratio) and high-frequency filtering in order
for balanced connections to effectively cancel crosstalk and interference. With 60dB
CMRR, Axia Livewire node inputs are designed to be no trouble in this respect.
46 Intro to Livewire
The pinouts for the RJ-45 style audio connectors are shown below:
Pin
Function
Color
Shield
Protective Ground
White/Slate & Slate/White*
1
L+
White/Orange
2
L-
Orange
3
R+
White/Green
4
N/C (GND)**
Blue
5
N/C
White/Blue
6
R-
Green
7
N/C (15-)**
White/Brown
8
N/C (15+)**
Brown
* Optional
** Used to power “spoke” devices such as balanced-to-unbalanced converters. LW nodes do not
supply this voltage, but external supplies can be used when needed.
Axia/Radio Systems Standard for Analog and AES wiring on RJ-45
Off-the-shelf or homemade RJ-45 cables, together with adapter
dongles (like these RJ to XLR adapters from Radio Systems) connect
the nodes to audio equipment.
Installing RJ-45s
It would be possible to build a sophisticated multi-studio facility without ever wiring a single RJ-45 plug – you would use modular patch fields or
jacks at each end of the long “horizontal” cable with punch-down 110-style connections. Then factory-made patch cords would be used to get from the switch or
Livewire node to the patch jack. And this might not be a bad idea!
Nevertheless, you’ll probably find yourself installing your own plugs at some
point, so here is some advice:
•
If you are making a patch cord, use stranded-conductor cable. Solid is likely
to break after some time being plugged and unplugged. However solid cable
should be used for backbone wiring as it has less loss.
•
Be sure you are using plugs designed for the cable type you are using. Plugs for
solid and stranded wires are not the same.
•
Plugs from different manufacturers may have slightly different forms. Be sure
your crimp tool correctly fits. In particular, the crimper made by AMP will only
work with AMP plugs. Buy a high-quality crimping tool to help prevent any
Intro to Livewire 47
problems.
•
The outer jacket should be cut back to about 12 mm (.5 inch) of the wire tips.
Check to be sure there are no nicks in the wires’ insulation where you cut the
jacket (an appropriate tool can be purchased to permit you to do so rapidly
without fear of damaging the inner insulation).
•
Slide all of the conductors all the way into the connector so that they come to
a stop at the inside front of the connector shell. Check by looking through the
connector front that all the wires are in correct position.
•
After crimping, check that the cable strain relief block is properly clamping the
outer cable jacket.
•
When checking the cable either with a tester or a real device, wiggle the cable
around near the plug to be sure that connector is working reliably with stress.
You’ll probably need a couple of times to get
it right the first time, but after some experience,
it will start getting pretty easy. Certain RJ connectors include a
small carrier that the wires can be fed into first,
and then slid into the connector itself. These are
recommended as they speed installation and improved accuracy.
48 Intro to Livewire
5: Designing & building your Livewire
Ethernet system
As with analog audio installations, Livewire set-ups range from the very simple to complex facility-wide installations with hundreds of ports. This section is aimed primarily at
those who will be building large systems.
CABLING
Ethernet is balanced and transformer coupled, so has very good resistance to
interference and has no problem with ground loops. However, frequencies ranging
to tens of megahertz are being used, so care must nevertheless be taken.
In the bad old days, wiring was specific to the task – and often to the vendor.
Each telephone, network, and audio had its own cable type and wiring protocols.
The idea at standards bodies like the Telecommunications Industry Association
(TIA) and the Electronic Industries Association (EIA) in the USA is to define classes or
categories of cables and accessories that can be used for all applications specified
for that class. With this, you have a vendor-independent way to wire buildings and
facilities so that services from many vendors can be supported over time without
replacing cabling and connectors. The name for this concept is structured wiring.
The long cables that go from equipment rooms to node locations are called
Charles Spurgeon in
Ethernet: The Definitive Guide says that
you should consider
wiring to be the essential skeleton for
your network installation. He goes on to
point out that network cabling skeletons are often hidden
in the time-honored
place for skeletons: a
closet. Rim shot.
horizontal cables. They usually terminate in RJ-45s, either in patchfields or on wall
jacks. Patchcords with RJ-45s at each end complete the system, connecting the
nodes and central equipment to the jacks.
Twisted-pair Cable Categories
Cable categories are key to the structured wiring concept. The cabling specifications for the various categories are in the TIA/EIA-568-A (and B) Commercial Building Telecommunications Cabling Standard. The following categories are defined:
•
Category 3: These are used only for telephone and Ethernet 10BASE-T, so are
not useful for Livewire installations.
•
Category 5: This designation applies to 100 ohm unshielded twisted pair
cables and associated connecting hardware whose transmission characteristics
Intro to Livewire 49
are specified up to 100MHz. Cat 5 cables are today’s most common because
they support Ethernet 100BASE-TX.
•
Category 5e: This is enhanced Category 5 cable. The main application is for
gigabit 1000BASE-T. While Cat 5 is acceptable for 1000BASE-T, 5e is officially
preferred.
•
Category 6: While not strictly necessary except for 1000BASE-T links, we recommend CAT-6 for all new Livewire installations. CAT-6 provides significantly
higher performance that that of CAT-5e. The main difference is that this cable
has a plastic pair separator inside that holds the wires in correct relation to each
other. This is what makes CAT-6 larger in diameter than CAT-5 cables.
Belden has a Cat 6 cable called Mediatwist that looks
very interesting. This
cable has a half-moon
shape and the pairs
are tightly held in
molded channels.
This product also has
the two wires in each
pair glued together
so that the twist
characteristic is fixed
and stable regardless
of manufacturing
tolerances, cable flexing, etc.
The most significant difference between cables from each category is the
number of twists per foot and the tightness with which the twists and the spacing
of the pairs to each other are controlled. The wire pairs in a voice-grade Category 3
cable usually have two twists per foot, and you may not even notice the twists unless you peel back quite a lot of the outer insulation. Category 5 is tightly twisted,
something like 20 per foot. This results in superior crosstalk performance at higher
frequencies.
Another characteristic of twisted-pair cables is the type of insulation used on
the wires and the cable jacket. “Plenum rated” cables are more stable with changing temperatures due to their using Teflon rather than PVC insulation. Plenum rated
cables are required in air handling spaces in order to meet fire regulations. Teflon
produces less smoke and heat in the case of a fire than PVC and does not support
the spread of flames.
Special Care for Ethernet Audio
“Normal” data over Ethernet is usually TCP/IP protocol. As discussed later, TCP
has a re-transmission mechanism that detects errors and fixes them by requesting
and obtaining replacement packets when one has been received with a defect. This
mechanism is not used for audio – it can’t be when you need low delay and multiple receivers. So it could be possible that a network could apparently be OK with
50 Intro to Livewire
computer data, yet exhibit errors with audio because TCP is covering-up underlying
problems.
A particular concern is to prevent impedance reflections at cable termination
points and to not disturb too much the position of the wires inside the cable. Here
are some specific recommendations:
•
Use the minimum number of terminations and patches that will support your
application.
•
Use patch cables, connectors, and other accessories rated at the same or higher
category level as the cable you are using. Generally, your best bet is to buy premade patch cables to both save money and time as well as assure reliability.
•
Keep a wire pair’s twist intact to as close as possible to any termination point.
For Category 5, this should be to within 1.3 cm (.5 inch).
•
Maintain the required minimum bending radius. For a 4-pair 0.5 cm (.2 inches)
diameter cable, the minimum bend radius is 4 times the diameter, or about 2
cm (.8 inches).
•
Minimize jacket twisting and compression. Install cable ties loosely and use
Velcro fasteners that leave a little space for the cable bundle to move around.
Do not staple the cable to backboards. If you tightly compress the jacket, you
will disturb the twists inside and the relationship of one pair to another, which
could cause crosstalk.
•
Do not overfill conduits.
•
Avoid stretching the cable. The official recommendation is to use less than 25
lbs. pulling pressure.
•
Avoid close proximity to power cables and equipment that generate significant
magnetic fields. The official recommendation is minimum 6.4 cm (2.5 inches)
from power cables when the Cat 5 is either inside a conduit or shielded. Care
should be taken also with fluorescent lighting fixtures, motors and transformers.
•
The pins on RJ-45 plugs are gold plated. But not all connectors are. For maximum reliability, use connectors with 50m gold plating.
Intro to Livewire 51
To Shield or Not to Shield
Unless you are in a high RF environment or you intend to run your network
cables close to audio cables with equipment that has poor balancing on the inputs,
you should be able to use unshielded twisted pair for your Ethernet connections.
If you decide to shield, the usual procedure to attach it only at one end applies in
order to prevent ground loops.
Unbalanced Connections
The Livewire nodes have very good common mode rejection. This, coupled
with the highly twisted CAT-5 or -6 cable works extremely well in the balanced proaudio environment. Unbalanced interconnections are problematic however
and should be avoided for the usual reasons. If you need to interconnect
a Livewire node to unbalanced gear we strongly recommend that you
use a balanced to unbalanced buffer amplifier or transformer located as
close as practical to the unbalanced equipment. There are a number of commercial
off-the shelf options to accomplish this. In particular the Radio Systems StudioHub+ Matchjack series pictured here offer plug and play compatibility between the
RJ-45 balanced and consumer unbalanced worlds.
More than Four Pairs in a Cable
Back in the 10BASE-T days, it was usual to have phone-type 25-pair cables carrying data signals. But the standards for Cat 5 and better call for individual cables for
each connection due to the possibility of multiple disturber near end crosstalk – or
many signals adding up to create combined crosstalk at too high a level.
On the other hand, Belden has some papers on their website proposing that
their finest cable, Mediatwist, would support even 100BASE-T and analog audio
inside a shared sheath. Nevertheless, they offer
the cable in only 4-pair versions at this time.
Patch Panels
Patch panels come in versions for rack or wall
mount and with varying numbers of jacks. Cat
52 Intro to Livewire
5/5e cables are punched down at the rear into 110-style insulation displacement
connectors using a tool very similar to the one that is used with traditional “66
blocks.”
Wall Jacks
Again 110-style IDC connectors terminate the cable. Then these wired-up “Keystone” RJ-45 jacks are pushed into a hole in the wall plate to complete the job. The
diagram to the right shows the simple steps involved in terminating these.
Cat 6 Jacks
Cat 6 cables and their accessories need more care to
maintain the twists as close as possible to the end. At left
is a high-end Cat 6 jack assembly ready for installation into
either a rack-mounted patch-field or a wall jack. This is a shielded version, so the
shell is made from metal to maintain the shield all the way
to the edge of the jack.
Next, you can see the components that make up the
jack disassembled. This is now the non-shielded version, so
the shell is plastic, the blue piece in the photo below.
At the lower left is a closer look at the part that holds the wires. Assembling one
of these can be done in a minute or two. First the wires are
put into the slots and the ends are trimmed. Then this piece
and the front part of the jack are pushed together. The shell
is then placed over these pieces and pushed over them,
which draws the wires into the insulation displacement
forks and locks everything together.
How to wire a Keystone
RJ-45 jack
Architecture Options
There are a lot of ways to build a Livewire network. For many people a simple
one-switch layout will be perfectly sufficient. Others will want to build sophisticated networks to support multiple studios and perhaps hundreds of audio channels.
Fortunately, Ethernet scales easily – and therefore so does your LW installation.
Intro to Livewire 53
Here are some examples and ideas to get you started.
Simple One-Switch Network
Common 1U switches can have as many as 48 ports. That is a lot of audio! Here’s
a setup that supports an on-air studio and a production studio.
The switch is a 24-port 100BASE-TX + 2-port 1000BASE fiber version.
There is the microphone version of the LW node in the on-air studio and the 8x8
line version in the rack. The production studio connects with a Router version node,
which has one send channel and a selectable receive channel.
The Element power supply includes plenty of GPIOs for starting CD players,
lighting on-air lamps, remote mic on-off, etc.
The StudioEngine conDelivery PC
nects with a 1000BASE-T
Mixing Engine
copper link to one of the
Production Studio
two 1000BASE-T SFP ports
with a standard copper
transceiver module.
The delivery PC connects
directly to the audio net-
Server
Example of a single-studio Livewire Routing/Mixing Solution
work with the Axia IP-Audio
Driver software. Control for
it may be directly over the network or could be with a hardware parallel connection. Servers and additional PCs can be connected to the switch for file storage and
delivery systems.
Peripherals such as codecs, telephone systems, and satellite receivers may be
connected into the network wherever it is convenient. In the diagram, the Zephyr
codec and Nx12 phone system are shown attached directly to the switch; many
Telos and Omnia devices, as well as those of other manufacturers, can be attached
directly with Livewire connection ports. Equipment without Livewire can be attached directly to an Audio Node, as is the satellite receiver in the drawing.
You could expand this to two Elements and Engines to support two studios
54 Intro to Livewire
since the switch has two 1000BASE-T
ports. Or you could substitute an additional 1000BASE-T switch to support
as many studios as you want.
In this photo, you see a typical setup with a node, engine, switch, and
patchbay. The patchbay is being used
to terminate cables from remote locations before being connected to the
switch with short patch jumpers, while the node and engine connect directly using
longer patch cords. Using a patchbay and off-the-shelf patch cords in this fashion
minimize the need to install RJ-45 plugs.
Daisy Chained Multiple-Switch Network
While one switch can support multiple studios, you would have a single point
of failure. Here’s another approach that gives each studio its own switch. The example on the next page uses three switches, one for each studio group. This layout
style could easily be expanded to any number of switches and studios.
The switches are connected together so that audio sources are shared. A
1000BASE-T link between the switches allows hundreds of audio channels to flow
from one group to another. With
more than two switches you could
have a “circular backbone” with
redundant Spanning Tree links
(described below) between the
switches.
Peripherals that are used in
common such as codecs could
be plugged to any of the studio
switches, or there could be a separate switch to pick up such feeds.
Daisy-chained Ethernet Switches Support Multiple Studios
Intro to Livewire 55
Hierarchical Multiple-Switch
­Network
This is a layout that could support a very large facility. A gigabit
switch is at the center and 100/1000
switches are used at the edge with
one for each studio or logical group.
A Router Selector node is kept in
the central equipment room for test
and monitoring. Additional nodes
could link audio from non-Livewire
studios.
While we could plug the Engines
A Two-level Hierarchical Network for Support of Larger Studio Facilities.
into the central switch, if we keep
them coupled to the individual studio switches, there is no single point of failure
for any studio.
Gigabit links are used between the edge switches and the center. These could
be copper or fiber with a suitable switch.
The physical location of the switches is a matter of taste and trade-off. Putting
the edge switches near the studios saves cable runs, but locating all the gear in a
central room simplifies engineering activities.
As this is written, an appropriate switch for the center costs $2k and the studio
switches are under $1k. So this is a quite reasonable-cost option that provides a lot
of power, flexibility, and expandability. Dozens of studios and thousands of audio
channels are possible.
Options for Redundancy
Ethernet switching has a built-in scheme for redundancy, called Spanning Tree
and standardized as 802.1D. A newer variant is called Fast Spanning Tree. Switches
with spanning-tree enabled exchange information with each other about the
topology of the overall network. You can have redundant backup links that are
56 Intro to Livewire
automatically activated in the case that a main link has failed. Depending on the
switch and layout, it could take as little as a second or as much as a half-minute for
a redundant link to be connected.
Link aggregation (sometimes called port trunking) is another method. With
Spanning Tree, even if you have two links between two switches, only one of them
at a time will be active. But, it’s often better to have both active simultaneously
because you get twice the bandwidth during normal operation and instantaneous
backup should one fail. The link aggregation standard is 802.3ad. To use it, you usually have to specifically enable it on your switch. Incidentally, this is supported on
some PC network interface cards intended for servers, so it’s not only for switch-toswitch links.
Most Ethernet switches offer a redundant power supply option.
We’ve been talking here about automatic on-line redundancy, but there is also
manual swap-out as a reasonable option. Because RJ-45s are so easy to unplug and
re-plug and because switches and other Livewire components are much cheaper
than traditional alternatives, you can have spare units on the shelf for fast substitution.
Fiber
Fiber optic links can extend the range of Ethernet. Because they are not subject
to crosstalk and magnetic interference, they also can solve problems that might
crop up in difficult locations with copper cables.
External media converters can be very simply plugged to Livewire nodes and
switch 100BASE-T ports to convert copper connections to fiber.
This unit from Allied-Telesyn uses 100Mbps ST multimode fiber for
up to 2km range. Units supporting SC single mode fiber can extend up
to 75km.
Modern Ethernet switches often have the option to plug a media
converter directly into a special socket so that fiber may easily be connected from
switch to switch. This is useful to make high capacity backbone links without any
external boxes.
Intro to Livewire 57
Here is a typical case. The switch shown at left has four “uplink”
ports for use with 1000BASE-T SFP (Small Form-factor Pluggable) copper transceiver modules. The device below it is a typical modern media
adapter in the “SFP/mini-GBIC” size – about the same
in width and height as an RJ-45 jack. They come in
different flavors, for 1000BASE-SX, 1000BASE-LX, etc.
Generally, SX cables have a range to 500 meters, LX
to 5km, and LH to 70km.
Radio Links
You probably expect
something with
“multi” in the name to
have more capability
than the same thing
designated “single”.
But this is not the
case with fiber optics:
single-mode cables
are better and more
expensive than multimode. These names
refer to how light is
contained within the
fiber. Single-mode
strands are smaller
and more carefully
control the light so
that it doesn’t bounce
around so much
inside, thus are more
efficient and permit
longer ranges.
There are Ethernet IP radios with surprisingly high bandwidth – and at surprisingly low cost. Not all units are capable of achieving true Ethernet performance in
terms of error rates, so some caution is in order. Most of these operate in the unlicensed ISM bands, but with modern spread-spectrum technology and elevated
directional antennas, interference doesn’t look to present much problem. Licensed
radios following the new IEEE 802.16 “Wimax” standard are starting to appear.
Bitrates range to 150 Mbps and distance to 25 miles depending on power level,
antenna, and terrain.
For studio-to-transmitter link, remote pick-up, and studio-to-studio applications, these offer multiple channels of uncompressed audio, two-way transmission, and the ability to multiplex VoIP telephone, remote control, and general data.
When audio and general data are mixed, the Ethernet switch provides the prioritization function. As with all Livewire elements, you can check them with a web
browser on a network-attached PC.
We have been studying many of these Ethernet IP radios, testing for Livewire
compatibility and general performance. You should consider these like the Ethernet switch – please let us advise you on the best
choice and help with your application; there are
several manufacturers and models listed at
www.­AxiaAudio.com/stl/. Of course, you’re always
welcome to contact us for our latest advice as well.
58 Intro to Livewire
Designing For Security
You will have 100% security if you keep the Livewire system completely isolated
from any other network, local or wide area. Those very concerned with protecting
the studio system may well want to take this approach.
But there are advantages to sharing with or linking to an office network. You
can configure and monitor the system from any connected PC and audio can be
monitored on any desktop with access. In this case, separate switches or VLANs (described later) can be used to provide isolation. An IP router passes only the correct
packets from one to the other and thus provides a firewall function.
The next step up in connectivity would be to have a network linking co-owned
or otherwise affiliated stations. In this case, a network engineer is probably in the
picture and he can take the necessary steps to protect your audio.
Connection to the public Internet brings the advantage that you can monitor and configure from a remote site, but you now have much risk from unwanted
intruders, viruses, etc. A qualified network engineer
should be consulted to be sure you have an appropriate firewall and other protections in place.
In Livewire nodes, web and Telnet access are password protected to
provide some measure of security.
But we do not use exotic techniques like SSL (Secure Sockets
Layer), so please understand that
our devices were not designed to
be exposed to the public internet
without external protection.
Intro to Livewire 59
60 Intro to Livewire
6: The Ethernet switch
This is what makes it all possible. Here are some details on requirements for a capable
Livewire switch.
Livewire packets include both the Ethernet and IP headers. This means that
Livewire streams may be either “switched” at layer 2 or “routed” at layer 3. For most
installations, we recommend a “managed Layer 2 switch” or a “Layer 3 switch” that
includes the required IGMP Querier and snooping functions. IP Routers are able to
do layer 2 switching as a subset of their more advanced capabilities, so may also be
used. Switches and routers range from very simple to amazingly elaborate.
Livewire Ethernet Switch Requirements
•
Sufficient backplane bandwidth. It is required to be fully “non-blocking” to
handle all ports at full capacity.
•
Sufficient frame forwarding rate. LW Livestreams have small packets at a fast
rate. The switch needs to handle this.
•
Correct handling of IEEE 802.1p/Q frame prioritization. Livewire audio frames
must be given priority without too much delay or jitter. The IEEE standard specifies 8 levels of priority, but few switches support all the levels. Many support
only 2 or 4, lumping some of the incoming levels together. We recommend 4 as
the minimum for a Livewire system.
•
Support for multicast, with sufficient filter entry capacity to cover the total
number of audio streams you need. This latter is important, because when the
filter capacity is exhausted, switches forward multicast packets to all ports, subscribed or not. This would cause serious problems. You will probably want 256
minimum.
•
IGMP control for multicast. Traffic must be under IGMP control – strictly no
flooding of ports with multicasts under any circumstances.
•
Support for both port-based and tagged-frame-based VLAN. This latter is the
IEEE 802.1Q standard and is what allows the switch to determine priority on a
frame-by-frame basis. Port-based VLAN can also be useful: it lets you “hardwire”
a particular port for a single VLAN, useful to be 100% sure an office PC can’t get
Intro to Livewire 61
onto the LW audio VLAN.
•
If you will use a separate VLAN for Livewire,
the switch needs to have an “IGMP querier”
on each one, which also means that you can
assign an individual IP number to each VLAN.
This is a rare capability and its absence disqualifies many switches.
•
Management. This is how you get remote
Cisco Catalyst 3750 Family
monitoring.
The practical bottom line is that you should
use a switch that has been selected and tested
by Axia. When we check a switch, we use a
laboratory setup that lets us send frames on a
number of ports at a high rate, while switching
channels on/off with IGMP, etc. We have a lot of
Cisco Catalyst 3560 Family
experience with different switches and know
what to look for. Using a recommended switch
will also help you when you need customer support because we will be familiar with it. You are
encouraged to contact Axia Customer Support
or visit www.AxiaAudio.com/switches/ for our
full list of supported switches..
Cisco Catalyst 2960 Family
Some Switches We Like
There are new switches introduced everyday
it seems, with ever increasing performance and
falling prices. Please check our website for our
latest recommendations.
Having said that, the Cisco Catalyst 6500,
62 Intro to Livewire
Cisco Catalyst 6500 Family
4900, 3750, 3560, and 2960 series switches are some of the units that are fully
compatible and qualified for use with Axia. There are literally hundreds of models
that can be used with your Axia networked audio system. 24 10/100BASE-T ports,
48 10/100BASE-T ports, with or without 1000Mbps SFP (copper/fiber) transceiver
ports — connect them together and have virtually unlimited flexibility and redundancy. All of these units include built-in simple router and IGMP Queriers on
every VLAN. They also come in a powered-port version that can be used with VoIP
phones.
Switch Configuration
Most switches offer three connection options: an RS-232 console port, Telnet
over Ethernet, and web over Ethernet. For Axia-supported switches, we offer a
configuration “cheat-sheet” that gives you the basics. We also will be happy to preconfigure your switch and test it at Axia before shipping it to you. There are two
different software images that Cisco switches can be configured with. They are
available with either the standard multilayer software image (SMI) which includes
advanced QoS, rate-limiting, ACLs, and basic or the enhanced multilayer software
image (EMI) which includes a richer set of enterprise-class features and advanced
hardware-based IP unicast and IP Multicast routing as well as policy-based routing
(PBR). We do not expect or require anyone to be Cisco Certified Network Engineers
so we provide you everything you need to know about configuring your network
switch based on the software image of your unit. Configurations for both the SMI
and EMI versions of Cisco software can be found on our website.
Intro to Livewire 63
7: Testing, 1-2-3…
There are tens of thousands of people installing Ethernet networks every day, and many
millions of working installations. So there are a lot of tools to help you. Livewire equipment have a lot of diagnostic functions built-in as well.
General Ethernet Troubleshooting
Ethernet is a mature technology, with years of proven reliable service. You are
not very likely to see problems in the fundamental technology if you follow the
network wiring and layout recommendations.
Prevention
The best way to avoid downtime is to build the network well in the first place.
Use high-grade cables, good quality factory-made patch cords, etc. And be careful
with the punchdown and plug installation.
More on the topic of patch cords. If you really must make your own, they should
be built with stranded wire cables. Solid conductors are likely to crack when flexed
a lot, usually right at the RJ-45 plug. From this you can get intermittents and bit errors. Also, as mentioned in the cabling section, be sure you have the right plug for
the cable you are using. An RJ-45 plug designed for stranded wire will cut through
a solid conductor.
But you know all that. So let’s get on to troubleshooting, when despite all due
care something still goes wrong.
The Basics
Link Test
A layer 2 test, this checks the connection between the switch and a designated
network device on the same LAN. During the link test, IEEE 802.2 test packets are
sent to the designated network device in the same VLAN or broadcast domain. The
remote device must be able to respond with an 802.2 Test Response Packet. Most
switches support this test via a web or command line interface.
Ping
A layer 3 test, a simple and effective way to check basic “reachability” of an
64 Intro to Livewire
IP-enabled device. Ping sends a test packet to a device and waits for an echo response. A Windows PC can do this within the command prompt window. Just enter
ping x, where x is the IP number or the domain name (if a DNS server is available)
and see the result. If you get the echo, the basic connectivity (including Layers 1, 2,
and 3) is OK. Most switches and almost all IP-enabled devices support this test.
Switch Diagnostics
Ethernet switches have many diagnostic tools, ranging from front panel LEDs to
sophisticated software monitoring functions. See the switch manual and software
description for details for your unit.
Simple Network Management Protocol (SNMP) and Remote MONitoring (RMON)
are part of the TCP/IP internet suite. (RMON is built on SNMP so they are closely
related.) They offer a way to probe and monitor network equipment operation in
a vendor-independent way. For example, an Ethernet port has a standard way of
communicating its status that is supposed to be used by all products with these
ports.
Almost all sophisticated Ethernet switches offer these, and they are useful tools
to monitor traffic, check operation, etc. You can do a lot of this with web and Telnet
based communication but SNMP usually offers a deeper look.
You will encounter the acronym MIB, for Management Information Base. This is
how information is organized within SNMP.
To use SNMP and RMON, you will need a software application that presents the
information. A popular tool is H-P’s OpenView, for example. HP ships a simple version called TopTools with many of its switches.
A full discussion would be too much for this document, but there is a lot of info
that comes with Ethernet switches, and a lot more in bookstores and on the web.
Some Things to Check
•
Switch configuration must be correct. IGMP must be switched on, VLAN parameters set if you are using them, etc. In our experience to date, this is the most
common cause of problems. (With the exception of cables, of course.)
•
Ethernet links can be 10, 100, or 1000 Mbps, and full or half-duplex. We always
want the maximum rate and full duplex. You can configure the Ethernet ports
Intro to Livewire 65
on some devices for specific modes – but you should not do this. The Auto mode is the correct setting, which will cause the device and node to automatically negotiate to the appropriate condition.
If you manually set the mode to full-duplex, the switch – in compliance with a flawed IEEE standard
– will set itself to half-duplex (!), leading to many problems. Axia Livewire h/w nodes are always set to
the auto mode, so this problem will not arise with them, but with other equipment such as PCs.
•
If you want and have multiple redundant links using port trunking or spanning-tree, you have to set
up the switch to support them. Taking the default will usually not work.
•
The “activity” LEDs (usually amber or green) on many network cards and switches will be on continuously when any LW audio streams are present on the link. That is because the logic that drives the
LED extends the on time so that you can see it with normal traffic. LW packets are traversing the
network at such a fast rate that the LED never has a chance to turn off.
•
Mode of the Axia Livewire hardware nodes have status LEDS. The provide useful information and
should be checked. This is covered later in this section.
Cable Testers
“It’s the cable – it’s always the **@@ cable!” said my first boss. About half the time, he was right. That
percentage is probably a bit higher in Ethernet systems. Indeed, a number of surveys have put the “network medium” to blame 70-80% of the time. This being the cables, connectors, and hardware components that make up the signal-carrying portion of the installation.
Wiggling and unplug-plug operations are legitimate and effective troubleshooting methods. But
there are plenty of cable testers to help you perform more elaborate checks. These range from simple
conductivity testers to sophisticated units that test cables for adherence to the TIA/EIA standards, detect
breaks with a Time Domain Reflectometer, and more. Contact info for the main manufacturers of these
are listed in the Resources section.
Four Cable Testers
The testers shown here represent something of the range available.
The Agilent Framescope 350 on the next page at upper left, and the unit from the Fluke DSP-4000
family at lower left can certify that your cable meets the appropriate category requirements with regard
to crosstalk, attenuation, etc. and perform a number of sophisticated tests. The adapter at the top of the
Fluke can be changed to allow the unit to work with both copper and fiber cable type.
The third unit, at upper right, is a much simpler and cheaper variant from Fluke that checks for con66 Intro to Livewire
ductivity and correct wiring. It can also tell you the distance to a break with a TDM
function and can do tone trace with an optional remote unit. Finally, the ByteBrothers 2-piece set is a basic wiring tester and tone line-finder.
Sniffers
These are s/w applications that run on PCs and can listen-in on the packets flowing on an Ethernet link. Usually used in conjunction with an Ethernet
switch’s port-mirroring function. This lets a designated monitoring port to mirror that traffic on any other port you select. Livewire audio packets are small in
length and very frequent compared to general data traffic so are quite challenging for a sniffer. To be useful, you will need a good one and a
fast computer to run it. Very useful, but expensive. Perhaps
best borrowed from your company’s network guys’ kit.
Diagnosing Problems using Livewire
Components
All Livewire components have built-in diagnostic tools. For
example, audio nodes have a loop-back testing procedure that
measures audio noise and distortion. The web interface lets
you check a number of internal values.
The Router Selector node is a useful device for displaying available audio streams and listening to them. It has one
channel of send, so is useful as an audio source injector as well.
The LW IP-Audio Driver has a diagnostic window that tells you a number of things about the system clock and audio streams.
Hardware Node Indicator LEDs
Four LEDs indicate the status of the Livewire™ and Ethernet connections, as well as system synchronisation as follows:
LINK: When illuminated continuously, this LED represents the presence of a live Ethernet link to another Ethernet 100 Base-T device. This
Intro to Livewire 67
LED indicates that a connection is present and some device is connected. It does
not indicate the quality of the connection however. If no Ethernet link is present,
this will flash slowly.
LIVEWIRE: This LED indicates that the connected Ethernet segment has
Livewire™ traffic present. If the link LED is illuminated, and the LIVEWIRE LED fails to
illuminate, there are either no other Livewire™ devices connected, or the Ethernet
switch has not been programmed to pass such traffic through to the port to which
this node is connected.
SYNC & MASTER: Only one of these two LEDs should be illuminated. If neither
LED illuminates, something is not correct. The SYNC LED indicates the receipt of
clock information from another (Master) Livewire Node. The MASTER LED indicates
that this node is acting as the master clock source for the Livewire network. More
specifically:
•
SYNC – If Sync packets are being received by the Livewire™ node, this LED
will begin to flash. The LED will continue to flash until the Livewire™ node has
locked its local clock to the network master. Once the local node’s PLL is locked,
the LED will illuminate solidly.
•
MASTER – The Livewire™ system employs a sophisticated master/slave clocking system over the Ethernet network. By default any device may become the
clock master, however this can be changed if desired. The system has the ability
to automatically change to a different clock master should the current master
become disconnected, or otherwise inoperable. This happens transparently
without any audio glitches. This LED indicates that this node is currently acting
as MASTER.
68 Intro to Livewire
8: Network engineering for audio engineers
You don’t need to know most of what’s in this section to use Livewire. Just as a beginner
can plug analog XLRs successfully together without knowing anything about op-amps,
you can connect and use LW without knowing details about packets. But just as fixing tricky problems in the analog world calls for higher-level understanding, so does an
awareness of Livewire’s internal technology help you to solve problems and build complex
systems.
This section introduces basic concepts – enough for you to get a feel for how data networks work and to understand the lingo so you are ready to ask intelligent questions of
network guys and vendors. It also explains a lot of Livewire-specific points.
Livewire is built upon standard components, so if you understand data networking generally, you’ll be ready for the specifics of Livewire audio networking. Network
engineering is a rich topic, abounding with information and nuance, and in constant
flux. Fortunately, Livewire uses only a small subset that is easy to learn and understand. That is mainly because most of the complexity comes with IP routing and widearea networks such as the internet – and we don’t use much of that, staying only with
the much simpler Ethernet LAN level. Even if you don’t know anything yet, you’ll get
pretty much what you need in the next few pages. If you want to know more, bookstores have shelves loaded with networking advice and information. We offer a few
starting points in the Resources section.
As always, Axia support is at your side to help with any specific practical issues
that may come your way.
If you are developing for Livewire, this will offer only a brief introduction, and
you’ll want to know more. Please contact us for any of your needs, such as software
API documents.
Ethernet / IP Networks
Layering Model
You need to know layers to know networks. The
notion of layers and the open systems they support are central to network engineering. Because
Intro to Livewire 69
layering is a key to enabling multiple vendors for each function, this design has also
been a major factor in the growth and operation of the internet. It’s also one of the
keys to Livewire, allowing us to build our professional audio transport application
on existing standard lower layers.
For many years, the Open Systems International
Layer
Function
7
Application
6
Presentation
5
Session
working. For example, the ISDN D-channel communi-
4
Transport
cation between nodes and the telephone network is
3
Network
2
Data Link
1
Physical
(OSI) model was the reference paradigm for data net-
loosely based on this model, shown at left.
But this proved to be too complex for most practi-
cal applications, and an architecture has evolved that is simpler than the OSI model.
The chart at right shows how
Layer
Function
Example
that simpler model applies to
5
Application
the IP-over-Ethernet combina-
HTTP (Web), POP
(mail), etc.
4
Transport
UDP/TCP/RTP
tion we are using.
3
IP Routing
IP
2
Switching
Ethernet Addressing
1
Interface
Ethernet Physical
Layer 1, Physical Interface:
This layer is responsible for
hardware connectivity, which is provided by Ethernet.
Layer 2, Ethernet and Switching: This layer is Ethernet’s end station addressing and everything related to it. An Ethernet switch is working at Layer 2 because it
forwards packets based on Ethernet Media Access Control (MAC) addresses which
are unique ID numbers assigned by the Ethernet-capable equipment manufacturer.
Layer 2 does not ordinarily extend beyond the corporate boundary. To connect
to the internet requires a router. In other words, scaling a Layer 2 network means
adding Layer 3 capabilities.
Officially, the transmission units comprising header and data are called frames
at this layer. At Layer 3, the correct designation is packets. But, since Ethernet
frames are almost always carrying IP packets, the word used to describe the combination most often depends upon the context or the author’s preference. Unless we
are referring to layer 2 functions, we usually use “packets” because Livewire audio
has the IP header – and because “packets” has become the usual way to describe
70 Intro to Livewire
this sort of thing generally.
Layer 3, IP Routing: In addition to Ethernet addresses, each IP packet on a LAN
also contains source and destination IP addresses. These were intended to be used
by routers to forward packets along the most efficient route and link LANs of different types. When the internet was invented, there were dozens of LAN technologies in use and this was an important capability. Now, IP addressing is used both
within LANs as a way to access servers from clients, etc, and to connect to internet
resources offsite.
IP in itself is not a particularly complex protocol, but there are numerous capabilities supplied by the other components of the IP suite. The Domain Name System
(DNS) removes the burden of remembering IP addresses by associating them with
real names. The Dynamic Host Configuration Protocol (DHCP) eases the administration of IP. Routing protocols such as Open Shortest Path First (OSPF), Routing
Information Protocol (RIP), and Border Gateway Protocol (BGP) provide information
for Layer 3 devices to direct data traffic to the intended destination.
Layer 4, Transport: This layer is the communication path between user applications and the network infrastructure and defines the method of communicating.
Transmission Control Protocol (TCP) and User Datagram Protocol (UDP) are wellknown examples of elements at the transport layer. TCP is a “connection-oriented”
protocol, requiring the establishment of parameters for transmission prior to the
exchange of data and providing error recovery and rate control services. UDP
leaves these functions to the application.
Layer 5, Application: Web browsers, audio editors, and email clients, for example. And our Livewire audio.
Applications developers decide on the type of Layer 4 transport necessary. For
example, database or Web access require error-free access and use TCP, while live
streaming media use Real-Time Protocol layered on top of UDP/IP.
Making Packets
Livewire Standard Streams use all of the recommended internet protocols and
are constructed in the usual layered fashion. Here is one representation of the
Intro to Livewire 71
packet structure:
In the next graphic, below, you can see this structure in more detail. This is the
way network engineers usually visualize a packet. It’s not important to know what
each of the fields means; the idea is for you to see how a packet is constructed
generally. Each of the horizontal bars are 4 bytes. At each layer, devices are operating only with the information contained within the associated header. An Ethernet
switch only cares about the layer 2 headers and everything else is just payload.
An IP router only “sees” the layer 3 header and doesn’t care about the lower-level
transport. Applications don’t care about headers at all – they just deliver their data
to the network and expect to get it back at the other end (There are, however,
exceptions, such as sophisticated
Ethernet switches that can inspect
layer 3 headers for some advanced
functions).
IP and Ethernet
­A ddresses
As with everything connected to
IP/Ethernet networks, Livewire devices require both IP addresses and
Ethernet MAC (Media Access Control)
addresses.
72 Intro to Livewire
IP Address
IP addresses are four bytes long and are written in “dotted decimal” form, with
each byte represented decimally and separated by a period. For example, in the IP
address 193.32.216.9, the 193 is the value for the first byte, 32 for the second, etc.
Since a byte can hold values from 0 to 255, this is the range for each decimal value.
Host IP addresses are assigned to your organization by your internet service provider and parceled out to individual host computers by your network administrator.
He may give you this number to be entered manually, or could opt for DHCP (Dynamic Host Configuration Protocol) to let your computer get the address automatically from a pool. Because Livewire devices are permanently attached and because
it is more convenient to know the IP address attached to a particular node and
perhaps assign them in some kind of logical pattern, we do not support DHCP for
our hardware nodes. Therefore, you will need to enter an IP address into each node.
In addition to the address, there are a few more numbers to enter into an IP
configuration:
Subnet mask
Subnets allow a network to be split into different parts internally but still act like
a single network to the outside world. There are three logical parts to any internet
address: the main network address, the subnet address, and the particular device
address. The mask marks the dividing point in the address between the subnet part
and the device (host) part. What is meant here by “network” and “subnet” depends
on your internet provider. A network in this context could mean all of the address
space allocated to the provider, and the subnets could delineate the individual customers. Or the network could be all the addresses allocated to a university or major
corporation and subnets could divide the address space to correspond to departments. Network addresses are assigned by IANA, the internet names and numbers
authority, while subnets may be changed without any official approval.
32-bit (4-byte) IP address space
Intro to Livewire 73
The mask is written in the same dotted-decimal form as IP addresses. In the
example a very large network supporting 64k hosts is divided into 64 subnets, each
with 1k hosts. The subnet mask would be 255.255.252.0, which is just another way
of writing the binary ones and zeros value shown above.
As a practical matter, you usually just take the number given to you by you network administrator or service provider and enter it.
Gateway address: This is the IP address of the device that passes traffic out of
your local network to the internet. This is usually a router.
DNS server address: This is the address of the computer that provides name
look-up service, translating text domain names like www.axiaaudio.com to IP address numbers.
In careful language, devices that attach to the internet and have IP addresses
are called hosts, a name that probably made sense in the early days. (They “host”
the IP stack and interface.) And Ethernet-connected devices are officially called stations to keep the radio/ether analogy going. But what do you call something that is
both host and station, as almost everything is “Host” doesn’t sound very natural for
our audio devices and “station” would be very confusing, indeed. As you’ve noticed,
we usually just say Livewire node in the context of our audio equipment, which
should be clear enough. They won’t be nodes, will they? Unless something better
comes along, we’ll probably say Livewire device. As to “host” and “station” for other
devices, we’ll just use connected PC or some variant, thank-you very much.
Ethernet Addresses and Address Resolution Protocol (ARP): Machines that
use IP and are connected to an Ethernet have two addresses, IP and Ethernet MAC.
While the IP address is user-determined, the Ethernet MAC address is usually programmed into the network card or interface by the manufacturer.
You will probably never have to deal with them directly, but who knows? Ethernet addresses are 6 bytes long and are written in “dashed hexadecimal” form like
this: 5C-66-AB-90-75-B1. (Sometimes colons are used as the separators.) Hex notation is just another way to write binary values. Single digits range from 0 to 9, A, B,
C, D, E, F and byte values from 00 to FF. The value FF means all the bits in a byte are
1s and is equivalent to decimal 255. While this notation may seem strange at first
74 Intro to Livewire
sight, it is very useful to programmers, who need to think in powers of two.
There is a unique Ethernet MAC address for each and every network adapter
ever made in the world. IEEE handles the allocation among manufacturers and
each manufacturer is responsible to ensure that they make no two alike within
their assigned range.
There is a need to translate between IP and Ethernet addresses. Consider a
server sending data to a machine it knows only by IP address. To communicate,
it has to generate an Ethernet frame including the Ethernet destination address
corresponding to the desired IP address. To do this, every IP-based device has an
ARP module, which takes an IP address as input and delivers the corresponding
Ethernet address as output. It maintains a local table with the associations. When
it encounters one it doesn’t yet know, it broadcasts an ARP query packet to every
device on the LAN and the device that owns the specified IP address responds with
its Ethernet address. If there is no owner, the packet is presumably intended for an
off-site device and is sent to the gateway address of a router. How does the transmitting device find the router’s Ethernet address?
With ARP, of course. Entering arp -a into Windows’ command prompt will give
you the current list of IP addresses and associated Ethernet addresses – the ARP
table for that machine.
I (Steve) used to feel
bad about all those
wasted addresses
from obsolete and
thrown-away network cards – guess
that’s the Protestant
USA mid-westerner in
me – but supposedly
6 bytes is enough
that each of Earth’s
grains of sand could
have its own address,
so not to worry.
Multicast Addresses: All of the above discussion was only relevant to the usual
unicast situation that is used for web surfing, emails, file transfers, etc. We also use it
in LW for configuration and control, such as when a web browser is connected to a
hardware node. But audio is multicast because we want it to be available to multiple destinations. The principle is simple: rather than specifying a specific destination, a special “virtual” multicast address is used that is not assigned to any particular device. Audio nodes can listen-in in a party-line fashion by receiving any packets
at this address.
Our audio streams are multicast at both Layer 2 and Layer 3, using the set-aside
multicast addresses at each layer. The Livewire channel number is automatically
translated to the appropriate addresses at both layers internally.
Livewire uses the IP address range starting from 239.128.0.0. This choice is
Intro to Livewire 75
based on the assigned numbers from the IANA (Internet Assigned Numbers Authority) allocation of this range for use within organizational and site specific
scopes. These addresses are to be used for multicast applications that are not used
across the global Internet. Since our application will be used within a single facility
on a single switched LAN, this range is appropriate.
Over 8 million unique IP multicast addresses are available with each address
mapping to a globally unique Ethernet multicast address.
Even so, IP is relatively stingy with its multicast space. Ethernet has set aside half
of all destination addresses for multicast - 140,737,488,355,328 addresses, which
should be enough for even the very largest broadcast facility! The designers clearly
had big plans for multicast that have not yet been realized.
The distinction is made in the first transmitted bit of the 48-bit address that
divides the total available address space in two: a 1 in this position signifies a multicast.
Ethernet Switching
Ethernet switching has caused a revolution in data networking. With switching,
each device owns all the bandwidth on its link. No sharing and no collisions. Incoming frames are forwarded only to the nodes that need them.
Despite their amazing power, the invention of switching was more akin to
falling off a log than sawing one in two… The switch builds up a table of what
addresses are attached to what ports, which it does by merely by examining the
source addresses of sent packets. When frames come in, the switch looks into the
table, discovers what port owns the destination and forwards the data only to that
port. In the rare case that no entry exists for an address, the frames are “flooded” or
broadcast to all ports to be sure the intended recipient gets it. If a connection is unplugged or there is no data for a long time, the entry is removed. Pretty simple, eh?
Multicast
The operation described above is for the common unicast, or point-to-point,
76 Intro to Livewire
communication that you have for typical traffic such as web, email, etc. But Ethernet supports three communication types:
•
Unicast means point-to-point. The usual mode for traffic.
•
Multicast means that multiple receivers may “tune in” to the transmission from
a source so that a selected subset of nodes is served.
•
Broadcast means that packets are sent to all receivers, which is quite common
on Ethernets. Microsoft file sharing, for example, advertises the PCs on a network this way. ARP uses this to get a query to all machines on the network.
We use multicast for Livewire audio streams because we want to emulate distri-
bution amps and audio routers, with multiple receivers being simultaneously able
to listen in to a source. The automatic procedure described above does not work for
multicasts because they are not associated with a particular output port and node.
Fortunately, switches offer a way to control these one-to-many streams. A multicast Ethernet frame has a “virtual” destination address that is just stopped inside
the switch if there are no interested receivers. When receivers want to tune-in, they
send a message to the switch telling it to turn on the stream to their port.
The switch knows what frames are multicasts because the destination address
belongs to the set-aside multicast pool.
Livewire uses one Ethernet/IP multicast address for each audio stream. These
are derived automatically from the LW channel numbers you assign. Streams are
multicast at both Ethernet and IP layers using the assigned multicast addresses at
each.
IGMP (Internet Group Management Protocol): We need some way to tell the
switch what streams go to what ports – that is, a way to control multicast switching.
IGMP was designed for just this purpose.
IGMP is part of the IP suite and is a Layer 3 function that was designed to communicate with IP routers to control multicasts. But switch manufacturers started
to implement “IGMP snooping” on the messages between hosts (computers) and
routers as a way to control multicasts at Layer 2. In recent switch implementations
of IGMP, this is taken further and a router is not necessary as long as a switch is con
Intro to Livewire 77
figured to support IGMP with the “Querier” feature enabled. We want this because
there is often no router in the system. Even were there to be one, better to have this
capability in the switch as a back-up.
IGMP uses three types of messages to communicate:
•
Query: A message sent from the querier (multicast router or switch) asking for a
response from each host belonging to the multicast group. If a multicast router
supporting IGMP is not present, then the switch must assume this function in
order to elicit group membership information from the hosts on the network.
•
Report (Join): A message sent by a host to the querier to indicate that the host
wants to be or is a member of a given group indicated in the report message.
•
Leave Group: A message sent by a host to the querier to indicate that the host
has ceased to be a member of a specific multicast group.
An IP multicast packet includes the multicast group (address) to which the
packet belongs. When an IGMP client connected to a switch port needs to receive
multicast traffic from a specific group, it joins the group by sending an IGMP report
(join request) to the network. (The multicast group specified in the join request
is determined by the requesting application running on the IGMP client.) When a
networking device with IGMP enabled receives the join request for a specific group,
it forwards any IP multicast traffic it receives for that group through the port on
which the join request was received. When the client is ready to leave the multicast
group, it sends a Leave Group message to the network and ceases to be a group
member. When the leave request is detected, the appropriate IGMP device will
cease transmitting traffic for the designated multicast group through the port on
which the leave request was received (as long as there are no other current members of that group on the affected port).
Thus, IGMP identifies members of a multicast group and allows IGMP-configured hosts (and routers) to join or leave multicast groups.
The function of the IGMP Querier is to poll other IGMP-enabled devices to elicit
group membership information. The switch performs this function if there is no
other device, such as a multicast router, to act as Querier. The switch automatically
78 Intro to Livewire
ceases Querier operation if it detects another Querier. A switch with IGMP querier
capability will become a Querier in the absence of any other Querier on the network. If you disable the Querier function on a switch, ensure that there is an IGMP
Querier (and, preferably, a backup Querier) available. If the switch becomes the
Querier, then subsequently detects queries transmitted from another device on the
same VLAN, the switch ceases to operate as the Querier for that VLAN. In the above
scenario, if the other device ceases to operate as a Querier, then the switch detects
this change and can become the Querier as long as it is not pre-empted by some
other IGMP Querier.
In a Livewire system, it is the responsibility of the audio nodes to generate the
IGMP messages.
Prioritization
Within a link, we sometimes want to have audio mixed with general data. This
happens, for example, when a delivery PC is playing audio and downloading a file
at the same time, or when our Studio Engine is sending and receiving audio and
control messages simultaneously. To be sure audio always flows reliably, we take
advantage of the priority functions that are part of the switched Ethernet system.
Compared to the original, modern Ethernet has an additional 4 bytes of data inserted into the frame’s header. One field provides a 3-bit priority flag, which allows
designation of eight possible values.
Priority Level
IEEE Recommendation
Livewire Assignment
7
Network Control
6
Reserved
Livewire Audio
5
Voice
Telephone Audio
4
Video Conferencing
3
Call Signaling
2
High Priority Data
1
Medium Priority Data
0
Best Effort Data
Livewire Control &
Advertising
Ethernet Priority Assignments
Highest-priority packets have first call on the link’s bandwidth. If high-priority
packets are in the queue and ready to go, the lower-priority ones wait. If there is
Intro to Livewire 79
not enough bandwidth for both, low-priority packets will be dropped – but this is
not a problem, as you will soon see.
The adjacent graphic
shows only two queues,
but the idea is the same for
four or eight. Switches used
for Livewire must support a
minimum of four queue and
priority levels. Some low-end
switches have no support or may have only two queue levels.
If you have multiple switches in a hierarchical configuration, the priority information is carried automatically to all the switches in a system.
This prioritization scheme works only within a facility’s local area network.
Because it is at the Ethernet layer, it has no effect past the router boundary into the
internet. However, we also set the priority bits in the IP header to match the Ethernet priority so that as LAN switching evolves to use more Layer 3 intelligence, our
packets will be ready.
The Role of TCP
TCP is a key to sharing high-priority audio with best-efforts data on a single
network link. Because the acronym TCP/IP is so often written, many people think
that they are necessarily and always joined. This is certainly not so. IP is independent from TCP and may well be used without it. For example, RTP/IP is specified for
streaming media, and UDP/IP is used for a variety of transmissions, such as DNS,
the internet’s name look-up service.
TCP has two functions: Ensuring reliable transmission and controlling transmission rate.
Routers and switches may drop packets when there is not enough bandwidth
to transmit them or when they are overloaded.
They also do not guarantee to deliver packets in the same
order as they were sent. And there is no protection for bit errors from signal cor80 Intro to Livewire
ruption. None of this is a mistake or oversight in the design of the internet. The
inventors knew what they were doing: they wanted control of any needed correction process to be as close as possible to the endpoints, consistent with the general
internet idea to move as much as possible from the center to the edges.
Certainly we need 100% reliable transmission for most data files – even a
missed bit could have bad consequences. TCP gets this done by using a checking
and re-transmission approach. Whenever TCP detects any corrupted or missing
data, it requests another copy to be sent and holds any data it might already have
in its queue until the replacement has arrived. Packets are numbered by the sender
so that they can be delivered to the application in correct order. The application
always gets good data – but it could be after significant delay.
Transmission rate control is essential for most internet applications because
the bandwidth of the many transmission “pipes” from sender to receiver are almost
always different. And the available bandwidth to a particular user is constantly
changing as the demands from the many users sharing the net ebb and flow. Think
of the common case that you are at home with a 56k modem connected to your
office server. The server and its local network can certainly send data faster than
your modem can take it. And the available bandwidth on the public part of the net
is varying. So something needs to slow the sending rate to match both the network and your modem’s ability to receive. That function is performed by TCP. This is
called flow-control. While the details are complicated, the principle is simple: a TCP
sender monitors the condition of the buffer at the receiver so it knows how fast the
data is arriving and can adjust its transmission rate to maintain the correct average
buffer fill.
TCP also has a function called congestion-control. While it also controls rate,
it does it with a different mechanism and for a different reason. The re-transmission procedure we discussed earlier addresses a symptom of network congestion,
but not its cause – too many sources trying to send at too high a rate. To treat the
cause of congestion, we need to have some way to throttle senders when needed.
TCP’s congestion control is unusual in that it is a service to the network at large
rather than to the individual user. It was conceived as a way to fairly ration network
Intro to Livewire 81
bandwidth to all users. To do this, TCP monitors dropped packets, assuming that
lost packets signal congestion. When a new connection is established, a slow-start
function causes the rate to start low and ramp up until a lost packet is detected.
Then the rate is cut in half and the ramp up begins again. In this way TCP is always
probing for the maximum available bandwidth and always adjusting its transmission rate to match. Its really a very slick technique, one that is very well suited to
getting the fastest transmission of bursty data over a shared links.
We’ve gone into a lot of detail on TCP because it is one of the keys to Livewire’s
audio being able to share a network link with other general data. The Ethernet
switch handles congestion in a similar way to the routers in the internet – when
there is too much traffic, it drops packets. But we have something very important:
Priority. Audio packets are assigned higher priority than general data. So they are
never dropped before all TCP packets are. The usual condition is that some percentage of the link is filled with constant audio streams and the remaining capacity is
left for data. For example, an 8-audio channel LWIO with all channels active will
take about 40% of its 100BASE-T link, leaving 60% for data. But, we could have one
or we could have a dozen audio streams active on a link – and this number could
well change over time. TCP automatically finds how much bandwidth it can use
and adjusts it rate naturally to match.
You might be thinking, “All well and good, but what if we put too many highpriority packets into the link? Won’t we have drop-outs then?” Yes, we would. But
we never allow this to happen. Remember that each Livewire node knows about
the link attached to it because it “owns” it. The link from a node to a switch is fullduplex point-to-point with no sharing. The node knows how many streams can fit
and never is allowed to send more into or request for reception more than can be
supported by the link.
All of the above applies to a shared link, such as for a delivery PC that needs
both audio and data. It is the Ethernet switching function that allows the overall
network to be shared, since general data never even gets to a port connected to a
Livewire node.
82 Intro to Livewire
Virtual LANs (VLANs)
This is a technology that came to Ethernet along with switching. It is a way to
have “virtual” isolated LANs, while using common hardware.
Remember those Broadcast packets? They go to all devices, even with an Ethernet switch in the picture. If there are a lot of computers on the network, there could
be a lot of traffic generated by these transmissions. VLANs can be used to contain
broadcast packets, since they are not propagated outside of their assigned VLAN.
VLANs can also be used for security. If the LW network is on a different VLAN
than the internet, a hacker would not be able to gain access to your audio streams
or send traffic on the audio network.
In a LW network that is shared with general data, VLANs offer protection against
a computer that could have a problem with its network software or interface card.
The Ethernet switch can be configured so that the ports to which general computers are connected are not able to forward packets outside of the assigned VLAN, so
can never reach LW audio ports.
Finally, VLANS protect against the rare case that an Ethernet switch has not yet
learned an address and has to flood all ports until it knows the specific destination.
All LW devices allow choice of VLAN. We recommend:
•
If you have a separate network for Livewire audio, you can just stay with the
default VLAN 1 and pay no more attention to this topic.
•
If you have your Livewire network connected to the internet, or shared with
A router that bridges
VLANs is sometimes
called a “one-armed”
router because it has
only one Ethernet
port, rather than the
usual two. But you
can use the same
router that you have
for your internet
link to provide this
function. Or maybe
better: Some sophisticated Ethernet
switches provide
an internal routing
capability that can be
used to bridge VLANs.
Simpler and saves
boxes.
a large group of office computers, use the default VLAN 1 for general data and
VLAN 2 for LW audio and control.
A router must be used to bridge the traffic between VLANs, while providing a
“firewall” function. So if you have PCs on the LW network that will be used for audio
and web surfing, etc, you will need to provide this bridge. You will also need this
to access LW devices on VLAN 2 with PCs connected to VLAN 1 for configuration,
monitoring, etc.
Tagged vs. Port-Based VLAN Operation
When the VLAN information embedded in the Ethernet frame is used to direct
Intro to Livewire 83
the switch, this is called tagged VLAN operation. With LW devices, when you configure a VLAN value, the device will transmit Ethernet frames with the embedded
value you specify. But some devices are not able to do this. As if this writing, Windows does not support VLAN tagging, for example. That means the switch itself
has to insert the tag – a procedure called port-based VLAN. In this case, all frames
that enter from a particular port are tagged with a certain value, defined by switch
configuration. To enable this, you must configure the switch appropriately.
There is one special case: Frames tagged with VLAN=0 are called priority frames
in 802.1p standard. They carry priority information, but not the VLAN ID. The switch
will translate to whatever VLAN is default for that port. This is useful if you want to
use port-based VLAN assignment at the switch, rather than tagging from the LW
device.
Many switches allow a combination of port and tagged VLAN. In this case you
assign a default value to the port and frames either with no tag or with tag=0 go to
this default VLAN, while tagged frames override the default.
It would be possible to use port and tagged VLAN in combination. For example,
you use LW node configuration to put all your audio devices onto VLAN 2. But since
Windows doesn’t support tagged VLANs, how would you connect a PC for configuration and monitoring? Using the port-based assignment, you can set a port to be
always VLAN 2 and plug your PC into it.
Some switches have other options for assigning VLANs. Assignment could be
“hard-coded” to MAC addresses with a configuration set-up. Or layer 3 protocols
(TCP, RTP, etc) could be detected and used as a way to make VLAN assignments.
These may have their place, but since Livewire devices provide the tagging, it
doesn’t seem that these methods make much sense. The less you have to configure
the switch, the better.
Ethernet Switching versus Routing
Both switches and routers examine packet addresses and send them appropriately on their way. So what is the relationship between these technologies? Why
and where would you use one versus the other? Routing works at Layer 3, where IP
84 Intro to Livewire
information resides, while Ethernet switching works at Layer 2. Routing is a much
more complex operation than switching, where multiple paths from one site to
another are the norm, and it is the job of the router to find the optimum route (get
it?), which may well be changing from minute-to-minute. On the next page is a
comparison of the two side-by-side:
As do switches, routers also support multicast and prioritization. So it would
be possible to have a routed LW network on top of a switched one. You’d still need
the layer 2 switching because Ethernet would still be the transport layer. Livewire
fills the IP header with all required information and does it in a standard way. So if it
ever becomes a good idea to route LW, we are ready.
Switch
Router
Layer
Layer 2 / Ethernet
Layer 3 / IP
Function
Determines to which port the addressed node is connected and
switches incoming frame to it
Finds the best route from among
many and forwards packet to next
router along the path
Terminology
“Switching”
“Forwarding”
Technology
Simple table look-up in hardware
Complex dynamic best-route determination in software
Standards
IEEE
IETF
Ports
Many, connecting mostly to end
nodes
A few, connecting to networks and
Telco lines
Cost
Low
Expensive, but coming down
Cisco is the most
famous and by far
the most widely deployed router brand.
They pretty much
have a lock on the
router market, while
there are a bunch
of vendors selling
Ethernet switches. Is
it any surprise that
Cisco wants you to do
everything at Layer 3?
Ethernet Switch vs IP Router Comparison
Traditionally, routers did their work with software, while switches had dedicated
hardware chips. Now there is something called Layer 3 Switch, a hybrid of traditional routers and Ethernet switches. Layer 3 switches perform their forwarding
– whether Layer 2, Layer 3, unicast, multicast, or broadcast – in hardware. Software
handles network administration, table management, and exception conditions.
As the cost of such devices falls, it could well be useful to have them at the core
of a LW audio system. Indeed, already some low-cost switches have basic Layer 3
functions such as simple routers that can pass packets from one VLAN to another
Intro to Livewire 85
Livewire Networks
So now we are ready to consider all that has gone before in the context of
Livewire. And to begin the discussion of Livewire-specific technologies.
Quality of Service (QoS)
An important concept in a converged network is Quality of Service. When general data is the only traffic on a network, we only care that the available bandwidth
is fairly shared among users and that the data eventually gets through. But when
our studio audio and general data are sharing the same network, we need to take
all the required steps to be sure audio flows reliably.
Our method for achieving QoS is system-wide, with the following components
each contributing a part of the whole:
•
Ethernet switch. Allows an entire link to be owned by each node. Isolates traffic by port.
•
Full-duplex links. Together with switching, eliminates the need for Ethernet’s
collision mechanisms and permits full bandwidth in each direction.
•
Ethernet Priority assignment. Audio is always given priority on a link, even
when there is other high-volume non-audio traffic.
•
IGMP. Ensures that multicasts – our audio streams – are only propagated to
Ethernet switch ports that are subscribed.
•
Limiting the number of streams on a link. Nodes have control over both the
audio they send and the audio they receive, so they can keep count and limit
the number of streams to what a link can safely handle.
The result is rock-solid QoS, combined with the ability to share audio and data
on the same or interconnected networks.
Source Advertising
Audio source nodes advertise their streams on a special multicast address. Receive nodes listen to these advertisements and maintain a local directory of available streams. The advertisements are sent when the streams first become available
and at 10-second intervals after that. (Actually, only the data version number is sent
86 Intro to Livewire
every 10 seconds. The full data is advertised only upon entering the system, on any
change, and on explicit requests from those having detected the data version number increase.) If a node’s advertisements are not received for 3 consecutive periods,
it will be assumed to be removed from service. There is also an explicit “stream
unavailable” announcement.
Receive nodes maintain a local table of available streams and their characteristics, updated as any new information arrives. Sources are cleared from local tables
when an explicit message is received announcing that a stream is no longer available, or when three consecutive advertisements have been missed.
A receive node may be configured to be permanently connected to particular
multicast streams, or users may select audio sources from a list. The list may display
all available sources, or a filtered subset.
Synchronization
You may ignore this matter completely – and your Livewire system will work automatically “out of the box”. But there are times when you might want to modify the
default behavior of the clock sync system, so here is some detail on how the system
works.
Livewire needs careful system-wide synchronization in order to have small
buffers for low-latency streams. If we did not have a distributed way to derive a bit
clock, we would eventually have buffer over or under-flow, resulting from the input
and output node clocks being not exactly the same frequency.
A PLL (Phase Lock Loop) in each Livewire node recovers the system clock from
multicast clock packet that is being transmitted at a regular interval. At any given
time, one Livewire hardware device is the active system clock master. In the event
the master develops a fault or is removed from service, the local PLLs in the nodes
are able to “ride out” the brief interruption and there will be no problem with operation.
All nodes are capable of being a clock source, and an arbitration scheme ensures that only the unit with the highest clock master priority is active. Clock mastership priority may be set by the user, or left to the default case of all being equal
Intro to Livewire 87
priority.
When the clock goes away for 3 consecutive periods, all capable units begin
transmitting clock packets, after a delay skewed by their clock mastership priority.
When a unit sees clock packets from a unit with a higher mastership priority on
Jitter in the timing
and PLL functions
ultimately set a lower
bound on output
buffers and therefore
audio delay. And
any drift in the time
calculation produces
buffer pointer drift.
Further, jitter in the
derived A-to-D and Dto-A bitclocks causes
sampling uncertainty
that can generate
unwanted noise in
the audio.
The LAN network is a
“noisy” environment
with packets of various kinds and lengths
being numerous and
unpredictable. Thus,
the PLL system needs
to be quite smart so
as to generate a reliable, consistent, lowjitter output, while
not being confused
by dropped or jittered time packets.
Our method for
handling this PLL
problem is subject to
a patent application,
to give you some idea
of the novelty and
complexity.
the network, it stops its own transmit of clock packets.
You can specify the clock mastership priority behavior. The clock mastership can
be made predictable, rather than end up being any node in the plant – maybe the
one down in an out of the way equipment closet.
Each node has a clock mastership configuration setting that can range from 0
to 7.
•
‘0’ means never - slave only
•
‘7’ means always - forced master (don’t use multiple forced masters in a system)
•
Factory default is 3. So all units have equal priority out of the box, and the
following is used to break ties (in descending order): lowest LW audio transmit
base channel, then lowest IP address, then lowest Ethernet address.
Livewire nodes have an LED labeled Master on their front panel that illuminates
when that unit is the clock master.
Synchronizing to AES3 Systems
To avoid passing audio through sample-rate-converters, we recommend that
Livewire be synchronized to your AES master clock, if you have one. Our Livewire
AES node provides this function, recovering the clock from an attached AES input
and creating a Livewire sync packet. The AES node must be active clock master.
Network Standards and Resources
We use standards whenever possible. Ethernet is standardized by the IEEE and
information is available on their website at www.ieee.org. Internet Protocol and associated technologies are standardized by the Internet Engineering Taskforce (IETF)
and much can be learned from their website at www.ietf.org. Documents are a free
download. Bookshops are full of books on Ethernet, IP, and networking and we offer a list of suggested reading.
88 Intro to Livewire
Livewire operates at both Ethernet and IP network layers, taking advantage of
appropriate standards-based resources at each layer.
Here are the resources we are using at the various layers:
Layer 1:
•
IEEE Ethernet Physical
Layer 2;
•
IEEE Ethernet switching
•
IEEE 802.1p/Q prioritization
•
IEEE 802.1p multicast management
Layer 3:
•
IETF IP (Internet Protocol)
Layer 4:
•
IETF RTP (Real-Time Protocol)
•
IETF UDP (User Datagram Protocol)
•
IETF TCP (Transport Control Protocol)
•
IETF IGMP (Internet Group Management Protocol)
Layer 5:
•
IETF NTP (Network Time Protocol)
•
IETF DNS (Domain Name Service)
•
IETF HTTP/WebIETF ICMP Ping
•
IETF SAP/SDP (Session Announcement Protocol/Session Description Protocol)
Here’s an interesting
application of LW AES nodes: Two LW AES nodes can be used as
a way to synchronize
two AES systems located apart, but with
an available IP path
between them. One
becomes the master,
connecting to a LW AES input. The slave
attaches to a LW AES output and is configured to recover clock
from it.
(in the Windows PC Livewire Suite application)
Network Time Protocol (NTP)
This is the internet’s standard for conveying time. There are a number of servers on the net that users can connect to in order to retrieve accurate time. There
are also boxes from manufacturers such as EXE that receive radio time signals and
translate them to NTP packets. Livewire does not need NTP, but some peripherals
do. For example, our studio mixing surfaces and Omnia processors use NTP to automatically synchronize to the correct time.
Intro to Livewire 89
A Note about Protocol Design
There is no question that among network protocols, the internet has been an
impressive success. One of the reasons for this was the approach its designers took
– and still use – when inventing its protocols. These are outlined in the IETF RFC
1958 document. We’ve taken the principles to heart in the design of Livewire. Here
they are, in priority order, and with our comments in parenthesis:
1. Make sure it works. Make prototypes early and test them in the real world before writing a 1000-page standard, finding flaws, then writing version 1.1 of the
standard. (Telos and Axia are practical commercial oufits, not an academic or
governmental organization. We had two years extensive lab tests of prototypes
in two locations and then real-world field tests at radio stations before locking
the core tech down.)
2. Keep it simple. When in doubt, use the simplest solution. William of Occam
stated this principle (Occam’s razor) in the 14th century. In modern terms, this
means: fight feature creep. If a feature is not absolutely essential, leave it out
– especially if the same effect can be achieved by combining other features.
(We believe firmly in this principle. We tried very carefully to add nothing unnecessary.)
3. Make clear choices. If there are several ways of doing the same thing, choose
one. Having multiple ways to do something is asking for trouble. Standards
often have multiple options or modes or parameters because several powerful
parties insist their way is best. Designers should resist this tendency. Just say no.
(It was just us – and we did say no. No committees or politics to cause bloating.)
4. Exploit modularity. This principle leads directly to the idea of having protocol
stacks, each of whose layers is independent of all the other ones. In this way, if
circumstances require one module to be changed, the other ones will not be
affected. (We built Livewire on all of the available off-the-shelf lower layers.)
5. Expect heterogeneity. Different types of hardware, transmission facilities, and
applications will occur on any large network. To handle them, the network
design must be simple, general, and flexible. (We had to accommodate both
dedicated hardware audio nodes and general-purpose PCs being used as audio
90 Intro to Livewire
nodes.)
6. Avoid static options and parameters. If parameters are unavoidable, it is best
to have the sender and receiver negotiate a value than defining fixed values.
(These were avoidable – we don’t have any such negotiated parameters. We do
have the receiver selection of stream types, but this is simple one-ended selection.)
7. Look for a good design, not a perfect one. Often designers have a good design
but it cannot handle some weird special case. Rather than messing up the
design, the designers should go with the good design and put the burden of
working around it on the people with the strange requirements. (Steve, Mike,
and Greg’s mantra! Make it work, make it solid, build just enough flexibility to
get the job done – and no more.)
8. Be strict when sending and tolerant when receiving. In other words, send only
packets that rigorously comply with the standards, but expect incoming packets that may not be fully conformant and try to deal with them. (We told the
s/w guys to do this. Hope they did!)
9. Think about scalability. No centralized databases are tolerable. Functions must
be distributed as close to the end-point as possible and load must be spread
evenly over the possible resources. (We kept very close to this idea – which is
the main spirit of the internet. We don’t have any central databases or other
pieces along these lines. We have a fully distributed system. If one part fails, the
others keep going.)
10. Consider performance and cost. If a network has high costs and there are cheaper variants that get the job done, why gold-plate? (Compare the power and cost
of our solution with others. Using simple off-the-shelf commodity parts was the
guiding principle for our work.)
Intro to Livewire 91
9: FREQUENTLY ASKED QUESTIONS
We know there will be questions. Here are some we’ve already heard, and some we imagine.
General Questions
Audio over IP reliable, it’s a new technology?
Axia uses the same technology that underlies VoIP telephony. Did you know that over half of
the Fortune 100 companies now use VoIP? Or that VoIP PBX systems now outsell the old kind by
a wide margin? With these systems, telephones plug into a standard Ethernet/IP network. Contrast this with traditional PBX phone gear — proprietary devices which required you to purchase
phone sets and parts exclusively from the company that built the mainframe. You were locked
into a single vendor, because the technology that ran the mainframe was owned by the company
that made the gear. (Kind of like the TDM router companies.)
IP is growing as a universal transport for almost any kind of signal. You see it now in television
studios, business teleconferencing, government communications, banking, etc. And it’s hardly
unproven, even for applications specific to radio studio infrastructure. There are plenty of people
successfully using it – now.
Can the network be used for general data functions as well as audio?
Most certainly, should you choose to do so. The Ethernet switch naturally isolates traffic. You may
even use one link for both audio and data, since the audio is prioritized. This will probably be the
case when a PC is connected to the network – you will sometimes want to download files, receive
email, etc. in addition to the audio stuff. Switch selection is important, though, and you must use
one tested and recommended by us. You could have two networks and link them as described
below.
Of course, we would never mix on-air audio and business functions or open ourselves up to hacking. Can I make this a completely separate network?
Yes, we understand and agree. You have a few choices:
•
Have a completely separate and isolated network for Livewire. Take advantage of Ethernet,
but don’t combine any internet or business functions with studio audio.
•
Have two physical networks and link them with an IP router. Correctly configured, the router
provides a security barrier.
92 Intro to Livewire
•
Share the network hardware for audio and general functions but isolate Livewire to its own
VLAN. Again, an IP router could be used to link the two networks.
I’ve heard that Axia costs half as much as the other guys. What did you leave out?
Nothing. Our cost savings compared to traditional routers are achieved by using standard, offthe-shelf switching hardware rather than custom-built solutions. It’s a lot less expensive to use
a mass-produced Ethernet switch available from any network vendor than it is to construct a
customized cross-point routing switcher, with its cards, frame and peripherals. This is the same
principle that has driven almost all stations to use PCs for audio playout and editing – they are a
lot cheaper and more powerful than any broadcast-industry specific machine could be.
Another way Axia saves you money is by eliminating sound cards. Professional multiple-output
sound cards are expensive. Instead, we wrote an IP-Audio Driver for Windows PCs that looks just
like a sound card to the OS, but streams audio in and out of the computer’s network card instead.
This driver provides 16 simultaneous channels of stereo I/O for less than half the cost of some 8channel broadcast audio cards.
Furthermore, eliminating the sound cards also eliminates the cost of the I/O needed to get their
audio into the switching network. With a traditional router, PC audio must be brought in through
a router input card or console module; bringing multiple channels of audio into the system in this
manner (from workstations or digital delivery systems) can significantly increase the overall cost
of the system. Our IP-Audio Driver eliminates this cost. So Axia clients usually realize several thousand dollars worth of savings over and above the cost of the sound cards themselves.
Do Axia networks have any single points of failure? Is there a central ‘brain’ I can lose that will
take the system down?
Axia networks are distributed, with no central box. Ethernet networks can be designed any number of ways, including those that are fully-redundant and self-healing. Normally, our clients build
larger facilities with “edge switches” serving each studio, connected to a redundant core. Each
studio is able to operate stand-alone.
How do contact closures get in and out of the network?
The Element power supply also has 40 GPIO connections. We make the same box without the
power supply, so if you need more GPIOs elsewhere, such as in a Tech Center rack, just install a
GPIO box there.
But we expect more and more, control functions will move from “dumb” contact closures to
Intro to Livewire 93
smarter network transactions. For example, a delivery system that now uses a closure to start play
could just take a packet over a network for this function. But, beyond this replacement of today’s
closure-based functions, you could have song title text or other information flowing between the
systems. A satellite receiver could have program information and requests for specific local tasks,
not just a “start something” closure.
Are there any problems with delay of control commands over the network? I’ve heard of other
systems using TCP/IP that have problems in this respect.
No, Livewire control latency is very small – no more than 50ms for hardware GPIO closures from
Surface button pushes. We are using a special network protocol we invented called R/UDP (Reliable UDP) rather than TCP/IP, in part to be sure control delay is low.
Can I use Livewire without the Element console?
Yes, of course. You could just use it as a snake or router system and connect whatever consoles
and other equipment you like.
How does Livewire compare to other audio networking systems?
Livewire is an audio networking system which allows real-time uncompressed digital audio to be
conveyed over standard Ethernet hardware. Livewire is extremely low latency, which is especially
important for broadcast facility operation, where live monitoring and cascaded links are common.
Second, Livewire includes all the technology you need for practical studio application: Switches
are controlled, sources are ID-ed and advertised to receivers, GPIO over the network is covered,
etc. Third, Livewire connects directly to PCs – no soundcard or other hardware is required.
Livewire is a not just a technology, but rather a get-the-job-done solution. We offer you all the
pieces you need to build a modern broadcast studio. Nodes, Engines, Surfaces, PC drivers. We are
experienced broadcasters, so we know how to support radio studio applications.
So, what about that delay?
For live monitoring, such as when an air talent hears his own microphone in headphones, 10ms
is the limit before noticeable problems. We’ve kept Livewire link delay to below 1ms, so a number
of links can be successfully cascaded. To put this in perspective, a normal professional A-to-D or
D-to-A converter has about .75ms delay.
Does latency increases whenever you add inputs. In other words, the more sources you add,
the higher the delay, right?
No, Livewire’s latency remains fixed at the same low value regardless of the channel count. You
94 Intro to Livewire
can run a system with a thousand channels and the latency will be the same as for a single stereo stream. Indeed, the delay is so consistent that channel-to-channel phase shift is less than 1/4
sample.
The total latency of an analog input to analog output using the Axia Livestream format is about
2.75 milliseconds:
•
The time through the A/D and D/A converters is about 1.5 ms.
•
The network transit time is 1.25 ms.
To put this into perspective, the analog input to output latency on a self-contained BMX-Digital is
about 1.75 milliseconds.
The backplane of a modern Ethernet switch can handle full duplex traffic on all ports simultaneously without any packet loss. And since Axia component links are designed so that they never
exceed any port’s capacity, we never exceed the switch capacity. The way we prevent port overload is simple: we “own” each port. Every Axia audio node is plugged into an unshared 100Base-T
port on the switch. Even when all of a node’s inputs and outputs are active, we are still well under
the bandwidth of the ports, and the switch is completely under control. Because the switch has
the backplane capacity to handle all ports fully loaded, the system performance doesn’t change
from one to thousands of audio channels. Let’s explore the issue of switch capacity a little further. We know how much capacity is required per port for each node, and we know that a node
will never produce or consume more than 16 stereo streams total. But what about the studio
mix engine? To support a large console with a lot of buses, inputs, mix-minus outputs, etc., you
may have 40 or 50 simultaneous signals (or more). Because this could exceed the port capacity
of a 100Base-T port, the mix engine is connected via Gigabit Ethernet only. Using Gigabit for the
engine, we could support a 200 fader console with 200 outputs and still have room to spare! Each
console’s mixing engine gets its own Gigabit port.No. We wouldn’t be proposing any system that
wasn’t full broadcast quality. With Ethernet switching, each device owns all of the bandwidth on
a link so there is no possibility of contention or audio loss. If a node needs both audio and data,
such as a PC running an audio editor and a web browser, audio is prioritized and always has precedence. We’ve had thousands of hours of testing in our lab with careful logging of packet transmission, not to mention the hundreds and hundreds of consoles that are already in the field. So
we can assure you that it works.
Intro to Livewire 95
How can you promise live audio over Ethernet? Won’t it drop out?
No. We wouldn’t be proposing any system that wasn’t full broadcast quality. With Ethernet switching, each device owns all of the bandwidth on a link so there is no possibility of contention or
audio loss. If a node needs both audio and data, such as a PC running an audio editor and a web
browser, audio is prioritized and always has precedence. We’ve had thousands of hours of testing
in our lab with careful logging of packet transmission, not to mention the hundreds and hundreds of consoles that are already in the field. So we can assure you that it works.
This idea comes from many years ago when Ethernet used a shared coax cable. In rare cases two
devices would grab the bus simultaneously. When this happened, one would back-off and send a
few milliseconds later. These were the famous collisions. But With today’s switched Ethernet, there
is no shared bus – each device completely owns its own full-duplex link. There are never collisions
or lost packets as a result of network congestion; it’s physically impossible.
But the Internet is a packet network and the quality is not very good for audio.
Right. Internet bandwidth is not guaranteed, so there can be problems when there is not enough.
But you completely own and control all the pieces of a Livewire system and there is more than
enough bandwidth on a switched Ethernet LAN, so performance is fully reliable.
Are you sure this is robust enough for 24/7 operation? My Windows networks always have
downtime.
Livewire equipment is based on tight, embedded hardware and software. The Ethernet switches
we recommend are fully professional devices with high reliability and options for redundancy.
Is your system compatible with Program Associated Data?
Yes. Devices that generate PAD plug into the Axia network; the information they supply is sent
along with its associated audio, and any devices that need it can also plug into the network and
retrieve it. This means that you can send audio and PAD together, without incurring extra costs for
separate audio and data networks.
Are both logic and audio routed together?
Of course. IP is great for data, no? Logic commands from external devices like CD players, DAT
machines, etc., enter the network using GPIO Nodes. The logic data is then “bound” to the audio
stream, and is routed with it to whatever console the source is loaded on.
Devices equipped with Livewire interfaces (like the latest Telos Zephyrs and phone hybrids,
Omnia audio processors and IDC satellite receivers, for example) supply audio and control logic
96 Intro to Livewire
directly from the device to the Ethernet switch over a single CAT-5e connection, further simplifying in-studio wiring and making Livewire’s audio+logic routing even more convenient.
Do you use any compression? I am concerned about codec cascading.
Livewire audio is uncompressed 48kHz/24-bit. It would be possible to have compressed streams
sharing the Ethernet, but this is not a part of Livewire.
Can I connect two studios across town with a T1 line?
Yes, but not the way you’re probably thinking. Remember that LW audio is uncompressed 24-bit
48kHz, so each stereo stream is 2 Mbps. A T1 is 25% less than that. To get this done over a reasonable phone line, you’d use Telos Xstreams to reduce the bit rate for connection across town via T1
or fractional T1 or use the Telos Zephyr iPort MPEG Gateway. This unit can be used in the way that
you were thinking. The unit allows multiple channels of stereo audio across any network with
guaranteed QoS, such as T1 and T3 connections, or MPLS networks. It contains six stereo MPEGAAC codecs in one box and can connect to your Axia Network using a single CAT-6 cable for all
I/O — or pair with an Axia AES or Analog Audio Node for use as a standalone multiple-stream
codec.
How do I connect this to my Zephyr?
Easy. Use any analog or AES I/O node ports or directly with the embedded LW port.
I’ve heard that there’s a PC inside your nodes, I don’t want to trust all my audio to a PC, is this
true?
No, Axia audio nodes do not have a PC inside them. They are “embedded” designs.
Our mixing engine, on the other hand, does. These days, Intel processors and motherboards are
the best way to get tremendous power at low cost. But we are using these as if they were embedded DSP processors. There are none of the components that cause problems in PCs, like hard
drives or general purpose operating systems. The software core is a special high-reliability realtime Linux variant that is dedicated only to the mixing application. Our engineers designed the
engine for bulletproof, 24/7 reliability.
Is Axia more expensive to install than traditional routing systems?
In fact, Axia costs lots less to install, because everything in an Axia network connects using offthe-shelf Ethernet cables, which carry multiple uncompressed channels of stereo audio. 100BaseT links can carry 25 audio channels simultaneously; Gigabit links can handle 250. The money
saved just from the elimination of expensive multi-pair cable for studio interconnects can be
Intro to Livewire 97
significant.
Even our audio connectors are designed to promote fast, inexpensive installation. All of our Audio
Nodes use the Radio Systems StudioHub+ RJ-45 standard for I/O jacks (except for mics, which
use standard XLR connectors); a huge variety of adapters are available from Radio Systems for all
kinds of devices. Tally up the savings in labor realized from not having to purchase and hand-solder hundreds of XLR and RCA connectors, and the money saved becomes even more impressive.
There’s considerable time saved during Axia installations as well. Due to the reduction of cabling
and the quick connection of devices, our clients tell us that installation of Axia networks goes 30%
to 50% quicker than wiring studios the traditional way.
PCs and Livewire
Tell me about your “sound card” driver for workstations.
The official name is “Axia IP-Audio Driver for Windows”. It makes the Livewire network look like a
sound card to a PC Windows application. Most audio applications should work unmodified.
Building Livewire Facilities
I’ve got a large facility. How many studios can I interconnect?
There is no limit. You may have as many studios and audio channels as your Ethernet switch can
support. Switches come in all sizes, some with hundreds of ports. And multiple switches may be
cascaded to expand ports. We recommend that you use a switch per studio to isolate any problems to a defined area. These are then interconnected with a backbone. Switches may be physically associated witch each studio or may be all in a central location, as you prefer.
What about for smaller stations? This all sounds pretty sophisticated for a simple set-up.
Look at Ethernet for data applications… You have everything from a single PC connected to a
printer to a few PCs in a small office tied to the internet and a couple of printers to huge campus
networks with thousands of nodes. This is one of the reasons we went with Ethernet – you can
use it for big and small facilities. The technology and economics naturally scale to suit the application size. We figure, in fact, that small stations may benefit the most as they gain routing capability at a very modest cost.
98 Intro to Livewire
This seems like a lot of IP to keep track of. What administration tools does Livewire have?
All Livewire devices have a web browser control and monitoring capability. Keep the IP numbers
in a “favorites list” and you can easily check them. Or make your own web page with all the links.
An additional tool is the Axia iProbe Network Management Console. iProbe is an intelligent network maintenance and diagnostics tool that makes managing, updating, and remote controlling
your Axia system very easy and intuitive.
How do analog sources become part of the network?
With Axia Livewire nodes. These come in variants for line and microphone application. Over time,
you can expect that codecs, hybrids, processors, etc. will have direct Livewire connection ports.
What about AES?
We have a node that interfaces your AES audio to the network. This is a direct bit-to-bit procedure
with no conversion of any kind. You can also sync a Livewire system to an AES master clock.
What is the best audio format to use with Axia systems?
Axia networks don’t care what format your music files are stored in. During the playout process,
your playout software will uncompress any compressed-format files (MP3, MP2, apt-x, etc.) and
present them to the Axia IP-Audio Driver. What this means is that all audio that moves within the
Axia system is the same - uncompressed.
So, the question really becomes, what audio format is best for your storage needs, your convenience, and the desired audio quality you want to have on-air. Our feeling, since large capacity
hard drives are very cheap nowadays, is that it’s better to store all audio in a linear fashion, as the
resultant audio quality will be higher, especially after any audio processing.
Mix-minuses are always tricky and I’ve heard that your consoles are good at mix-minus. What’s
so different about the way you do it and how are they generated?
We’re part of Telos, so as you might imagine, we’ve studied mix-minus for a long time. And we’ve
always been amazed at how complicated and confusing it is to set up correctly. With today’s radio
shows relying heavily on phones and remotes, something needed to be done.
Our Element consoles generate mix-minuses automatically, on-the-fly, without any intervention
needed from talent. The way it works is simple: when a caller is on the air, he hears the main Program feed, minus himself. When off the air, he hears a special “off-line” phone mix that can contain
audio sources: pre-fader audio from the host mic, other phone callers, etc.
Best of all, mix-minus settings for audio sources such as phone hybrids and remote codecs are
Intro to Livewire 99
assigned to the source itself, not the console fader — so when a source that needs a mix-minus is
loaded onto any fader, on any console, the mix-minus settings are automatically loaded too.
At the physical level, mix-minus is easy, too. Livewire carries audio in both directions, so one RJ-45
covers everything.
How many mix-minuses can your consoles have anyway?
One for every fader! There’s a lot of processing power in our Studio Mix Engines, enough that Axia
consoles can provide automatic mix-minuses simultaneously for every fader on the console. You’ll
probably never need to have 40 mix-minuses running at once, but isn’t it nice to know that you
could?
You said I can get RS-232 data through the system. How is that done?
Using 3rd party devices, such as from Lantronics, your serial data can go anywhere across the
network and be used where it’s needed.
Most companies recommend that I bring them on-site to help install and configure their systems. Do I need your help to install an Axia system?
With those other guys, you’d better hire their systems engineers. With us, it’s much easier! While
we’re happy to come and help commission your new Axia network, it’s not necessary. If you know
how to use a Web browser and plug a telephone into the wall, you’ve got all the skills needed to
install and configure your new Axia network. And Axia Technical Support is there to help if you
need it, too.
Ethernet Media
Are optical audio links supported?
Livewire is fully compatible with copper and fiber connection types. We imagine a common configuration to be switches dedicated to studios with 100BASE-T copper connecting nodes, engines,
surfaces, etc. A fiber backbone connects the switches in order to share audio among the studios.
What Ethernet rates do you support?
Nodes connect with copper 100BASE-T links. PCs may use 100Mbps or 1000Mbps, copper or fiber.
Our processing engines use 1000BASE-T. Switch-to-switch links may be any supported Ethernet
media. Media converters allow the use of fiber on nodes, such as for extended-range snake applications.
100Intro to Livewire
What happens if someone accidentally unplugs a cable?
Axia audio nodes “advertise” the presence of their audio streams to the entire Livewire network.
So if someone unplugs a node, the sources attached to it will be offline. But all you have to do is
plug that node back in, and the node will “advertise” that the audio streams are available again.
Within 10 seconds, all destinations that need those sources will be back up and running.
Additionally, among the many features of Pathfinder is a silence detect function that can be programmed to switch to another feed should one stop working for any reason.
One of your competitors says that you’ll have to replace the Ethernet switches every 3 – 4
years. And when you do, you’ll have to load all new software to work with it (if you can even
make your stuff work with it).Is this true?
Ethernet is a standard, IEEE 802.3. Axia gear works with any switch or router that supports the
standard. We do generally need higher-end switches because Livewire uses advanced multicast
and Quality-of-Service features that are not included in low-end switches.
You might want to upgrade to a larger or more powerful switch for some reason in the future; for
example, if you were to add more studios to a cluster. Or maybe you would like to change from
copper to fiber for some kind of remote uplink connection. Or you might want to replace an older
switch at some point. Thanks to IEEE 802.3, the replacement switch would simply plug into the
network and function, with no software changes needed on any equipment. But there is certainly
no requirement to do this at any particular time interval. Ethernet’s open and evolving nature
gives you choices that closed systems don’t. That’s an advantage.
With all this talk about switches and switch configuration, do I have to be an IT expert to run an
Axia system?
We’re engineers, and we like to talk tech. Sometimes, we talk about the tech more than we need
to! But in this respect, an IP-Audio network is like a car: you don’t have to understand how the engine works in order to drive it. Just connect two pieces of gear together with CAT-5e and they will
talk to each other — like plugging a mic into a mixer. The Livewire protocol takes care of routing
the audio without any need for intervention from you. And the equipment interface is all webbased with GUI control. It works intuitively, and you don’t have to know anything about the tech
inside to make it work.
That having been said, another of the advantages of Ethernet and IP is that bookshop shelves are
full of well-written books that can explain any aspect of standards-based networking at any level
Intro to Livewire 101
of detail you might want.
The Internet and Livewire
What about hooking up over the internet? With my studio audio in IP form, can I just plug a
port from the switch into an internet router? Why do I need ISDN anymore?
As the internet becomes ever more ubiquitous and bandwidth more plentiful, arguments for using it for audio transmission become more convincing. A gateway device could perform compression from LW’s PCM to a lower-rate bitstream using a codec like MPEG AAC. The main problem to
be overcome is the internet’s lack of any Quality of Service guarantees; a “net storm” that starves
bandwidth and drops audio might not be a big deal to a kid at home with his computer, but it
sure wouldn’t be good for an important on-air feed. Private networks with reserved capacity are
one answer. Another could be the “resource reservation” and “differentiated services” technologies
that has passed out of the laboratory and might eventually be implemented by Internet Service
Providers – at a cost, of course.
In theory, RSVP, diffserve, mpls, IPV6, and other emerging technologies will in due course offer
us reliable audio transmission. However, given the slow pace of new tech adoption at the core of
the public internet (nothing much has changed for a decade) and the problems with scaling the
lab work to the real world, perhaps the following observation applies: Sometimes there is a gap
between theory and practice. The gap between theory and practice in theory is not as large as
the gap between theory and practice in practice.
In our view, the wait for ISPs to offer QoS guarantees at a reasonable price is likely to be long. And
when they do, transmission delay is still probably going to be an issue for live interactive broadcasts. So, it looks as if ISDN is going to be the best option for most remote hook-ups for awhile.
Our Zephyr codecs support direct LW connection, so you can use them to get a remote link into
your Livewire network that way and effectively have the same result – albeit at ISDN’s per-minute
cost.
All that having been said, for some non-critical apps, an occasional drop-out might be acceptable
and a gateway with appropriate buffering and error recovery might be useful.
The best option would be a product we developed just for this type of application. The Telos
Zephyr IP offers high quality, reliable audio-over-internet for remote broadcasts. It featuring 24bit
102Intro to Livewire
AD/DA, AES/EBU, direct Livewire connection, and USB audio, the Z/IP will transmit from 16kBit to
256kBit compressed audio via UDP/RTP and TCP/IP. Network address traversal (NAT) and directory
assistance are provided by Z/IP Server technology developed and housed by Telos. Wireless connectivity through Ethernet via PCMCIA and USB are also options.
Can an Axia network can catch viruses?
There are no general purpose operating systems in Axia devices, so the answer is “No.” You can
keep computers attached to your Livewire audio network safe by keeping it isolated from data
networks.
The StudioEngine and Consoles
Can a single Mix Engine handle two or three Consoles?
Each Mix Engine allows a huge amount of power and flexibility to each Element, so it’s a one-toone ratio.
You are using a PC motherboard for the Studio Engine, right? It’s hard to believe that an offthe-shelf PC can do high-quality audio mixing. Are you sure there’s enough power there?
The amount of processing power in a Pentium-4 motherboard is staggering. If you don’t burn
it up with fancy graphic user interfaces, it’s amazing what you can do. With optimized software
design, a single P4 can outperform the largest, multi-DSP consoles and routers. We use a special
realtime and minimized version of Linux as the operating system, so there is no overhead needed
for the graphic displays, etc. that burden desktop PCs.
Will it be as reliable as the cards-in-a-frame approach? I sure don’t want this thing to crash.
Modern PC hardware is very reliable. The parts and board count of the PC solution is much lower
than a card-frame approach, so statistically the h/w failure rate is almost sure to be lower. The
most failure-prone device in a PC is the hard drive and we don’t use one; our software is loaded
from Compact Flash memory. There are no plug-in PCI cards to cause connector-related problems. We have redundant large panel-mounted fans turning at a relatively low RPM, rather than
the usual small heatsink-mounted high-RPM cooler. But more important for reliability is the
software. We are using an off-the-shelf Intel-made PC motherboard and processor, but we are
treating it from the software perspective as if it were an “embedded DSP” platform. We’re running a pared-down and highly-optimized version of the Linux operating system and our engine
Intro to Livewire 103
processing application code is carefully “written to the metal”. Unlike general PCs that must host
a lot of different application s/w, which are coming and going, sharing and releasing resources,
and potentially causing conflicts, we have only one application running in a carefully controlled
environment.
What does Telos know about console design, anyway?
Quite a lot, actually. Axia President Mike Dosch, the man who designed PR&E’s Radiomixer console, also designed Element. We’ve got the biggest R&D department in the broadcast industry,
and it’s filled with folks who’ve spent years around consoles, figuring out what works and what
doesn’t. We have over a dozen ex-broadcasters on our staff, by the way. It’s fair to say we know
our way around faders and buses.
The other guys say Axia consoles don’t have all the features they have. Like voice processing, is
this true?
Sure we do. Voice dynamics is a standard feature on all Axia consoles — and it was developed by
Omnia’s Frank Foti, who knows a little something about audio dynamics processing. Other standard features include per-source EQ and panning, headphone EQ and a one-touch off-air recording function, all of which can be set and saved to instantly recall each jock’s favorite settings. The
Element console features four stereo program buses, and four stereo aux sends with two stereo
aux returns for off-air production.
Element also does things that no other console can do. For example, you can connect the latest
Telos broadcast phone systems with one RJ-45 — all audio I/O for four hybrids plus line-selection
control from a tightly integrated drop-in panel enter the system using one skinny Ethernet cable.
It has optional dynamics processing on the headphones. It has a one-touch record-mode button
for recording phone bits. It has a powerful system that lets you set, save and recall events and
personal preferences for each jock. A clock that can be synched to NTP network time. A motor
fader option. Add our Pathfinder routing control package, and you get extensive user-programmable event handling, built-in silence-sense that automatically switches to a backup audio feed,
interactive on-screen virtual control panels for studio PCs, and much more.
I like the Element’s features and design, but I’m not ready to commit to Livewire for my full
facility. Can I just use your Surface and Mix Engine as a drop-in console replacement?
Sure, you can. Take an Element and Mix Engine, add the audio I/O you need and an Ethernet
switch and you have a stand-alone console that interfaces via analog or AES to your other equip104Intro to Livewire
ment.
I have decided that Axia is what I want; do you have a modular console?
Our Element console is modular. Element modules contain groups of four faders, which are easily
accessed for service by removing just two screws and a cable or two. They’re hot-swappable, too
– since all the mixing is done away from the board, in the Studio Mix Engine, you can even take
out a module while it’s on the air without affecting the audio in any way. Indeed, you can hotswap the entire console without disrupting audio.
Analog Audio & AES On RJs And Cat 5
Do I have to use CAT-6 to connect everything? That could get pricey.
CAT-6 is used for only two heavy-traffic network segments: connecting Axia mixing engines to
Ethernet switches, and connecting switches to each other. All other equipment is connected with
common, inexpensive CAT-5e cable.
You recommend an outer shield for analog audio. Why?
As a precaution. Shielded cable protects against RF and eliminates any possible crosstalk between cables in multi-cable bundles.
Is there any crosstalk between the pairs within the Cat-5 cable?
As long as your circuits are balanced, there is almost no left/right crosstalk inside the cable. With
a balanced input circuit that has 50 dB CMRR (Common Mode Rejection Ratio), separation will be
greater than 90 dB.
So, must all the audio and digital signals be balanced?
Generally, yes, or crosstalk will degrade. Unbalanced connections can be used for short runs only
and preferably with separate cables for left and right if you care very much about stereo crosstalk.
Radio Systems makes small devices that adapt unbalanced RCAs to balanced RJs for their StudioHub system that could be used to convert any unbalanced sources you have. AES3 digital audio
signals are always balanced and require no conditioning.
Is Cat-5 OK for AES3 digital audio?
A 1997 report, Review of Cables for AES/EBU Digital Audio Signals, conducted by the BBC Research and Development Department, concluded that Cat-5 shielded twisted audio pair cable
“offered the highest performance of all the cables tested here.” Their tests included coaxial cables
Intro to Livewire 105
and special cables specifically designed for digital audio. They preferred Cat-5 cables for their consistent performance and because they have the flexibility to support other signal formats.
Cat-5 cables are engineered for data rates up to 100 Mbps to support networks such as 100BASET. Since AES3 signals have a bandwidth of about 3 Mb/sec (depending on sample rate), AES3’s
requirements are well within the Cat-5’s guaranteed performance parameters. Dependable error-free transmission is possible at lengths up to 920 meters (over ½ mile). Cat-5 cables perform
well for AES3 because they are engineered to have characteristic impedance of 110 ohms and
extremely low capacitance – in the 12 pF/ft range. This yields low signal reflection and excellent
high frequency response, thus lowest error rates.
Is Cat-5 OK for analog audio?
Sure, it is! The low capacitance, needed for data’s high velocity and wide bandwidths, yield exceptionally flat analog audio frequency response, even over very long cable lengths. The tight,
controlled twists are good for hum and crosstalk rejection. Steve Lampen, a senior audio video
specialist for Belden Wire & Cable writes, “Digital cables make the absolute best analog cables.
You can go farther with flatter frequency response than with any cable designed for analog”.
Steve Lampen of Belden graciously wrote us a paper on the subject entitled “The Axia Guide to
Choosing Category Cable; you can get it from out website at www.axiaaudio.com/tech/cable/.
Also see Belden’s web site for more interesting and revealing papers on the subject of using Cat 5
and 6 cables for analog signals.
Livewire, Standards and Other Vendors
Is Livewire standards-based?
As much as it can be, yes. Standard Stereo Streams use all the relevant internet standards, the
main one being the RTP format defined in the IETF document RFC1889. Thus standard PC audio
players can play this audio. But, there is no standardized way to convey low-delay full-fidelity audio over Ethernet because you need a synchronization system and that doesn’t exist in either the
Ethernet or internet standards. So we had to invent that. Still, they are as standard as is possible to
be.
Also, we needed to implement a protocol for tagging audio sources with names and advertising
these to receivers. Nothing was available off-the-shelf, so we had to invent something for that,
106Intro to Livewire
too. Same for the GPIO-emulation functions.
Are you planning to share information so that other vendors can make gear that directly plugs
to Livewire?
Yes. Software vendors for PCs can use our driver to easily make their applications compatible.
Many delivery system providers have built in LW compatibility. Some examples are: BE, BSI, ENCO,
Google, OMT, and PSI. Makers of audio hardware such as 25-Seven Systems, AudioScience, International Data Casting, Radio Systems, and of course Telos and Omnia have already incorporated
LW into their products and others would have to coordinate with us to be compatible. Of course,
you can use whatever equipment you want via the analog and AES nodes.
Intro to Livewire 107
10: Resources
Networking is a field well covered by books and web sites. There’s plenty of information out there. Here is a
selection of some resources we’ve found useful. The links are active and the list is larger and up-to-date on the
Axia Livewire website.
Livewire/Broadcast
Axia Audio - www.AxiaAudio.com
Email updates by request at: [email protected] or by phone at +1 216 241.7225
Telos Systems - www.telos-systems.com/livewire
Radio Systems - www.studiohub.com — Vendor of Studio Hub components
Ethernet
IEEE - www.ieee.org
The standards body for Ethernet. The documents are now a free download, but will cost you a lot
of paper and toner – the basic Ethernet standard is 1,268 pages!
Charles E. Spurgeon, Ethernet: The Definitive Guide; O’Reilly & Associates, 2000
www.bellereti.com/ethernet/ethernet.html
Living up to its title, it is pretty definitive on basic Ethernet topics. Stops short of much detail on
switching and multimedia, however, and has a lot of coverage of older Ethernet technologies we
don’t use.
General Networking and Interest
IETF (Internet Engineering Task Force) - www.ietf.org
The Internet’s main standards organization. Look for the RFC (Requests For Comment) documents
to see in detail how the internet is built.
Andrew Tannenbaum, Computer Networks; Pearson Education/Prentice Hall, 2003
Our favorite general networking book. Popular college textbook covers it all, including multimedia, with a breezy style and at just the right level of detail: enough to be useful, but not so much
as to be overwhelming.
J. Naughton, A Brief History of the Future; Overlook Press, 2000
Not really so interesting for audio and Ethernet, but still worth reading for perspective. This history of the internet tells how it happened in a friendly – even charming – way. Lots of stories and
anecdotes. We particularly love AT&T’s repeatedly making clear that digital communication had
no future. (Something a lot like what we expect to hear from certain quarters regarding the future
108Intro to Livewire
of computer networks for studio audio.)
Cabling Information & Standards
Cabling Business - www.cablingbusiness.com
This magazine, targeted to cabling contractors, is a good way to keep abreast of the latest TIA/EIA
cabling specs. It is also a great source for innovative cabling accessories, testers, and installation
techniques. Those located in the USA can sign up online for a free subscription on the web site.
Jim Abruzzino; Technician’s Handbook to Communications Wiring; CNC Press, Chantilly VT,
1999.
This book is concise yet contains a lot of great information including proper technique for working with Cat. 5 cable and connectors. Small enough to keep with your toolbox.
Cabling Design - www.cabling-design.com — Cabling tutorials
TIA - www.tiaonline.org — Standards organization for cables
Global Engineering - www.global.ihs.com — Sells the TIA/EIA cabling standards
Cable and Contractor Supplies
AMP - www.amp.com — RJ plugs and tools
Belden Cable - www.belden.com — Leading cable supplier
Hubbell Premise Wiring - www.hubbell-premise.com — Devices for Cat 5, etc
Panduit - www.panduit.com — Marking and installation products
Siecor - www.siecor.com — Fiber optic cabling and components
Siemon - www.siemon.com — Punch blocks
Cable Testers
Fluke - www.flukenetworks.com —Full range of testers
Agilent - www.agilent.com — Top-end tester
ByteBrothers - www.bytebrothers.com — Low-end tester
Acterna - www.acterna.com — Fancy sniffers, too
Ethernet Switch Vendors
Cisco Systems - www.cisco.com
Intro to Livewire 109
Network “Sniffers”
Shomiti - www.shomiti.com
Network Associates - www.nai.com
Ethernet Radio Equipment
Aeras Networks - www.aerasnetworks.com
Dragonwave - www.dragonwaveinc.com
Exalt Communications - www.exalt.com
Mikrotik - www.mikrotik.com
Motorola Point-To-Point - www.motorolaptp.com
110Intro to Livewire
Appendix A: Livewire tech details
You don’t need to read any of this unless you want to know about the internal details.
Livewire Packet Structures
The speed of the link, the bit requirements of the header and payload, and the number of samples
that are combined into a packet determine link capacity. The facts are the more samples that are combined, the less the header overhead per packet, and the higher the efficiency and capacity.
There is a fundamental tradeoff: When we have more samples per packet, we have more capacity
– but at the expense of more delay. Good design means finding the best compromise.
The sampling rate and the number of samples that are combined into a packet determine delay:
Packet-time = 1/sampling-rate * samples-per-packet
There is one packet send buffering; three packets receive buffering, and the switch latency, therefore:
Link-delay = packet-time *4 + switch latency
Standard Streams
Standard Streams are compatible with internet standards. They use large packets so as to be very efficient with both computer resources and network bandwidth.
Function
Bytes
Notes
Interpacket Delay
12
This is not actually transmitted, but must be taken into account
for network bandwidth calculations
Ethernet Header
26
Includes the VLAN/priority fields
IP Header
20
Standard
UDP Header
8
Standard
RTP Header
12
Standard
Audio
1440
240 samples @ 48kHz, 24-bit, stereo
Audio (variant)
720
120 samples @ 48kHz, 24-bit, stereo
Standard Stereo Stream Packet Format
Total bytes per packet = 1440. Core delay = 5ms. (720 and 2.5ms with the variant format)
An Ethernet frame’s maximum data length is 1500 bytes, so you can see that we have chosen to pack
the Ethernet frame to nearly the maximum possible. There are two reasons for this: 1) the frame rate is
lowest possible to put the least burden on PC receivers, 2) the header overhead is applied to the most
data so the proportion of capacity devoted to audio vs. overhead is highest.
Livestream
Livestreams are specialized for low delay, so we can pack only a few audio samples into each packet.
Intro to Livewire 111
Because they are smaller, less buffering is needed and that means the time delay is lower.
Function
Bytes
Notes
Interpacket Delay
12
This is not actually transmitted, but must be taken into account
for network bandwidth calculations
Ethernet Header
30
Includes the VLAN/priority fields
IP Header
20
Standard
UDP Header
8
Standard
RTP Header
12
Standard
Audio
72
12 samples @ 48kHz, 24-bit, stereo
Livestream IP Packet Format
Total bytes per packet = 72. Core delay = .25ms.
The header load for RTP/UDP/IP is 40 bytes per packet, which is a significant piece of the network
bandwidth given that our audio is only 72 bytes. Most of the time this is of no consequence, since we
have plenty of bandwidth.
However, there are some situations where a “lean and mean” approach makes sense.
Surround Streams
By know, you realize that Livewire inherently carries multiple audio streams and surround mixing is a
built in feature in the Studio Mix Engine. This makes Livewire and Axia fully compatible with any HD and
surround capabilities that are being introduced into radio. It would also make for a very nice surface to
control your personal surround sound system in your living room.
Function
Bytes
Notes
Interpacket Delay
12
This is not actually transmitted, but must be taken into account
for network bandwidth calculations
Ethernet Header
30
Includes the VLAN/priority fields
IP Header
20
Standard
UDP Header
8
Standard
RTP Header
12
Standard
Audio
1440
60 samples @ 48kHz, 24-bit, stereo
Surround Stream IP Packet Format
Total bytes per packet = 1440. Core delay = 1.25ms.
Surround streams are actually multiple stereo streams carrying the 5.1 audio plus the original stereo
version simultaneously. Stereo streams carry 8 speaker channels in the following order: front left, front
right, center, low frequency enhancement, back left, back right, stereo left, and stereo right.
112Intro to Livewire
NETWORK LINK CAPACITY
Each Standard Stereo Stream has a bitrate of 2.304Mbps. A 100Mbps link can therefore carry 43 such
channels at full capacity and a 1000Mbps link can carry 430 channels.
Each Livestream has a bitrate of 3.776Mbps. A 100Mbps link can therefore carry 26 such channels at
full capacity and a 1000Mbps link can carry 260 channels.
In practice, links to hardware nodes will have a mix of Standard Stereo Streams, Livestreams, and control data. Our biggest node has 8 channels, so there is plenty of link capacity. PCs use the more efficient
Standard Stereo Streams and maybe only 6 of them maximum, so again there is plenty of capacity to
handle both audio and simultaneous file transfers, etc. Our Studio Mix Engines connect with 1000Mbps
links, so the sky is the limit there.
All of the above has been concerned with per-link bandwidth. The system bandwidth is effectively
unlimited with appropriate switches.
MULTICAST ADDRESS TRANSLATION
Livewire streams are multicast at both layer 2 and layer 3.
The Livewire channel number is automatically translated to the appropriate addresses at both layers
internally. You might want to know the translation algorithm because maybe you or your network engineer might need to check packets with a “sniffer” or Ethernet switch diagnostics. So here are the details.
Livewire channels range from 0 to 32767. Audio streams are mapped into IP and Ethernet multicast
addresses using the channel numbers for the lower 15 bits as follows:
IP Addressl
Type
239.192.000.0/15
Livestream and Standard Stereo Streams
239.192.128.0/15
4 addresses are our system defaults, the others not used (left for
expansion)
239.193.000.0/15
Back Standard Stereo Streams
239.193.128.0/15
not used (left for expansion)
239.194.000.0/15
not used (LW v1.0 used for Ethernet Livestreams)
239.194.128.0/15
not used (left for expansion)
239.195.000.0/15
Back Livestreams
239.195.128.0/15
not used (left for expansion)
239.196.128.0/15
Surround Streams
239.251.000.0/15
not used (left for expansion)
239.251.128.0/15
not used (left for expansion)
Intro to Livewire 113
The following special addresses are assigned:
IP Addressl
Function
239.192.255.1
Livestream Clock
239.192.255.2
Standard Stream Clock
239.192.255.3
Advertisement Channel
239.192.255.4
GPIO (UDP port 2060)
These all are within the range specified for “Organization-Local Scope” use by IANA – the Internet Assigned Names and numbers Authority. Routers do not propagate traffic on these addresses to the internet; they stay contained within LANs. (We also set the “link local” bit and TTL=1 in the IP header to further
ensure that streams stay local.)
The range supports our 32k channels, with up to 120 stream types per channel. We are only using
four types now, but there is plenty of room for growth.
Our motivation for mapping each type to a contiguous block rather than having the type in the
lower-order bits is to allow configuration of switches and routers on a per-type basis by specifying an address range. This direct mapping of channels to addresses also makes sniffing easier: it is simple to know
where to look for a particular audio stream.
IP addresses are mapped into an Ethernet MAC layer multicast, according to a de-facto standard process for this procedure. This process is as follows:
Using the Class D address, identify the low order 23 bits of the class D address.
Map those 23 bits into the low order 23 bits of a MAC address with the fixed high order 25 bits of the
IEEE multicast addressing space prefixed by 01-00-5E.
Example:
•
Assume: Channel = 80
•
Assume: stream type is Standard Stereo Stream
•
Then: IP address = 239.192.0.80 (dotted decimal)
•
And then: Ethernet MAC Address = 01-00-5e-00-00-50 (dashed hex)
Ethernet addresses are transmitted most-significant byte first, but least-significant bit first within the
byte, so in our example it is the 1 in the leftmost MAC address byte 01 that signifies a multicast address.
114Intro to Livewire
Intro to Livewire 115
Axia Audio, a Telos Company • 2101 Superior Ave. • Cleveland, Ohio, 44114, USA • +1.216.241.7225 • www.AxiaAudio.com
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement