Network Cables & Switches – Myths & BS
Netgear FS750T2 Switch
Introduction
I have worked at the pointed or origination end of audio and video all my professional life (40+ years) from professional recording studios to professional radio and TV broadcast and production. So I couldn’t let the following comments evaporate into the ether after reading about the new Melco S100 and Russ Andrews network switches for audio. Not to mention the new silver Ethernet CAT(x) cables from Artisan.
I make no apologies for what I am about to say as I try to bring some level of reality to all those unsuspecting amongst us who are being ‘sucked into’ this ridiculous hyperbolae.
The company I work for is the largest independent AV systems integration company in the world. We have some of the most skilled and talented design engineers creating the VERY BEST of designs for production, broadcast, radio, TV and sports. From 3D, virtual reality, high end audio and A/V to professional radio, TV and film production in some of the most complex 4K all IP environments that you can imagine, including fail safe military systems. So it rankles me to read the BS from vendors trying to scam their ill informed and naive customers.
My 2 Cents
So what am I referring to? As we move into the ‘wonderful’ world of digital connectivity and signal processing there are certain realities that should be understood by those at the receiving end. YES THAT’S YOU!
The world of IT has already established fantastic and reliable high speed network performance for the 100% reliable transfer of digital data between two points exceeding 10’s of Tb/s per link. Far higher than anything you will ever have to deal with in a home theater and over far greater distances and more arduous environments. Unfortunately in the world of IT, technical parameters like jitter, buffering and timing generally have minimal impacts on the networks performance and are dealt with with well established hardware and software techniques. This is because most IT data is just that, just data or a phone call, and exactly when it arrives doesn’t matter ONLY that it arrives with no errors or with errors that can be PERFECTLY corrected.
In order to keep the development and deployment of professional high bit rate digital video and audio cost effective, it was realized that using Commercial Off The Shelf Switches (COTS) would provide a fully developed means of switching high speed A/V digital data packet streams. Especially once we entered 4K and 8K digital video and IP production. There are some commercial broadcast-A/V manufactures like Evertz that design their own switch fabrics to handle these data packets in a more elegant way meeting all the standards listed below. The downside is that this elegant “proprietary” integrated solution is more expensive than the more basic and less flexible COTS solutions.
These COTS switches are managed . NOT simple unmanaged plug and play (P&P) switches. They provide a whole range of firmware features that need to be configured by a network manager to support packet switching, quality of service (QoS), virtual LANs (VLAN), spanning tree protocol (STP), internet protocol version 6 (IPv6) and a whole host of other functions that enable them to correctly and efficiently be configured to handle audio and video data packets.
In the world of video and audio those earlier mentioned network performance issues have a critical impact on the quality of the final A/V stream. So what those COTS switches required was an additional management layer for the A/V data and timing/synchronization control which has now been developed, is continuing to be developed and for which there is a whole laundry list of standards that are still being developed and added to. To include:
- ST 2022-1:2007 – Forward Error Correction for Real-Time Video/Audio Transport Over IP Networks
- ST 2022-2:2007 – Unidirectional Transport of Constant Bit Rate MPEG-2 Transport Streams on IP Networks
- ST 2022-3:2010 – Unidirectional Transport of Variable Bit Rate MPEG-2 Transport Streams on IP Networks
- ST 2022-4:2011 – Unidirectional Transport of Non-Piecewise Constant Variable Bit Rate MPEG-2 Streams on IP Networks
- ST 2022-5:2013 – Forward Error Correction for Transport of High Bit Rate Media Signals over IP Networks (HBRMT)
- ST 2022-6:2012 – Transport of High Bit Rate Media Signals over IP Networks (HBRMT)
- ST 2022-7:2013 – Seamless Protection Switching of SMPTE ST 2022 IP Datagrams
- ST 2022-8:2019 – SMPTE Standard – Professional Media Over Managed IP Networks: Timing of ST 2022-6 Streams in ST 2110-10 Systems
- ST 2110-10 – System architecture and synchronization. Synchronization is based on SMPTE 2059.
- ST 2110-20 – Uncompressed video transport, based on SMPTE 2022-6
- ST 2110-21 – Traffic shaping and network delivery timing
- ST 2110-30 – Audio transport, based on AES67
- ST 2110-31 – Transport of AES3 formatted audio
- ST 2110-40 – Transport of ancillary data
- SMPTE 2059-1 – Defines signal generation based on time information delivered by the IEEE 1588 protocol.
- SMPTE 2059-2 – Defines an operating profile for the IEEE protocol optimized to the needs of media synchronization.
Assuming that you can recover your data signal from the copper or fiber interconnect then the ONLY parameters that impacts the aural or visual quality of a digital signal are jitter, buffering and timing. Given sufficient buffering and appropriate control, loss of data due to various system delays can be completely negated so it will not even be considered.
Now remember, I am dealing with the very highest of quality A/V systems for audio and video in the recording, TV and film production industries all the way up to 4K and beyond. So whatever quality of audio and video those systems produce is the very best, NOTHING that you do at home is going to improve that quality, although it MAY maintain it.
So we are left with two issues in the digital domain that need to be dealt with; timing and jitter.
Timing
Lets start with timing as this is critical in high speed digital networks in order to ensure that all the data arives where it is required both when it is required and synchronously with other associated data streams, like audio, timecode, control data etc. There are a whole host of timing models and techniques that can ensure correct timing of audio and video data streams. All hardware is locked to a central clock and/or sub clocks that in turn are ultimately locked to a Global Positioning System (GPS) timing reference signal and use Network Time Protocol (NTP) and Precision Time Protocol (PTP) timing data to accurately lock all data sources and destinations together. These reference generators contain crystal and temperature controlled clocks that provide a range of stabilities based upon the systems requirements. So reliably clocking data through a system synchronously is well established. These external master and sub-master clock systems are NOT required for residential A/V use…..EVER! Why? Because you are not producing independent digital baseband or IP video, music and data streams that need to be locked together or sent to some remote facility and need to be synchronized to that destinations house clock.
Jitter
In residential digital A/V networks jitter is a function the technical performance of the digital clocks within the associated hardware like the player, AV Receiver and associated buffering. Hardware can be locked together using one as a master reference or the clock can be extracted from the incoming digital A/V signal. Either way and with the use of modern master clock techniques jitter in most systems is rarely an issue.
NOTE: All digital clocks create jitter to varying levels depending upon their design. Just as their stability and frequency accuracy is design, component and temperature related. Generally speaking, the more expensive the master digital clock the better all these critical performance parameters are.
In a residential system (DVD player, receiver, AV processor and display/projector) clock data is extracted/used in one of three ways:
- Using proprietary and dedicated data links like Denon Link (DL 1/2 &3) and dedicated clock links like Denon DL4 (for bitstreams over HDMI) and Pioneer’s PQLS (Precision Quartz Lock System).
- By locking the receiving clock to the incoming data stream. (HDMI, AES, Toslink)
- The received data is locally buffered, think any network streamed digital data. The “fullness” of the buffer is used locally to control the data clocking rate to keep the buffer from over or under flowing. The jitter of the re-clocked data is then just dependent upon the receivers master digital clock, as in both of the above cases.
For residential use the AV receiver/pre-amp or processor is considered to hold the master digital clock. Approach 1 is used to lock the transmitting (player) clock to the receiving (A/V Receivers) master clock. In the case of DL1/2 &3 the link carries separate locked audio data and the HDMI bitstream connection is unused. DL4 locks the players clock to the receivers master clock in order to reduce HDMI and hence bit stream jitter, similar to PQLS. The primary benefactor to all this jitter reduction is the audio. It is preferred to approach 2 as the clock data is more stable (less jitter) than extracting it from an incoming data stream. Option 3 is also commonly used to remove incoming jitter while at the same time prevent data loss due to system data rate variations or interruptions.
In my system I can clearly hear improvements to audio applying these techniques to audio (DL 1, 2, 3 & 4) but reductions in HDMI jitter (DL4) are not observable in any video. I have seen differences between HDMI and SDI video when both are driven from EXACTLY the same serial data source. The SDI video always looks less noisy, is sharper and has more saturated colors. I have my own theories as to why this happens but this is not the place to extol them. In any case, 4K SDI connections are not available for residential use like HD SDI connectivity was. So the improvement is moot. (Until 4K came along my system connectivity was HD SDI).
In all cases the jitter stability and performance of the players and receivers clocks are still the major governing factor on many audio performance improvements, whether they are locked or not.
A note here on jitter created by cabling. Firstly this is nonsense unless the environment is so high noise or the cables so close to the limit of their performance, that it impacts the edge recovery of the data. Data that is locally buffered (3 above) after error correction and clocked out under local control is not subject to this issue at all. Ethernet is a balanced system, and I cannot imagine any residential environment where the system electrical noise is so high or the cables so long as to impact the balanced differential data signal. If the data can be recovered, buffered and clocked, only the local master clock jitter counts towards audible issues.
What Hardware & Cabling Does The Professional Industry Use?
I’ll start of with two quotes from purveyors of these ‘magical’ home theater products. My response to each one is italicized at its end:
Quote 1 – The Switch
“Digital Music relies on perfectly timed data in a low noise environment to achieve maximum performance. Partially correct, but not relevant to low noise residential systems where there are no timing issues as all relevant data is contained within one connection. It is only relevant in professional production systems.
The limitation to the ultimate sound quality of digital music on an Ethernet network is the network itself and the components used. Partially correct, but it’s never an issue in any COTS system that I have seen that uses professional cabling. Its about the A/V hardware clocks jitter performance and buffering.
IT components are low cost commodity items and are effective for PC data and gaming etc. But they are simply inadequate for high quality digital music. Total rubbish – COTS switches are used throughout the ENTIRE professional A/V industry.
One critical component in all digital music systems is the dataswitch which handles both the music data and control.” Correct, but not all music systems require or uses one. Furthermore control data is often (and by preference) sent separately to video and audio data, its is not subject to high speed data and timing problems as it is VERY LOW SPEED and not timing critical (unless in a professional A/V production environment) . Also if the control data is embedded into the video and audio streams its extraction occurs within the receiving hardware and the switch is irrelevant to the data path.
Quote 2 – The Network Cable.
“As we’ve all learned over the years, a chain is only as strong as its weakest link and all things are important to achieving the end result – and it turns out that network cables are no exception! Total rubbish when using professional network cables. There are so many other areas of real concern like power and grounding.
All of these measures add up to as near to bit-perfect, ultra-low-jitter replay as possible. Total rubbish, you don’t need all those esoteric measures just a good COTS switch and professional cabling.
Music simply becomes more vivid and real sounding. Ultra-pure transparent music, packed with every bit of low-level detail, superior tonal accuracy, timing and dynamics.” Total rubbish – FULL STOP.
So what have we learnt?
- ‘Special’ CAT(x) network data cables are a total waste of money as there is no such requirement. Belden CAT (X) cabling such as 2413 (CAT6) and 10GX62F (CAT6A) will meet your every need.
- Off the shelf switches from companies like CISCO, NETGEAR, ARISTA etc. will work just fine in ANY home environment and NEVER impact you visual or aural performance.
- If COTS switches with their management layers can support all the above EXTREME professional standards why are they not good enough for the consumer? They are. It’s just that somebody wants to make a fast Buck, Pound or Euro of the unsuspecting. And that’s YOU.
NOTE 1: For home A/V use most good quality unmanaged 1 Gigabit switches will work just fine. You DO NOT need a managed switch unless you have a sophisticated, central whole house network switch, with devices like IP PTZ cameras (that require PoE), security systems and multiple high bandwidth A/V streaming devices attached to it.
NOTE 2: The above implications also apply to all copper and fiber (optical) interconnects carrying any digital data, to include: HDMI, AES, SDI video and of course USB. (I have tested a large range of both HDMI, Toslink and SDI cables within my own system and there is no visual or aural differences between them.)
If you use professional industry standard hardware (COTS switches) and cabling (E.G. Belden/Clark/Klotz) your data will remain intact and a perfect replica of the originating source. It is then up to your receiving master clocks performance. All this silver and gold plated nonsense with high performance switch clocks and analog power supplies has no relevant impact on the distribution of residential A/V digital data between two points.
As for cables and power supplies impacting the performance of analog audio and analog video, that can be a different story for another time.
Ask yourself this. If all these cables and ‘special’ hardware actually did what these manufacturers said it did, companies like mine would be clambering for it and using it every day. Why? Many of our A/V clients demand the very best that is available and are fully prepared to pay almost anything to get it. Funny how we don’t use it, because if we did we could really increase our margins and bottom line, but we don’t, why?
For more on cabling and these “magic” products see:
UK ASA Adjudication on The Chord Company Ltd for their cable claims – found guilty.
The English Electric – AKA Chord – 8Switch.
The Chord C-Stream CAT6 Digital Streaming cable.
The Russ Andrews Network Switch
The list just goes on and on!!
For a breath of fresh air see:
