Building A DMX-over-Ethernet System
Ethernet has been creeping into lighting systems for some time now. Not only is it being used to simplify multiple runs of DMX, but additional tasks such as tracking backup, designer’s remotes and media server integration are all utilising the ubiquitous blue string. However, when trying to integrate these Ethernet interfaces into a system that has been cabled for the more familiar DMX, the system doesn’t get easier, it gets harder. In this article, we’re going to look at how to avoid this and keep it easy. We’re going to examine why Ethernet is becoming the preferred data medium for all lighting control data and the various DMX-over-Ethernet protocols already in the wild. Once we’ve covered these topics, we’re going to bring it all together and look at how we can build a data distribution system that integrates DMX and Ethernet, scales simply between small and large shows and, most importantly, remains simpler to use than the traditional DMX systems we are all so comfortable working with.
What’s Wrong With DMX?
Before we leap into building a network, it’s important to understand why we’re changing it at all. What’s wrong with DMX? Why do we need to change it? These are important questions, but we have some good answers too. The most widely known limitation of DMX is its 512 channel limit. As we no doubt all know, DMX works with a limited bunch of “slots.” This wasn’t a problem when we only had dimmers using up one slot each, but since the introduction of moving lights and media servers, these slots get gobbled up extremely quickly, requiring multiple DMX runs (universes) from the console – each of which needs to be carefully tracked and noted so they don’t get mixed up. With Ethernet, there’s no such restriction. Multiple universes can be sent down the same wire, with the bandwidth to handle hundreds at once. There’s no need to carefully track where each universe cable run goes, as we simply address our transmitting and receiving devices to the correct universe, without needing to worry about the physical infrastructure in-between (see diagram 1) – just like we do with DMX addresses now.
While the removal of the 512 channel limit is an important benefit of an Ethernet system, it certainly isn’t the only one. Thanks to the layered structure of the TCP/IP protocol suite that makes up the transmission standard used on Ethernet networks, we can send multiple, unrelated pieces data down the same network. Now we can run tracking backup information, MIDI-over-Ethernet (perhaps using the ACN protocol), remote video information, firmware updates, RDM (Remote Device Management) information, media server content and much more down the one network, without any of the costly, time-consuming cable re-arrangement that is normally required when a rig’s configuration is changed (see diagram 2).
A significant benefit of Ethernet is its “star-topology” wiring layout. While DMX uses a “daisy-chaining” method, the star-topology of Ethernet leads to a much more robust network. The lack of daisy-chaining support may seem like a disadvantage at first, but the extra resilience of a network utilising a star-topology means a fault in a device or cable will generally have a much smaller impact over the whole network and make it easier to trace. This coupled with the lower cost of cable (offsetting the extra length required) will make a network utilising Ethernet a safer, more robust system than anything built using the DMX protocol. Not only is it more reliable, but it is more flexible. The lack of “inputs” and “outputs” in Ethernet means you can plug any device (be it a console or fixture) into any port without needing to worry about the direction of travel. All devices will automatically communicate with each other as required thanks to the flexible TCP/IP protocol suite. Beware of systems that encourage the use of “Ethernet uplink” ports. These sound like a good idea, as they allow a daisy-chaining-like approach, and reduce the amount of cable needing to be run. However, there are a number of significant technical issues with implementing a system like this in an Ethernet system (unlike in a DMX system where this is generally not a problem). For starters, the Ethernet protocol specifies a set amount of “hops” allowed between any two devices. Use of even just a few “uplink” ports will exceed this limit and lead to latency problems and tricky-to-trace communication errors. Additionally, Power-over-Ethernet is generally not available in these “uplink” ports, as the power available on PoE is fairly limited, and again, just a few devices linked together would exceed the available power for that line. A much better solution is to run a second line, or put a cheap, small Ethernet switch in place to distribute out to each node in the traditional star pattern (see diagram 3).
Probably the most under-estimated advantage of Ethernet is its ability to leverage the wide range of technologies developed by the IT industry. Not only do we get cheap (yet highly reliable) switches and cables to use, there’s a host of sub-protocols supported by all but the cheapest of switches, many of which can be useful in an installation. An example of this would be Spanning Tree Protocol (IEEE 802.1D) which allows redundant links between switches such that should one go down for whatever reason, the other can take over seamlessly. Another useful technology is Virtual LANs, or VLANs (IEEE 802.1Q) which allows multiple venues all on the one network to segment the network into discrete virtual networks, guaranteeing no inadvertent control of one room from another. Streaming ACN (which we will talk about later in this article) makes use of the multicasting protocol to intelligently route universes of DMX only where it’s required, with no manual reconfiguration. All of these features (and many more) come in commonly available switches that are extremely reliable and often available much cheaper than the largely featureless DMX splitters we use today.
So, that provides an in-depth look at why Ethernet is better, and just how much it can simplify so much of the system. While there’s already much more than just DMX data that we may wish to transmit across the network, it’s probably the most important, so let’s take a quick look at some of the more popular DMX protocols out there. Just as in the early days of analogue (and digital) control of lighting, there is not yet universal agreement between manufacturers as to which DMX-over-Ethernet protocol to use. Some protocols are fairly open and used by many manufacturers, while others remain very closed and are typically only used by the manufacturer that developed them. There are probably five worth mentioning here: ArtNet, Pathport, ShowNet, ETC Net2 and Streaming ACN.
ArtNet is a good one to start with. It’s a simple protocol designed purely for encapsulating DMX data within Ethernet nice and simply. ArtNet has done a fantastic job in introducing the lighting world to DMX-over-Ethernet in a cheap, approachable way, and this is probably why it is the most widely used DMX-over-Ethernet protocol in use today. However, there are a few little rough edges to it that may limit its long-term usefulness. The numbering of universes from 0 rather than 1 is not very user-friendly, and can lead to confusion and incompatibility between manufacturers. Similarly, the lack of support for multicasting can make its transmission inefficient in large events utilising many universes. For these, and a few other reasons, ArtNet’s future is in doubt against the more robustly defined Streaming ACN protocol.
Pathport, ShowNet and ETC Net2 are protocols designed by Pathway, Strand and ETC respectively. Their penetration is not as wide-spread as ArtNet, but they are used by a few consoles and devices to make their inclusion in this list worthwhile. The Pathport protocol is supported by both MA’s GrandMA and Jands’ Vista for instance, while Strand’s ShowNet is the only protocol available in their 300 and 500 series desk, and the protocol of choice in their newer Palette consoles. ETC Net2 is the protocol used for ETC’s previous generation of consoles. While ETC has abandoned Net2 in favour of ACN in its latest Net3 products, Net2 is still commonly found in their older consoles, such as the Obsession and Expression range. If you wish your installation to support a wide range of devices, support for these three protocols is a good idea.
Streaming ACN is the newest kid on the block, having only recently been released. Its late arrival is due to rigorous defining from the ESTA-run Technical Standards Program. It is already the protocol of choice for ETC’s latest generation of consoles: Eos, Ion, Congo and Congo Junior under their “Net3” moniker. Thanks to careful design from the ESTA panel and the breadth of experience and knowledge from many of the world’s top lighting manufacturers sitting on the panel, Streaming ACN is probably the best DMX-over-Ethernet protocol available. Not only does it avoid some of the pitfalls of its predecessors, but it also adds some neat new features that extend the functionality allowed in DMX: a nice “preview” feature allowing your visualisation software to use a different stream of DMX than that is being used by the dimmers and fixtures (useful when in “blind” mode or for previewing the next cue), embedded priority information (useful for intelligently merging multiple sources without all the messing about required today) and an “end-of-sequence” function (allows devices to tell the difference between a cleanly-shutting down controller and one that has crashed or shut down unexpectedly).
Designing an Ethernet Lighting Network
Although lighting networks may one day be entirely Ethernet without any DMX wiring, that’s still many years away, so in the meantime we need a simple, convenient way to convert between DMX and Ethernet without any fuss. Many manufacturers are producing DMX-to-Ethernet and Ethernet-to-DMX nodes at various price points, and it can at first seem that the feature-set of these various nodes is equivalent. However, there are often some important things to watch for to ensure your network works simply, and with as wide a range of equipment as possible (both now and in the future). The most important thing to consider is the range of DMX-over-Ethernet protocols supported. While most support ArtNet (currently the most popular protocol), it’s worth remembering that there’s still a significant amount of equipment out there that does not support ArtNet, or has only added ArtNet as an after-thought. Also, as we discussed previously, there are some rough edges to ArtNet that may limit its long-term future, and there is a distinct possibility that Streaming ACN will become the preferred protocol driving lighting equipment: ETC have already done this with their range of products working under the “ETC Net3” label. Pathport nodes are unique in the large range of protocols they support. ArtNet, Pathport, ShowNet, ETC Net2 and Streaming ACN (ETC Net3) are all supported by Pathway’s Pathport nodes, meaning your choice of console is never limited by your DMX-over-Ethernet nodes. With a Pathport node system, there is no widely used console with an Ethernet connection that cannot be supported directly: allowing you to choose or tour consoles as required without needing to change your physical configuration at all (simply plug in your console and go).
Another important feature to look for is the range of merging features available. While two may seem enough, that’s only going to handle a simple master/backup setup. The need for more could quickly arise from many other setups: a simple console in prompt corner, suitable for a teacher or stage manager to operate in simple shows (allowing the venue to cut labour costs for shows not requiring a dedicated lighting operator). Alternatively, you might require a specific practical light to be operated by someone who has a clear line of sight from the wings rather than the control booth. With the aid of a system that can handle merging a number of universes simultaneously, you’ll be able to accommodate these scenarios (and many others) with no fuss and no extra cabling. Once again, the Pathport nodes have this covered. With support to merge up to 8 DMX universes simultaneously, even the most complex setups can be quickly and simply accommodated. Extra configurability is allowed for with a comprehensive soft-patch that can be configured for each node, allowing some consoles to have access to just a few dimmers, while giving others full control. The merging can be defined on a HTP- and priority-basis so you can specify exactly how much control each controller gets.
Once you have chosen an appropriate DMX <-> Ethernet node, it’s important to consider your placement. It can be very tempting to approach as you would a traditional setup: place one node in every location that might need one. However, this can be both expensive, and can lead to a system that will not easily adapt to a distribution system that becomes more predominantly Ethernet based. Another common approach is to install a traditional DMX system with an Ethernet network alongside it for future use. While this does cover the bases, it does present usability problems as the user tries to integrate their DMX and Ethernet devices alongside each other. A system like this would often rely on only one or two converter nodes with a very clear division between the DMX-side and the Ethernet-side of the network. This makes it both difficult to build and extend on, as well as limiting all the extra usability features an Ethernet based system produces.
The best approach for DMX and Ethernet integration is with infrastructure that is entirely Ethernet based. With a backbone that is Ethernet based, you can then place DMX-to-Ethernet nodes where required that convert the DMX-over-Ethernet signal to legacy DMX for only the last couple of meters to the lights. It’s generally important to feed more than one run of Ethernet to a specific location so that you not only have one for your DMX-to-Ethernet node, but also for native Ethernet devices (such as consoles, media servers, configuration devices and future devices) to plug directly in. While wall-mount nodes in commonly-used locations is generally a good idea, consider portable nodes for some of the less common locations. By using portable nodes that plug into Ethernet points when needed, you can save costs by keeping the number of nodes down. Some nodes (including the Pathport series) allow their ports to be configured as either output or input, regardless of the gender of the physical port. This allows you to turn a node typically used for output to become an input node with no more hardware except a simple gender-changer adaptor. Once again, this saves costs as you don’t need to worry about putting in nodes everywhere you need them. Instead, you put in cheap Ethercon points everywhere, and nodes only where you’ll often need them, leaving those unusual setups for portable nodes and gender-changed wall-mount nodes.
The Last Word
In this somewhat lengthy article, we looked at a number of important points. We looked at why upgrading to Ethernet was worth it at all: the support for more fixtures, a consistent, predictable structure in both small and large networks, that DMX-over-Ethernet signals can be shared with any other signals on the same network, a much better and more robust star-topology network layout and the availability of mass-produced IT switches cables which are not only cheaper and more reliable, but come with a bunch of extra features which can be just as useful for lighting networks as they are for the office networks for which they were designed. Once we’d covered why Ethernet was a necessary upgrade, we looked at the various DMX-over-Ethernet protocols that already exist and their various strengths and weaknesses. Of particular interest was ArtNet and Streaming ACN, with ArtNet currently being the incumbent DMX-over-Ethernet favourite, but under significant threat from the new Streaming ACN protocol which smooths a few edges, adds a number of significant features and already has the support of a number of significant lighting manufacturers, most notably ETC, who have used it as their preferred protocol in their latest range of consoles. Finally, we looked at building a network that can support DMX and DMX-over-Ethernet with equal ease, and no headaches. We raised some important considerations in node choice: the need to support as wide a range of DMX-over-Ethernet protocols as possible, merging support that not only covered the simple cases, but the ones that are a little more left-field as well with equal ease. We mentioned the Pathport range of nodes which has the widest range of supported DMX-over-Ethernet protocols out of any node on the market, and also supports merging for up to eight universes on an HTP- or priority-based basis; allowing plenty of room for growth. We also looked at some recommended cabling arrangements: running two or more Ethernet runs to a single location to allow not only for DMX-to-Ethernet nodes, but future Ethernet-only devices too, avoiding a reliance on integrated “uplink” ports that could cause troublesome network problems. Lastly, we looked at how costs can be reduced on smaller or start-up installations that might not have the budget for one or two dozen nodes in one hit: include plenty of the cheap Ethernet points, and rely more heavily on portable nodes and gender-changer adaptors to provide the DMX points you require.
While it may be tempting to include the same old DMX-based system with some disconnected Ethernet cabling for “future use” it is important to remember that systems such as these will no doubt cause usability issues and will fail to scale well with the Ethernet revolution that is so nearly here. By following some of the tips in this guide and looking to the future, your system can and will scale and adjust effortlessly as your venue’s equipment becomes more and more Ethernet based.
Lighting Applications Engineer – Jands Pty Ltd