New life for Co-Ax

Cat X cable is very much the golden child of the AV / IT convergence at the moment. However not so long ago, there was a quaint little technology much beloved by the AV industry called co-ax. In this world of RJ-45 and Cat5e, is there still a place for it? Rob Moodey of Matrox Graphics thinks there is, here’s why.

Anyone who talks about the cables going between a video player and screen will probably voice three themes – a short VGA/DVI cable, a CAT5 extender system or optical fibre. We all know that VGA/DVI doesn’t go far and that fibre is still comparatively expensive – but there is a copper-based cabling option more familiar to AV folks than CAT5 – co-axial cable.
If you talk to an IT engineer about co-axial cable they will wax lyrical about how the 10BaseT thin Ethernet used to be wired with Co-ax. If you let them, they’ll talk about how moving to ’duplex’ over twisted pair cables eliminated the data ‘collisions’ and enabled faster networks. Co-ax to them is a ‘has been’ that faded away before ‘convergence’ was a buzzword. But co-ax is back – and bidding for the IT projects that sit firmly within reach of the AV community – Digital Signage.

If co-ax and CAT5 are both ‘possibles’ for a project - how do we know when to use each? – read on

Cables: The choice –there are four criteria often used when choosing a cable for an installation
 Ease of termination (speed, convenience etc)
 Electrical performance (attenuation, insertion loss, phase linearity etc)
 Ease of handling (flexibility, ruggedness, longevity etc)
 Price (not just per metre on the reel, but installed and commissioned)

Different people will choose cables for a signage project according to a different mix of the above criteria
A wiring ‘operative’ could choose mainly on ease of termination – with little interest in the system performance requirements – perhaps even a belief that all cables are ‘the same’. A contractor might be more swayed by speed of deployment (how many can I do per day to reduce the bid price for the contract?). A system engineer needs the performance (the system is supposed to do a job – and that needs performance). This article concentrates on the ‘performance’ aspect – since this is behind the use of co-ax. It underlines what video people have known for a long while – co-ax is best for signal integrity.

The specification of cables is encapsulated in data sheets with many parameters. We can distil the effect of these parameters to four things.
 Dielectric loss – a result of the electrical characteristic of the cable construction.
 Skin effect – the tendency of higher frequencies to travel closer to the surface of the conductor
 Cross-talk between pairs of signal paths
 Conversion of differential mode signals to common mode signals
The first two effects are predictable (and can to some extent be allowed/compensated for) dielectric loss is a 1/f relationship and skin effect is a 1/( √f) effect. The second two are not such that we can eliminate them as they happen – these are addressed at system design stage.

Bandwidth is the future As nBaseT networks have evolved, the Unshielded Twisted Pair (UTP) construction of CATx cables has given way to progressively more shielding (in both cable and connector) to reduce the effects of crosstalk that plagued the introduction of higher data rates. Inevitably this has reflected in cost. Interestingly, the TIA/EIA specs that define the cable also define the network topology and connectors.

A building that was wired with a now obsolete CAT5 cabling might support a 100Mb/s Ethernet but not 1000Base-T . Cat6a cables provide performance up to 500MHz – more than double CAT5e (the current definition of CAT5 cabling).

So don’t take it as read that a building that already has “CAT5” wiring needs nothing else. Bandwidth is the future (Doesn’t everyone want broadband that they didn’t a couple of years ago?). And the CAT5 that is ‘installed’ may not be right even for the computer network! There may well be a ‘re-pulling’ project in the offing!

The burgeoning computer networks community speaks of ever higher bandwidths to eliminate bottlenecks – Gigabit Ethernet, 10Gbit Ethernet and so on – but whilst talk is cheap, some of the components needed for these are not. And the distances achieved with electrical connections are shrinking – hence the presence of so much optical fibre in this sector.

And we see the same when we use “CAT5 extenders” – the achievable distance for a given image quality falls with pixel count. More pixels in a given time is the same as higher frequency.

So what does co-ax bring (back) to the party? It has been used as the staple transmission medium by the video community since forever. PAL, RGB, YUV, 270Mb/s SDI, 360Mb/s wide SDI, 1.5Gb/s HD SDI and now 3Gb/s are all moved successfully over hundreds of metres of co-ax within facilities. In particular the interest here is that they are currently moving digital signals – signals that are intrinsically more robust – over point-to-point links. And remember that modern studios are knee-deep in IT infrastructure –which they use for moving files (datacoms!) – and yet they use co-ax to move pictures.

What’s the difference between their picture data and their network data? It’s all ones and zeros but the situation is different – a datacoms network has a protocol that enables repeat transmissions – if the receiver can’t make sense of what it receives, or detects uncorrectable errors, it asks for the data to be sent again. And if the network ‘gets busy’ everything slows down. In our ‘final hop’ to the screen scenario, that is not an option – any delay (congestion or retransmission) causes dropped frames, picture-freezes and stutters. Errors will be shown - not corrected. The picture playout is real-time – unlike file transfer. (Note there is a difference between ‘real-time’ and ‘live’). In that respect, the data network’s need for bandwidth is not so rigorous as ours. So don’t run away with the idea that CATx cable is dead – this format will continue to evolve and provide good service to the datacoms industry by providing appropriate bandwidth.

So let’s move back into the AV world and think about Digital Signage.

Clearly cross-talk doesn’t occur if you are only taking one cable from one player to one screen. If the distance is within the spec of the CAT5 extender this is a ‘home run’ for CAT5.

When there is a need for several screens of different content, a popular signage solution is to have a single player device driving multiple screens – from a multi-headed VGA graphics card with multiple CAT5 extensions. This ‘single player’ option is popular because it reduces the number of player licences required. Clearly this will mean several adjacent cables (for at least part of the run) risking cross talk. There will also be some ‘correction’ for the frequency dependent factors mentioned earlier. The stability of the correction is something that is rarely considered – but in an analogue world I’d suggest that everything changes with time.

Another approach is to use the technology embodied in the Matrox Veos – this enables a single player to output a very large number of pixels as a single physical output. That single player can provide enough pixels to drive 2 or 3 screens, each at their native resolution, with different content. The Veos then streams those pixels, representing real-time images, as a serial data stream. The catch? More pixels means higher bandwidth – and that, as we saw earlier, means co-ax rather than CAT5. And we are talking here of ‘bandwidth’ not ‘broadcast quality pictures’ - don’t be distracted!

So why would we want to use a co-ax system like this?
 As a serial stream there is no cross-talk between the three channels.
 There is no skew or EQ adjustment to be made – everything travels along exactly the same signal path. (And even if someone fiddles in the trunking, the geometry of the co-ax will not change, unlike with CAT5 where a skew re-adjust might be needed)
 There can be no common mode conversion – there is only one signal path
 Digital signals can be regenerated and passed on – without the accumulation of analogue distortions that occur in an analogue ‘daisy chain’.
 Only one cable to pull instead of three

Another issue is ‘pin count’. This is a frequently used indicator of potential unreliability – the more connector pins you have, the greater the chance of failure. Cables of course have connectors at both ends. The figure shows a setup for three screens of differing content – pincount speaks for itself.

Why not compress? The classic solution to ‘the pipe isn’t big enough’ is to reduce the amount of ‘stuff’ going through – by compressing the signal. But as the challenge increases, you need to lose more and more of the original (who hasn’t seen digital compression artefacts on their TV as the satellite or cable operator tries to squeeze an extra channel into their distribution network?) There comes a time when increasing the size of the pipe is needed.

The proliferation of IT into ‘our’ domain is something that AV folks often bemoan. We’ve probably all heard:

“RJ45s are easy to fit but break if you tread on them, snag as you move them, and soon degrade if you make and break them several times – why do we use them instead of ‘proper’ connectors (thinking of BNC, RCA etc) for AV signals?”
The answer, clearly, is that they have their place. The same can be said for the cable that runs between those connectors. As AV people we all know that there is more than one grade of co-ax and not surprisingly there is more than one sort of data cable – some are old, some are cheap and some are good.
We’ve probably also heard, read or even thought: “it’s a shame that computer applications can’t run over co-ax”.

Well, as we’ve seen – they can. For Digital Signage applications, the run between player and screen can be co-ax without it being an RF cable to a tuner. There are significant benefits in terms of either or both of image fidelity and distance between player and screen. And this ‘leg’ of a Signage system is firmly in the AV area of influence and expertise.
If there is a reason why convergence didn’t happen at the rate that the pundits initially anticipated a few years back, perhaps it was the realisation by IT folks that some 1s and 0s were more important than others. And if there are any ones and zeros that get ‘victimised’ it’s those that represent AV signals. The demands are increasing with the advent of higher definition pictures – running in stark opposition to the network people who want to reduce bandwidth consumption by compressing the video. And bandwidth limitations happen in post-player elements of systems too.

So, AV folks, play to your strength in signage projects – you have a powerful weapon, co-ax, in your armoury. But it’s only powerful if you take it out of the armoury and use it!

Article Categories