The future of modular infrastructure

Crystal Vision workflow diagram
Philip Scofield, Crystal Vision
March 11th 2016 at 3:51PM : By

What place do modular components have in an increasingly IP-centric industry? A very important one, says Philip Scofield, managing director of Crystal Vision

From about the dawn of the industry, broadcast engineers have needed bits of electronics to interface between one unit and another. In the early days of analogue black and white it started with distribution amplifiers. Master clocks were probably also in there for synchronisation.

As television developed – from black and white to colour, from analogue to digital, from standard definition to HD – so the need for interface electronics became more pressing. And, as the power of electronics became greater, so it became practical to shrink the circuitry for these functions from a box to a single board.

Thus was the modular concept born. It is now widely understood by vendors, engineers and purchasers alike. A box with a single power supply (or probably two for redundancy) has slots for a number of plug-in boards, each of which provides a specific piece of functionality.

Equally, as electronics grew in capability, so we found tougher jobs for modular products. From distribution amplifiers we moved on to analogue-to-digital converters, which in turn needed digital distribution amplifiers. SDI allowed us to embed audio into a digital stream, so we needed audio embedders and de-embedders.

Everything about television is based on ideas established when electronics meant valves 

The latest generation has brought devices which would once be thought of as challenging applications to the modular world. So companies like Crystal Vision now provide very high quality chroma keying on a card, for example, or logo generation, or clean switching.

In the analogue world, and now in the digital world, there has always been a need for modular devices. They interface from one thing to another, or they convert one thing to another, or they distribute one thing to many.

Now, though, the suggestion is to move from traditional broadcast connectivity and platforms to an IP world. What happens to modular products now?

IT ideas

As a business, we recognised the move to IP and IT hardware several years ago. To prepare ourselves, we started recruiting engineers from the IT industry. Having spent my entire career in broadcasting I felt I knew how the industry worked, and they would enjoy learning from me.

But the first thing they asked was why is everything still compatible with black and white television? Why do we need a vertical blanking interval that was only there to allow a CRT spot to get back to the top of the picture?

The IT industry never shrinks from moving on when it feels it needs to. Software products rarely worry about supporting hardware more than a few years old. Everything about television is based on ideas established when electronics meant valves.

The second point they made was that the video industry seems to have suggestions instead of standards. That may be a bit harsh: we can standardise when we need to – SDI is pretty universal. But we do have a lot of standards and a sometimes vague approach to them.

As a long-term video engineer I spent a long time encouraging my engineers to understand the unique issues of the video industry, and that there were good reasons for everything that exists in a television studio. But just as they have finally understood that the television industry is unique, they have convinced me that we can no longer go on working this way.

We have to find ways of transforming the technology through the move to IP. And if that is going to work the way we want it to, with off the shelf products that plug and play, we have to look at new standards.


If you think for a moment of the web, and clicking on a link which will launch a video. We can be reasonably certain that, in a second or two, the video will start to play. Hidden from us, but absolutely vital, is a set of protocols and standards that ensure the right content, in the right resolution and aspect ratio, appears on the screen.

That is what we have to achieve in IP-centric video. We have to be able to set up a route and all this happens.

The requirements for modular products will continue, now, through the transition and on into the IP era

We are beginning to develop standards around this. The SMPTE 2022 family is a good place to start, but it is important to understand that this only defines the data content, not its control. Various manufacturers are proposing control systems, or software-defined networks, but the protocols are by no means universal. There is no reason why there should not be a common standard, but it needs agreement.

Until this agreement happens – if it happens – then there needs to be a set of buffer applications which take the data and control from one environment and turn it into another.

It seems, then, that we might have a requirement for another set of modular devices here, as data translators. Certainly, during the transition to IP, yet another class of modular devices will be called for, which interface between SDI and IP, devices Americans call “on-ramps” and “off-ramps”.

The future for modular

I speak regularly to a large number of broadcasters, and their view is clear. They want to end up with best-of-breed, vendor independent technical architectures, just as they have now, but capable of working over IP. 10GB ethernet, and soon 40Gb ethernet, will provide a more flexible, more cost-effective connectivity than SDI. They recognise that 4K, and maybe other Ultra HD elements, will become more practical and more in demand.

There will almost certainly be changes to SMPTE 2022 in the near future (one major manufacturer in particular is pushing very hard), and no doubt in due course SMPTE 2022 will be replaced by something better.

We have gotten used to keeping audio and video together and in sync through embedding, but RTP and other internet standards allow audio, video and metadata to be separate streams, recombined and timed only when necessary. And on top of all these infrastructure-level functions, few of the traditional modular requirements have gone away. Channels still need a logo burnt in at the point of playout. Productions still need clean green screening. Audio channels need to be embedded, de-embedded, shuffled and combined. And so on.

Finally, remember that this is very specific functionality that is only appropriate to broadcasters. No-one else in the IT industry cares about this, so it is down to specialist vendors to design, build and deliver it. That is true whether it is specialist bespoke hardware, clever software running on FPGAs on a standard board, or indeed pure software running on a standard PC.

The requirements for modular products will continue, now, through the transition and on into the IP era. That is because the need for point functionality, to do something or to get from one environment to another, will always be with us. In broadcasting, we need to deal with live, realtime streams so we cannot stop what we are doing to route a signal through some other process.

It follows, then, that the smart developer of modular products will have a plan for the future which also allows a transition. Traditional modular frames will move towards a system which supports both SDI and IP. And ultimately the modules will be implemented purely in software, to run in a virtualised environment in the data centre that is the new broadcast centre.

There will always be a demand for live television, and that means there will also be a need for flexible architectures which depend on the glue of the infrastructure, and which can be configured as well as operate in real time. Only broadcast vendors understand the need and the construction of these modules, and the ones who are smart enough to track the move to IP, software and virtualisation will continue to find a role.