This year we took the to the in Amsterdam to launch it to the industry. The technology was well recieved, but it's absolutely cutting edge stuff, and getting it ready for a high performance demonstration at a major show such as IBC is far from straight forward. In this post I'll explain why the project started, state of progress at the start of the show planning process, and the amazing work our team acheived to get it into action last month in Amsterdam.
So, why are we doing this project at all? Radio spectrum is a valuable and finite commodity- there is simply only so much radio frequency space available. These days many more organisations and companies want a share of that spectrum. Once upon a time TV and Radio broadcasting and radio links for video/audio contribution links were rare and had relatively uncontested use of the airwaves in key frequencies. Nowadays though, with the rapid growth of mobile digital services, that same spectrum is under high demand, and governments around the world are carefully managing the licensing of it for different applications. As the popularity of high quality radio links for program making has increased, so too the move to High Definition (HD) video has also created a demand for yet more data and hence more pressure still on the spectrum.±«Óãtv R&D recognised that something needed to be done and three years ago we created the Advanced RF for HD Radio Cameras project to look at RF (radio frequency) techniques that could be used to make radio cameras more spectrally efficient. Our aim has been to use techniques such as MIMO (Multiple In Multiple Out) to create a system that uses half the spectrum compared with current commercial systems, both standard definition (SD) and high definition (HD).
In the spring I went to (Broadcast Video Expo) and was surprised to see someone selling a dual (Digital Video Broadcasting- Terrestrial, the standard Freeview uses) HD radio camera, using twice the bandwidth of SD radio cameras. This got me thinking and after a discussion with the rest of the team we decided that it would be very worthwhile showing our system at (International Broadcasting Convention) in Amsterdam.
At this stage our project had advanced to the stage where our transmitter comprised of what we call the 'wedding cake', which is a rather large unit about 300mm x 300mm x 200mm plus an antenna pole. This large transmitter was based on a DSP board that we had built with at least four purposes in mind. These four purposes were; to be the main board in our receiver, a platform to test advanced RF algorithms, a channel emulator and a channel sounder, so it had a lot more functionality than would be required for a transmitter back. The whole transmitter was built up with layers like a wedding cake, so that's where the nickname came from.
We had a number of things we needed to do before IBC; first and foremost we needed a more compact transmitter unit, which would comprise a video coder board, a modulator board, an up-converter board, antennas, a case and possibly some kind of display. We had already obtained a compact H.264 coder from NEC (based on the very capable ) so we set about designing the other boards. We knew time was short and that we'd only have time to produce one iteration of the boards, so they had to work first time. However that didn't stop us taking a few calculated risks along the way.
Previously we've always used IQ (In-phase, Quadrature) upconversion techniques to translate our signal directly from the modulator up to its final output frequency. Whilst this makes output filtering easy it does mean that there is some degradation in the transmitted signal, but this time we would use very fast Digital to Analogue Coverters (DACs) to provide an output at an intermediate frequency and then IQ up-convert from there. The risk in this case was that we hadn't used these specific DACs before. We were also using a brand new lower power Feild Programmable Gate Array (FPGA) device for the modulator- so new in fact that we could only obtain pre-production engineering samples, which also added an element of risk.
During the summer our CAD group and the team worked very hard putting in long days and working at weekends and we were running lots of tasks in parallel. We built a new larger receiver crate (wired up by one of lab technicians) so we could add a Xilinx development card which gave us the resources to implement an advanced Forward Error Correction (FEC) technique called Low Density Parity Check (LDPC) code which theoretically might give us as much as 5dB more signal ruggedness.
As soon as we had come up with a board design that was ready to send out to manufacture CAD started work on the design of a chassis to mount the boards on and a 3D printed case to protect them and make the transmitter look less like a R&D prototype. Antenna mounts and antenna radomes also needed to be designed and made. Meanwhile team members worked on getting our DSP board and the Xilinx development board we were using as the modulator to talk to each other using the fast Xilinx Rocket IO. The antenna mounts and chassis were made by our ±«Óãtv colleagues in Technical Support Services at Brock House. In parallel our Prototyping Support Engineer was kitting for boards, getting quotes to get the boards made, arranging to get the printed box made and the antenna covers made.
The new camera back prototype showing its workings - multiple components from different sources all integrated onto the chassis.
Bear in mind too that all of this work was racing for a deadline at a major international exhibition, so among other things hand-outs and posters were written and then designed, space booked on the (EBU) stand and we came up with a name for our system. Seeing as our system uses half the spectrum of conventional radio cameras it was decided that our system would be called halfRF.
When the cards came back from manufacture the fun really began. It was a very exciting time. Testing the basic functionality of the cards went well; things like the power supplies were fine. The up-converter card required some tweaks of the RF circuitry, but nothing major. However, the modulator card wouldn't recognise the coded video input. There was lots of head scratching and discussions with the suppliers, Xilinx and eventually it was found that the documentation from Xilinx was based on engineering samples and we had initial engineering samples and the input circuitry was different on the initial samples- such are the potential pitfalls of using the cuting edge of the cuting edge! Once that was discovered progress in getting the modulator card running was swift.
Bench tests of the transmitter looked very good- the quality of the output was close to that of a main station broadcast transmitter. We did some walk around tests at the South Lab at Centre House without LDPC and the performance looked good, if we could get the LDPC running for IBC the system would be stunning. Then we noticed occasional "splats" in our received spectrum. Initially we thought they were incoming interference, but investigations revealed our transmitter itself was producing these splats. That had to be fixed before we did anything else. On the Friday before IBC the splats were fixed just minutes before our senior divisional management board was due to visit the lab, so we did an impromptu demonstration to them after Stagebox had done theirs. Then we packed everything into crates to be shipped to Amsterdam for IBC.
During the summer we had arranged to be included as a demonstration as part of the EBU village at IBC. The EBU are the industry body for broadcasters in Europe, and very interested indeed in supporting technologies that can improve the use of scarce spectrum for broadcasting purposes. We've shown new technologies alongside them before, and this was a perfect alignment of their and our objectives. The diagrams of the stand which we'd been sent implied that we were on part of the stand facing into the hall. However, when we arrived at IBC and saw the stand we found that we were actually facing into the corner of the hall. At this point I lost my nerve slightly. We were planning to mount the directional recieving antennas quite high up on the stand out of the way so that they didn't cause too much visual impact and also would get a clearer view of the hall, so that meant that we wouldn't be able to move them. I knew we had to cover our demo area so I faced the antennas into the corner away from the rest of the hall. Sure enough, when we tried the camera it worked perfectly in the demo area. Then we ranged further out into the rest of the hall expecting slightly patchy coverage, but in actuality found we had seamless coverage of the whole hall, which meant our signal was being bounced off the roof and the walls of the hall- the advanced signal decoder ensured perfect reception in even these challenging conditions!
We wanted to highlight that we were providing 20Mb/s in a 5MHz channel so as part of our demonstration we had a spectrum analyser showing the received spectrum from one of the antennas. It was noticed the spectrum appeared to be stable and surprising free of multipath. One of the clever things we'd implemented in the transmitter was a second pair of transmit antennas so that we could utilise part of the forthcoming (Digital Video Broadcasting- Next Generation Handheld) standard. An option in the DVB-NGH standard is to transmit horizontally and vertically polarised MIMO terrestrially and then add a satellite with left and right hand polarised signals to provide transmit diversity and therefore more rugged reception on portable devices. Many things in the RF world are reciprocal so we turned this around somewhat by having a portable transmitter with diversity to provide rugged reception on a fixed device. What this means is that more paths from the transmitter to the receive antennas are utilised and in order to have a fade in the received signal more paths would have to cancel, so the signal looks and is more stable.
The spectrum analyser demonstrating the performance of the halfRF system at IBC
We demonstrated the system to many visitors, broadcasters and equipment manufacturers alike. All visitors were very impressed; these included the president of large Japanese electronics company, the chief of a Brazilian TV company and Business & Technology directors from a multinational company. A number of companies expressed an interest in licencing our technology.
One particularly notable visitor was the Chairman of the DVB technical module, who was very impressed that we'd managed to implement part (albeit a small part) of the DVB-NGH standard even before the standard was even finished.
After a very succesful IBC we have the kit back home and we plan to iron out the last few bugs so that we can carry on and research some ideas that we have to make the system even better. I imagine we'll be doing lots of demos too, and before too long this groundbreaking work will make it into a TV production you'll see on your screens.