Next-generation routers

While routing might not be as exciting as other tasks performed in today's high-tech digital facilities, the digital video routing switcher is typically the heart of a plant and often determines what types of workflows can be cost-effectively established. Add to this the successful integration of multiviewers, embedding/de-embedding capabilities and conversion tools, and you now have a heart on steroids. Not that this is necessarily a bad thing; it just requires careful consideration and planning.

Processing in the router: boon or bane?

When considering the size of today's routers, it is amazing to think that only a few years ago a 512 × 512 router was considered enormous for most facilities. Today's broadcast facilities, however, are distributing more content to more platforms and thus demanding more of their signal distribution infrastructure than ever before. Besides sending HD-SDI baseband video signals to a variety of destinations, there is an expanding need for additional outputs to handle a wide variety of signal types. The cabling requirements alone can easily overwhelm a facility. So how do you simplify this level of complex wiring?

With FPGA and other VLSI technologies advancing at a relatively high rate, broadcast manufacturing design engineers have an increasing number of gates to play with in ever smaller packages that consume less power. The result is the expanding migration of modular “glue ware” integrated inside the router. Initial forays into the world of modular router integration included simple line and frame sync capabilities. We now see the advent of embedders, de-embedders, signal format converters, scalers and multiviewers moving into the router as well.

Bringing all of these capabilities inside the router significantly reduces the complexity of plant wiring and offers significant savings in footprint as well as cable management and maintenance. It also adds versatility to the router by allowing users to select only the tools they need. Another advantage is that router control systems are becoming more powerful and can manage not only the router crosspoints but also internal tie lines and key parameters for the various modular functions now in the router or in external trays. There is, however, a catch.

Keep the noise down

As the amount and type of digital signal processing and signal speed increase in complexity, the noise generated and coupled into adjacent channels within a router can also increase. Today's newest routers address this issue by using the best practices of both RF and digital design. To keep spurious signals from affecting adjacent channels, the best designs use differential microstrip transmission lines that offer maximum common mode rejection, even at 3Gb/s, along with physical separation and symmetrical layouts. To keep unwanted signals from affecting signal performance, top-level routers use precision connections to “edge launch” the signals from the coax to the circuit board with minimal reflected energy. Another highly advantageous technique is to minimize the number of internal interconnections and long signal path lengths. Even with all of these techniques, additional internal processing and an overall increase in signals inside the router will inevitably increase the noise floor of the router.

Timing is everything

As we all learned in physics class, electromagnetic radiation (light) travels at a finite speed. In the world of advanced television systems, that can mean that small shifts in timing can make or break signal reliability. In addition, it takes a lot of processing time to perform all of the calculations needed for video conversion and manipulation. These processing times are significant when compared with 1080i and 1080p HD frame rates. Consider also that best practice requires television facilities to ensure that various signals arrive closely timed at specific destinations, such as video production mixers. If the router is expected to switch signals at the proper point during vertical interval, the signals also must be timed with the sync reference signal for the router. As a result, great care must be given to ensure that signals internal to the router are kept in time. This begins with bringing them into the router in time and requires that everything is kept in sync all the way to the output. This often means additional frame delays and more complex circuitry inside the router.

For audio embedding and de-embedding, timing is even more complicated. Routers that remove embedded audio from a serial digital video signal and then insert new audio take time to do their job. If embedded audio is to be transferred from SDI video signal A to SDI video signal B, it will take even more time to do this. If the two SDI signals are initially in time with each other, the second signal B must be delayed while the audio from signal A is received and inserted into signal B. Otherwise, the audio packet must be inserted in the next audio space in signal B.

The first scenario will cause the B SDI signal output to be delayed by a few microseconds. Repeated passes through processing may cause significant delay differences between processed signals and unprocessed signals. Frame synchronizers are often needed to solve these timing differences.

The second scenario will preserve the video timing but cause the audio samples to be delayed by one or more horizontal lines relative to the video signal. For normal audio, this will cause a delay difference of several audio sample times between processed and unprocessed audio signals. With care, this may not be a huge problem for AES audio unless delayed and undelayed audio signals from the same audio location are combined into a mix. However, Dolby E audio expects a specific relationship between the compressed audio packets and the video signal vertical interval. Audio-to-video delay errors of one or more lines can break Dolby E signals. Insertion of the Dolby E audio into a subsequent audio space will work if the Dolby E packets are lined up with the next video frame, but that means the audio will be one frame late, leading to the lip-sync problems prevalent in broadcast today.

The other timing issue in complex routers operating in the 1.5Gb/s and 3Gb/s range is jitter in the high-frequency digital signal. Jitter is the displacement or deviation of the amplitude, phase timing and pulse width of a received signal referenced to the signal's ideal location. The principle causes of jitter include electromagnetic noise or interference, adjacent channel cross talk, and temperature or voltage fluctuation inside of the router. This leads to a loss of signal integrity and detectability. Basically, noise in the router can dramatically increase jitter and reduce overall signal path reliability. The good news is that, with carefully engineered RF design in the router, this can be overcome.

While it is true that all of these timing issues occur in processing modules both external and internal to the router, their presence in the internal solutions are often far more difficult to troubleshoot, isolate and correct. The overall impact is that while we gain an advantage in footprint and cabling, we yield some of that advantage to timing risks.

Is it hot in here?

Another major issue with increasing the complexity of the router is increased heat buildup in the router. As we pass digital signals through the router, and especially when we go from 270Mb/s for SD to 1.5Gb/s and 3Gb/s for HD-SDI, the increase in the data rate translates directly into additional power consumption and causes the system temperature to increase. The primary components that drive up heat in a router are the output amplifiers, FPGAs and other signal processing chips that do the conversion and embedding/de-embedding work. The crosspoint gates and transistors, as well as the equalizer circuits and reclockers, also generate additional heat in the router. The bottom line is that the more features and circuits you add to the box, the more power you will consume, resulting in heat buildup.

Continue on next page

With good thermal management design, however, the best router frames can significantly mitigate this problem by ensuring that the air moving through the router is cooling those hot, power-hungry, monsters. This has an additional advantage in that if you keep that temperature down, chip performance becomes more stable (less jitter), and the chips won't wear out as quickly — especially for modern ultradeep submicron (90nm and less) processes. This lowers the overall cost of ownership. A good example of this type of design is a routing architecture that optimizes granularity (I/O) per card with overall component count and surface area. By optimizing the I/O per card, the total number of redundant or duplicated circuits in the router frame can be minimized, and you will have a more effective and unrestricted cooling surface area. This enables you to lower the amount of heat-generating electrical losses within the frame, minimizing heat buildup.

Doing your due diligence

When selecting your next router, ask the following questions:

  • Can I keep everything cool? With the additional heat generated in routers today, making sure that you can maintain adequate cooling is absolutely essential. This means carefully assessing the cooling capacity of your air-conditioning system as well as the airflow capacity and flow pattern to be sure that sufficient air is moving quickly enough to keep the router within specified thermal limits. Remember that power in equals power out, and all that heat needs to be dissipated or it will build up in your engine room. Adding the cost of more cooling capacity to the price of the integrated router solution is a good double check on determining need versus value.
  • Where do I save the most money? With today's tight budgets, reducing costs can often be the difference between success and failure. Making the right decision on cabling and complexity is crucial. Take the time to carefully evaluate the costs of your available space and cabling versus the cost of ownership of a new router with internal signal processing, not just the initial price of the router. It's not uncommon for the tried-and-true external solution to be cost-effective compared with the integrated products. If space and cabling cost are at a premium, use the best bang for the buck approach.
  • How much conversion do I really need? Just because you can do something doesn't necessarily mean you should. Having the right amount of conversion and other signal processing tools is always essential, regardless of whether they are in the router. However, remember that generating heat and noise with unneeded processing, particularly in the router, reduces reliability and increases operating costs without a corresponding return on investment.
  • Will I get the support I need to use and maintain the new router? Complex routers mean complex configurations and operational training. Be sure to select a manufacturer with excellent product support and training, as well as customer service. Don't skimp on the commissioning and training part. Factoring these costs and time into your selection and budget will save you frustration and embarrassment in the end.
  • What is the trade-off between reliability and performance versus built-in features? Knowing that more does not necessarily mean better, do your homework on product reliability versus built-in features. Remember, integrated solutions offer numerous advantages but also pose some risks in terms of reliability. Choose the solution that best fits your business and workflow model, regardless of the built-in features, and you'll sleep better at night.

Steve Dupaix is director of router and modular product management at Grass Valley.