R outers have been present in the broadcast industry for a long - PDF document

Routing demystified By Renaud Lavoie R outers have been present in the broadcast industry for a long time; I was in school when the first re- dundant router was introduced. The reason for redundancy is quite simple: Viewers do not like to see messages like “Com- ing soon” or “Technical difficulties, please wait.” Imagine the stress in the studio to re- solve these issues. These days, critical equip- ment on the live path has built-in redundan- cy, or at least an automatic changeover (ACO) implemented to make sure that viewers do not see the screen in Figure 1 too often. There are two main categories of routers: video and audio. Nowadays, these two types can be combined to form one signal router in- side the studio. Another major trend is the IP router and agnostic (universal) router, thanks to SFPs and emSFPs, capable of transporting any type of signals over any media (coaxial, fiber, Ethernet, etc.). The behavior of the traditional broadcast Shown here are input and router is simple. It takes numerous inputs and the switch line. In asynchronous routers, the output cards using SFPs and emSFPs that enable the reroutes, or connects, these inputs to numer- router has the ability to “frame synchronize” routers to handle formats ous output ports. This is the basic function the asynchronous inputs on the input cards such as HDMI, NTSC/ of the video and audio router. The second of the router. This means that all the SDI sig- PAL, SDI and 3G-SDI. important task of the router is to correctly nals sent to the crosspoint switches become switch the output source (i.e., output 1 equals synced with the studio reference — i.e., the input 1 and will be changed for input 2) at pixels of every video of the same format (SD, the correct time and even perform glitch- HD and 3G) are at the same frame rate at the free switching (i.e., no error in the data same time. (See Figure 2.) Modern systems or no visual effect, no audio glitches, no also can include advanced functions: audio metadata disturbances). embedding/de-embedding, up/down/cross- What does this mean? In a modern sys- converters, composite video blanking and sync tem, a variety of signals are present in the (CVBS) conversion, and advanced diagnostics studio, including SD, HD and 3G at refresh at the input and output. Figure 1. T echnical rates of 50Hz, 60Hz and 59.94Hz, depending The case in Figure 2 is a simple one that difficulty color bar on the destination. The router has to switch the signal correctly to one of the switch lines (RP168), depending on the new input to be sent In sync EAV VANC SAV Y Cb EAV VANC SAV Y Cb out. Naturally, if the router changes from an SD signal to an HD signal, this change will affect VANC SAV Y Cb Y Delay by 1 pixel EAV VANC SAV Y Cb visual effects. Figure 2 shows an example of switching be- SAV Y Cb Y Cr EAV VANC SAV Y Cb Delay by 2 pixels tween two asynchronous HD signals. There are two types of input routers: synchronous and asynchronous. In synchronous routers, all the inputs should be synchronized to allow a clean switch, or at least a change between signals on Figure 2. Simple frame synchronizer Special Report supplement | February 2012 | broadcastengineering.com 39

Special Report BUILDING PRODUCTION, POST AND BROADCAST FACILITIES the frame synchronizer should deal with — Format Typical switch line same vertical refresh rate, same format, line aligned (e.g. HD-SDI). In fact, a simple line SD (525) 10/273 buffer can be used, but the reality is much SD (625) 6/319 more complex. HD (1080i/720p) 7/569 Next up is the line switch. Table 1 shows the switching (or change) line number in the video 3G (1080p) 7 frame of switch lines typically allowed for by the T able 1. Switch lines typically used various standards. Nowadays, routers allow the for the various standards user to delay or change these lines to make the router even more flexible. The switch lines are in the vertical blanking, and sometimes ancillary data could be in conflict with the switch lines. The new routers bring flexibility that could cor- rect this potential problem. Monitoring and test registers Temperature sensors As a reference, the SMPTE RP168 gives a Loss of signal detection recommended switch-point location. The central piece of a video and audio Input buffer with programmable input equalization IN0P OUT0P Switch core IN0N OUT0N router is called the crosspoint switch card. The fully non-blocking matrix array IN1P OUT1P M x M IN1N OUT1N Output buffer with programmable major difference between the telecommunica- programmable swing control Loss of signal detection output de-emphasis and tion\data communication router and the video router is the fact that the video router does not interact with the content inside the stream. These days, the router may interact with the inputs and outputs. For example, if a broad- caster feeds a CVBS signal to his or her router, he or she now can convert the CVBS to SDI INMP OUTMP signal immediately at the input and propa- INMN OUTMN gate the SDI signal to any of the outputs. The configuration Output buffer configuration Input buffer Active switch configuration crosspoint switch can take any input signals XOUTPUTEN XINPUTEN Intermediate switch configuration present and redirect those inputs to output Global crosspoint configuration signals. At one extreme, one signal can be sent XSET to all the outputs (big cross point switches Software interfaces and JTAG boundary scan MF (11:0) CONFIGSEL 2-wire serial interface have around 290 inputs x 290 outputs), so 290 CONFIG (1:0) 4-wire serial interface TDI, TGLK Parallel interface TEN, TMS TDO times the input. At the other extreme, the con- figuration can be one to one. Figure 3 shows the typical crosspoint switch. Figure 3. Block diagram of a crosspoint switch. Image courtesy of Mindspeed The crosspoint switch card The new 3G-SDI crosspoint switch cards integrate new pre-emphasis and equalization functions. The pre-emphasis/de-emphasis function is to boost high frequency to com- Standard SDI signal pensate for the FR4 (PCB) trace attenuation. Level 1 Level 0 As will be discussed later, in the section on dis- tribution amplifiers, high frequency is more Fall time Rise time attenuated in the copper medium than low frequency. The pre-emphasis and the equal- SDI signal with pre-emphasis izer share the same function on the input/ output. The pre-emphasis is used at the out- Level 1 Level 0 put, and the equalizer is used at the input. Boost on high frequency I prefer using the equalizer, because the pre- emphasis brings more noise and EMI into a system, but the pre-emphasis is simpler to Figure 4. Digital signals without pre-emphasis use and to set up than the equalizer. (top) and with pre-emphasis (bottom) 40 broadcastengineering.com | February 2012 | Special Report supplement

Figure 4 shows two waveforms. The first one cards. Redundant power supplies are also an is a digital signal without pre-emphasis, and important feature of the router. The router is the second one is the same signals with pre-em- vital equipment in the studio, and it should phasis. As the figure shows, the pre-emphasis work around the clock without failure. increases the amplitude of the edges of the sig- Another interesting aspect of the configu- nal, so the high-frequency increases. ration presented is the 144 outputs per cross- By using multiple crosspoint switches, point. If the manufacturer uses the same the manufacturers are able to create a large crosspoint, then the architecture allows for router. Two behaviors of routers that could be double outputs, e.g., integrating DA function- implemented are blocking and non-blocking ality or even a non-square (576 x 1152) block- behaviors. A non-blocking router is one that ing router. As a side note, other connection allows all the inputs to be sent to all the out- architectures are also possible. Documenta- puts without having any limitations. Almost tion is available on the Web for more details all of today’s routers are non-blocking, but it on those architectures. The input card 1-576 inputs Now the major board has been covered. But 1-288 289-576 the router is not composed only of the cross- point switch cards — input, output, synchro- nizer and control. The input card in Figure 7 288 x 288 288 x 288 1-288 289-576 1-576 outputs Figure 5. Blocking 576 x 576 router might be interesting to explain the difference and to highlight how to create a non-blocking Figure 7 . Input card 576 x 576 router. Figure 5 shows the configuration of 288 x is composed of equalizers (like those discussed 288 crosspoints connected together to form a earlier) and reclockers. The trend of the mar- 576 x 576 blocking router. It can be seen clearly ket is now using SFP with reclockers instead that input 1 cannot be rerouted to output 576. of putting the reclockers onboard. Figure 8 From Figure 6, it can be seen that in non- represents a block diagram of the architecture blocking routers, the inputs should be split based on SFP. into at least two copies, but more ideally, into The advantages of using an SFP-based four copies to allow redundant crosspoint architecture are many. Optical or coaxial 1-576 inputs 1-576 inputs 1-288 289-576 1-288 289-576 288 x 288 288 x 288 288 x 288 288 x 288 1-144 145-288 289-432 433-576 1-144 145-288 289-432 433-576 288 x 144 288 x 144 288 x 144 288 x 144 1-144 145-288 289-432 433-576 1-576 outputs Figure 6. Non-blocking 576 x 576 router Special Report supplement | February 2012 | broadcastengineering.com 41