Distribution Amplifiers (DA's)

Distribution amplifiers are found all over a television plant. DA's take one signal in , and provide several identical copies of it. In the analog world, DA's are used to allow a signal to feed several things simultaneously.... a picture monitor, a switcher, a patchfield, etc. This function is still required in SDI, but DA's can also serve another function... they can completely restore the signal to "brand-new" after a long cable run.

SDI As It Starts Out

After a long Cable Run

The lower graphic shows the effect of cable capacitance and noise on the signal.... it is lower in amplitude, the rise and fall times of the pulses is longer (slope), it is noisy, and the slope and the noise have caused the receiver to jitter in time, because of the vagueness of where the pulse goes from high to low and vice versa. 

There are two kinds of distribution amplifiers for SDI. The simplest is the equalizing DA. It can restore the level, and make the signal capable of driving another length of cable, but the jitter and noise remain. Equalizing DA's are used to "fan-out" an SDI signal so that it can feed several pieces of equipment in an environment where the cable runs aren't too long (an edit room, for instance).

Re-Clocking DA's provide fan-out, but also use a phase-locked loop clock-recovery circuit to re-create a jitter and noise-free signal that is as good as new. 

Just to drive the point home, the cable-length limit in SDI is about 250 meters. After that, the signal may not be recoverable. An equalizing DA at the end of a long run will not make the signal suitable for another long run, but a re-clocking DA will.

Routing switchers usually feature re-clocking on every input.

Timing

In analog, we had to ensure that the video signals arrived within a couple of nanoseconds at the production switcher. Gross miss-timings could be fixed with a framestore synchronizer (AKA framesync) device. Why didn't we just throw framesyncs all over the place to fix the timing? There was a slight degradation from passing through a framesync, but the biggest problem was that the video was one or two frames late coming out of a framesync, and that caused the audio and timecode to be early. 

As long as the video hits a framesync early enough, there doesn't have to be a one-frame delay at the output... 

With digital equipment, having a frame sync on every input of a switcher wouldn't be too difficult, but the accumulated one-frame delays would be a mess. In an SDI installation, timing-sensitive equipment like production switchers have small re-syncing buffers equal to about one-half a horizontal line-period  (about 32 microseconds). As long as the signals arrive within this time compared to reference, all is well.

Reference 

Yes, well. What do you suppose this fancy new system uses for a timing reference? Wait for it...

The timing reference for SDI is good old analog black, or bars. This makes sense... black has all of the important timing marks in it in a simple-to-decode fashion. Burst is ignored. Not only is analog black the perfect reference, after switching to SDI the average TV plant will have lots of leftover analog DA's and video cable. In most cases, the existing reference wiring can be left in place.

Routing

SDI signals can be switched at the router whether they are synchronous or not. The signal will re-sync much like the old system. Timed switching is done where it matters, as in a router that is an on-air switcher. In the edit room, where routing is simply used to change configurations not on air, timing doesn't matter.

In synchronous switching, the switch is done during the vertical interval.

The one switching caveat has to do with embedded audio. Suddenly switching bit-streams can cause a pop at the switching point. Some routers buffer the inputs to ensure that the signals are timed exactly.