Jitter is a word that we hear often and manufacturers often quote. When we hear “Low Jitter” we all know that this is a good thing – and it is. The causes and cures for jitter are particularly complex and many discussions of jitter are very misleading. A quick overview of jitter is important because it is the most common source of timing errors, poor synchronization and ultimately bad sound in the digital studio.
I am going to borrow Julian Dunn’s definition of jitter, which is: “The variation in the time of an event – such as a regular clock signal – from nominal.” The waveform representing the Word Clock signal is a textbook perfect example of a square wave. In the real world a “perfect” square wave would have a finite rise and fall time with some under and overshoot. At the end of a cable it would be somewhat less than square, and in many situations where long cable lengths or bad termination were involved it would not be square at all and would become a problem causer instead of a problem solver.
Jitter is – for whatever reason and however caused – when the transitions in the square wave do not line up in time to the expected or nominal periodicity compared to the expected frequency. Jitter is created when the edges, due to their distortions, are misinterpreted as timing signals. This is a dynamic condition and the timing information will be random and constantly changing sometimes ahead and sometimes behind. The word “jitter” is appropriate because this change from nominal is never a stable change; it is always one that varies. In effect the timing information jitters back and forth in time.
Many different conditions can cause jitter to be produced. One of these is any interference signals that modulate the Word Clock lines, for example a ground loop. The type and severity of jitter will be determined by the ground modulation as these ground signals can range from 60 Hz hum all the way up to dimmer noise.
To properly measure or observe jitter it is necessary to first have a stable frequency reference, and then a method of dynamically comparing this reference to the actual signal. Many people attempt to measure jitter by simply looking at a Word Clock or AES3 waveform on an oscilloscope. I have one magazine article that erroneously came to conclusions on several high-end Word Clock products using several completely faulty methods to make their jitter measurements and aural comparisons.
The reason we are concerned about jitter is because each internal clock in our studio looks for and locks onto the edge of the incoming Word Clock square wave and uses the distance between each transition to determine the operating frequency of the Word Clock which it then uses as its own timing reference.
If the edge of each Word Clock square wave is occurring at different times or the Word Clock receiver is triggering on it at different times then each clock will have slightly different time bases – exactly the problem we are trying to eliminate.
Jitter is also a concern because it can be caused in so many ways. It is common to find Word Clocks that generates very low jitter from their internal reference but pass on high amounts of jitter when locked to an external reference like Video Sync or Time Code.
Excellent lock between an external reference like SMPTE time code or Video Sync and Word Clock are only excellent if they lock without transferring any jitter.
If you have a master Word Clock generator in your studio and everything works beautifully until you lock your Word Clock generator to incoming video, then high jitter passed thru to your Word Clock outputs from the video is your problem.
Assuming that the master Word Clock generator has no significant jitter itself the problem then becomes one of minimizing jitter that can easily be induced or transferred into the timing signal as it moves from device to device and, in a larger facility, from room to room. This involves paying careful attention to cables, connectors and termination.
Next to poorly designed digital recording equipment and poorly designed master Word Clocks, cables, equipment interconnection and improper termination are the most common causes of timing errors because they are a direct cause of jitter. All Word Clock signals must be connected using RG59 coax cable and properly terminated. RG59 cable has a characteristic impedance of 75 ohms and the BNC connectors that go on it are specifically designed for use at this impedance. RG59 cable provides a signal path for Word Clock waveforms that minimize cable induced jitter when fed from a 75-ohm source and must be terminated with a 75-ohm termination to properly function.
Cable runs must be kept as short as possible and separate isolated Word Clock outputs should be used to feed each Word Clock input, if at all possible. The most jitter critical equipment in your studio are your A/D and D/A converters. To ensure the lowest possible jitter these must be physically located as close to the master Word Clock outputs as possible in order to benefit from short cable runs. To do this, mount your Word Clock generator near your converters in the same equipment rack. The metal grounding of the rack will also help minimize ground induced jitter.
While it is common to use BNC-T connectors to distribute a single Word Clock output to several devices, this must only be done after you are certain that the 75-ohm termination for each Word Clock input you are looping to can be shut off. Restrict the use of BNC-Ts and loop no more than three Word Clock inputs. Restrict your looped cable lengths between Ts to very short distances and remember that the last BNC-T in the chain must be terminated.
A Word Clock output cannot be terminated by more than one input. Double termination is a common problem when looping Word Clock using BNC-T connectors. One termination per Word Clock output is required for optimum performance. No termination or more than one termination per output will cause excessive jitter and seriously degrade performance. Beware of 50-ohm coax cables and connectors – they are not the same thing.
It is well accepted that Jitter introduced into the AES3 bit stream manifests itself in clearly audible sonic degradation. Low jitter generation of reference Word Clock timing signals, their jitter free synchronization to video and SMPTE and their proper distribution between digital audio devices throughout the studio will result in the precise timing of digital audio bit streams as well as deliver a noticeable improvement in the quality and subtlety of recorded and mixed music as well as the reliability of transfers, spotted sound effects and layback to video.
Bob Snelgrove is the President of GerrAudio Distribution and the Canadian Product Specialist for Audio Precision test instrumentation.