Serial transmission of data representing video and audio signals is standard practice across the broadcast industry. The advantages are many, yet with these advantages come a new set of parameters that need to be monitored and controlled to ensure proper data transmission. Eye pattern analysis is a common method of testing, but other, more advanced testing means are required for full jitter analysis.
But first, let's take a quick look at serial video transmission. The standardisation of encoding and serial transmission specifications is one of the major advantages to the present state of the industry. The Society of Motion Picture and Television Engineers (SMPTE) has codified the methods used in the broadcast industry, and these are now universally accepted by equipment manufacturers, broadcasters, and all the intermediate contributors.
Serial data transmission involves taking the stream of 10-bit words used in uncompressed component video schemes and manipulating it into a final electrical signal to launch onto a cable. To do this, the parallel stream is turned into serial, a new serial clock is generated, and the serial stream is encoded to lower the average frequency - and to more evenly distribute frequencies within the allotted bandwidth.
Further encoding is performed to more evenly distribute transitions in the final serial stream. This encoding uses an SMPTE-defined scrambling polynomial and ensures that a long series of binary '0' values does not cause a corresponding lack of transitions in the serial stream.
Some parameters influence the serial transmission integrity. As effective as these techniques are, there are still numerous challenges in transmitting data at these high rates.
Transmission path bandwidth is especially crucial for high-definition (HD) serial data transmission. Using this method, the data rate is 1.485 Gb/s, which requires approximately 2.1 GHz of transmission bandwidth to maintain the data waveform integrity. Limiting bandwidth produces the same effects on the data transmission waveform as on any other non-sinusoidal wave shape - rise and fall times are slowed, causing the waveform to appear more sinusoidal, and other time-domain effects may appear, such as over- and under-shoots.
Reduced signal amplitude can indirectly cause frequency domain effects as well. All kit used with serial data transmission uses a cable equalizer component to automatically restore the received signal to nominal amplitude. Obviously, the lower the received amplitude, the more gain is required to perform this equalisation. Noise introduced in this analogue process will appear as jitter in the equalised data waveform.
This is, however, only one way that jitter can be introduced into the serial transmission path. The generation of the serial data-rate clock signal is typically performed by a phase locked loop (PLL). The design of this PLL circuit is critical in the overall jitter performance.
Some amount of residual jitter is always present, but this process can either attenuate or amplify the jitter present on the parallel clock (or do both simultaneously, but at different jitter frequencies). This circuit can also introduce patter-dependent jitter, where the jitter changes throughout the video frame based on the video signal contents.
While the overall jitter amplitude is crucial in assessing a serial video transmission system, it is meaningless without knowledge of what jitter frequencies are present. Jitter at a particular amplitude (time value) will have a drastically different effect on a system if its frequency is widely varied. The two jitter quantities, amplitude and frequency, cannot be separated when analysing a system - knowledge of one parameter is meaningless without knowledge of the other.
Today, there are several advanced measurement features. Eye pattern analysis is the most basic method of analysing jitter.
Although useful, eye waveform analysis has numerous weaknesses. An eye pattern display can tell us the total amount of jitter, but gives no indication of jitter frequency or jitter variances over time. Some other techniques are required for these analyses.
The output of the jitter demodulator gives an accurate view of jitter vs. time. This output is usually plotted as jitter amplitude (as the Y axis) and time (as the X axis) at video sweep rates - one or two lines or fields. This yields a plot of jitter that shows any pattern dependencies, such as jitter steps during vertical interval or due to video content. Examining only the eye waveform cannot reveal this change of jitter with time.
Further analysing this waveform can yield information about the frequency content of the jitter. Analysing jitter frequency is crucial since it is a critical part of a system's jitter tolerance. The total amount of jitter that any system can process with acceptable performance is frequency dependent. Systems can usually tolerate more absolute jitter at very low frequencies. Jitter tolerance decreases at mid-band frequencies (roughly 100Hz to 1 kHz) before starting to decrease again at high frequencies outside the response range of the serial receiving circuitry. So, knowing the absolute jitter value is not enough to predict the response of the system; the frequency must be known as well. The jitter histogram displays this aspect of the jitter analysis.
End users, therefore, need to look out for more advanced signal analysers that can display both parameters, such as jitter frequency, as well as pattern dependence.
Michael L. Richardson is director of Product Development for Videotek Test and Measurement products at Harris Corporation.