Table 1. 4:2:2 component digital format sampling frequencies. Click here to see an enlarged diagram.
North American and European digital standardization efforts resulted in CCIR Recommendation 601, Encoding Parameters of Digital Television for Studios, now known as ITU-R BT.601. This recommendation established an agreement on a component digital approach that is compatible with both the 525/59.94 and 625/50 scanning standards. It is at the root of all subsequent component digital developments.
The coded signals
The recommended and most commonly used digital coding is based on the use of one luminance (E'Y) and two scaled color-difference (E'CB and E'CR) signals. The coded signals are defined by the following expressions:
Table 2. Sampling rates of various members of the family. Click here to see an enlarged diagram.
E'Y = 0.587 E'G + 0.114 E'B + 0.299 E'R; referred to as Y in North America E'CB = 0.564 (E'B - E'Y); referred to as PB in North America E'CR = 0.713 (E'R - E'Y); referred to as PR in North America
These signals have the following characteristics when representing a 100 percent color-bars signal without setup (known as a 100/0/100/0 color bars signal):
Table 3. 4:2:2 sampling structures and horizontal resolution. Click here to see an enlarged diagram.
- The luminance signal has a peak positive value of 700mV and no setup.
- The scaling factors for the E'CB and the E'CR color-difference signals were chosen to obtain a bipolar signal with a p-p amplitude of 700mV. These scaling factors differ from those used with the composite analog PAL and NTSC signals or those used by the Betacam and MII component analog VTR formats.
The sampling rates
Early proposals for the values of the sampling frequencies of the Y signal specified a multiple of the subcarrier frequency (fSC) of the associated composite video signal. This resulted in the 4:2:2 concept, whereby the E'Y signal is sampled at a frequency of 4fSC, and each of the two color-difference signals is sampled at 2fSC, hence 4:2:2. The major achievement of CCIR 601 is specifying a set of sampling frequencies common to both the 525/59.94 and the 625/50 scanning standards. The selected frequencies are common multiples of the horizontal scanning frequencies (fH) of both standards, as well as 3.375MHz. Table 1 shows the derivation of the 4:2:2 standard sampling frequencies.
Figure 1. Details of the 4:2:2 sampling grid. Click here to see an enlarged diagram.
A family of sampling rates common to both scanning standards and based on 3.375MHz has evolved. Table 2 shows how the sampling rates are derived.
The sampling frequency has a direct bearing on the frequency response and the number of horizontal picture elements (pixels) that the system can resolve. Recommendation 601 specifies the low-pass filter (LPF) characteristics of the anti-aliasing (ahead of the A/D converter) and reconstruction (after the D/A converter) filters, which determine the analog in/out frequency response characteristics. For the 4:2:2 format, the resulting luminance bandwidth is 5.75MHz. This is a compromise between the slightly higher requirements of the 625/50 scanning standard and cost-optimized A/D conversion circuitry. It is worse than the analog studio signal-distribution elements and state-of-the-art cameras. The color-difference signal bandwidth for the 4:2:2 format is 2.75MHz. This exceeds the chrominance bandwidth of the NTSC or PAL analog composite standards. Figure 1 shows details of the 4:2:2 sampling grid.
The specified sampling frequencies result in an integer and equal number of sample periods during the active line periods for the two scanning standards. The sampling strategy is called orthogonal sampling. In the 4:2:2 format, with twice as many luminance (Y) as chrominance samples, the chrominance samples (CB and CR) are time-coincident (cosited) with odd Y samples. Table 3 details the 4:2:2 format sampling structure and the LPF-related analog horizontal resolution. It shows that the two scanning standards have an equal number of samples per active line.
Figure 2. Simplified block diagram of a Rec. 601 component digital 4:2:2 encoder. Click here to see an enlarged diagram.
The sample resolution
The sampling process results in signal amplitude values measured periodically at the sampling rate. Analog signals may assume an infinite number of amplitude values inside established limits. The quantizing process results in converting the measured voltages into digital data. It results in “quantizing errors” (Qe), inaccuracies in the digital representation of the analog signal related to the number of bits per sample. CCIR 601 specifies a resolution of eight bits per sample, which allows for 256 levels (28) of amplitude information to be represented for each component. This number is reduced slightly by the need to provide headroom, which helps avoid analog signal clipping and results in a specified quantizing range. With eight or more bits per sample, Qe manifests itself as random noise. The choice of eight-bit resolution was based on the state-of-the-art technology of the 1980s and is satisfactory only with analog source equipment having a signal-to-noise-ratio (SNR) of about 50dB, which effectively masks the eight-bit Qe.
Figure 3. Relationship between 100/0/100/0 component color-bars signals and 10-bit and eight-bit digital sample values. Click here to see an enlarged diagram.
Figure 2 shows a simplified block diagram of a 4:2:2 encoder. The sampling of the three components results in three bit-parallel datastreams, which are time-division-multiplexed into a sequence of cosited CB,Y,CR samples followed by an isolated Y sample and so on. The multiplexed datastream has a rate of 27Mwords/s. Figure 3 on page 30 shows the relationship between the component signals representing a 100/0/100/0 color-bars signal and 10-bit and eight-bit digital representations. It also shows the relationship between the E'Y, E'CB and E'CR analog component signal levels corresponding to a 100/0/100/0 color bars signal and the 10-bit and 8-bit Y, CB, CR digital sample values.
Currently, studio equipment uses a resolution of 10 bits per sample. In a 10-bit system, there are 1024 digital levels (210), expressed in decimal numbers varying from 0 to 1023 or in hexadecimal numbers varying from 000 to 3FF, which can represent the sampled analog video signal amplitude.
Note that the sync portion of the luminance signal is not sampled. Digital levels 000, 001, 002, 003 and 3FC, 3FD, 3FE, 3FF are reserved to indicate timing references. The 700mV luminance signal occupies a range extending from blanking (64 decimal or 040 hexadecimal) to peak white (940 decimal or 3AC hexadecimal). The bipolar (±350mV) color-difference signals are shifted up to fit the A/D converter, which requires unipolar signals. They occupy a range extending from the digital equivalent of the maximum negative level (64 decimal or 040 hexadecimal) to the digital equivalent of the maximum positive level (960 decimal or 3C0 hexadecimal). A small amount of bottom and top headroom allows for misadjusted or drifting analog component signal levels. The resulting Y SNR is 70.35dB for 10 bits and 58.3dB for eight bits.
As noted, the analog sync is not digitized. For synchronizing purposes, two four-word timing reference sequences (TRS) are carried. These are the end of active video (EAV) and the start of active video (SAV) TRS signals. Figure 4 (525/59.94 scanning) and Figure 5 (625/50 scanning) show details of the horizontal blanking interval. In both standards, there is a considerable ancillary data space that can be used to carry up to 16 audio signals (eight stereo signals) as well as other data such time code. In a future article, we will discuss the characteristics of bit-serial signals.
Figure 4. Details of the 525/59.94 scanning standard horizontal blanking interval showing the composition of the 4:2:2 digital data multiplex and the position of the timing reference signals, EAV and SAV. Click here to see an enlarged diagram.
Figure 5. Details of the 625/50 scanning standard horizontal blanking interval showing the composition of the 4:2:2 digital data multiplex and the position of the timing reference signals, EAV and SAV. Click here to see an enlarged diagram.
Michael Robin, a fellow of the SMPTE and former engineer with the Canadian Broadcasting Corp.'s engineering headquarters, is an independent broadcast consultant located in Montreal, Canada. He is co-author of Digital Television Fundamentals, published by McGraw-Hill and translated into Chinese and Japanese.
Send questions and comments to:firstname.lastname@example.org