1. INTRODUCCION
4.2. DISCUSION
MAC has a different approach in using time multiplexing for luminance, chrominance and audio transmission. Each 64ps line is broken as follows:
• lOps for audio/data transmission; • 18ps for chrominance transmission; • 36ps for luminance transmission.
Audio is transmitted digitally in duobinary, along with additional data, which means the audio/data modulation provides stronger modulation energy. Chrominance and luminance signals are compressed by ratios of 3:1 and 3:2 respectively, and this increases MAC video bandwidth to 13.5MHz. Two sub-standards exist, D-MAC (A/D transmission rate is 20.25Mbit.s’^ ), and D2-MAC (A/D transmission rate is reduced to 10.125Mbit.s'^ ).
3.3.2 : RF Modulation Standards
Certain modulations create strong sideband energies at particular frequencies. This may cause interference with other signals, but since terrestrial TV transmitters tend to have a close radio horizon this is generally not a problem. DBS TV transmissions have wider coverage and a greater capacity to interfere with other communication links, terrestrial or space-based. For this reason, three precautions are recommended to mitigate this problem: • Frequency modulation is recommended by the CCIR for DBS TV transmission;
• Pre-emphasis functions re-distribute rf spectral energy density to compensate for the uneven probability density function of luminosity and colour information in TV pictures, and thus provide a more even spectral density;
• A dispersion signal is used to displace the carrier using a sawtooth signal of +/-600kHz every 2 0ms, and reduce risk of persistent interference.
3.2.3 : Simulated TV Waveform Ambiguity Functions
To facilitate comparison between ambiguity functions of PAL and MAC video standards, digital representations of each TV waveform are sampled at 20.25MHz. According to CCIR recommendation 601, baseband video sampling rates should be 13.5MHz for luminance and 6.75MHz for chrominance. For D-MAC, the compression ratios of 3:2 and 3:1 respectively ensure a consistent digital rate of 20.25MHz though the compromised D2-MAC standard halves this to 10.125MHz. In the case of PAL, the ideal UHF/VHF transmitter filters the rf band in a peculiar way, such that the lower band edge is IMHz below the carrier, while the upper band edge is 5.8MHz above the carrier. This ‘vestigial sideband’ modulation means that PAL TV waveforms at UHF/VHF are double side-band within IMHz of the carrier, but single side-band between IMHz and 5.8MHz away from the carrier. PAL TV waveforms sampled at 20.25MHz must be appropriately filtered.
CfiapterJ ŒXBS ^ Waveform
AM to FM conversion is the most significant difference between terrestrial and DBS broadcasts. The effect of broadcasting TV waveforms by frequency modulating the cairrier on the autocorrelation functions is strong. For example. Figure 3.4a shows the AM PAL video signal for the eight colour band testcard, and its auto-correlation function. This does not drop below -2dB at any delay time. Figure 3.5a shows the same signal, this time frequency modulated at 13.5MHzV \ Sidelobe levels are now approaching those of a sine function, while the mainlobe has a 3dB-level width of the order of Ips. For comparison. Figure 3.2a illustrated a chirp of dispersion 13.5MHz per 64ps.
Some manipulation can be performed to reduced sidelobe levels. Figure 3.4b illustrates the effect on the testcard ACF if the dc signal of 0.5V is trimmed out. DC energy is the most obvious cause of high mean ACF sidelobe power. However, for FM signals the general approach is amplitude weighting. Figure 3.2b illustrated the typical example of cosine weighting a chirp of T=64ps over each period. There the sidelobes reduced to - 60dB. The same technique applied to the FM PAL testcard (Figure 3.5a) does not have such a great effect though, any difference actually being hard to discern.
The reason for this difference lies in the relation of instantaneous frequency with time. For the chirp, this was linear and therefore deterministic, while for the example TV line, this may have been determinate here, but is not a simple function of time, and in general may be expected to vary in an indeterminate way. The point is that with a simple time- frequency relation, amplitude weighting in the time domain is effectively amplitude weighting in the frequency domain. For example, with the chirp the lowest and highest frequencies were attenuated by the most. For a typical TV waveform, it is not clear which frequencies would be attenuated. Therefore to properly weight a reference TV waveform, direct weighting in the frequency domain is required.
This adds another reason for preferring an FFT based method of pulse compression. Before the inverse Fourier transform, the reference TV waveform energy spectrum can be weighted to improve compressed TV waveform shape. Two processes that have not been pursued in practice, but could be performed are:
• Balancing of waveform spectral density by attenuating higher energy frequencies to the level of weaker energy frequencies. This will reduce the ripple of sidelobes due to particular spectral energies, but may result in a severe loss of power. A Gaussian post weighting spectrum may be opted for to avoid sidelobes altogether;
• Notch filtering at 15.625kHz and frequencies that are factors of this, to suppress ambiguities at 64ps.
CftapterS 1XBS ‘TV Waveforms