The function of a video decoder is to separate the luminance and chrominance information contained in a composite video signal. There are several ways to achieve this, each with its own advantages...
 

The choice of decoder for a particular application is not a straightforward matter. In order to approach the problem logically, it is useful to consider the encoding process used to convey the information contained in RGB or YUV signals as a single composite signal.

R, G and B are matrixed to give the Y, R-Y, B-Y component video signals. The chrominance information contained in the R-Y and B-Y signals is modulated onto a subcarrier signal with restricted bandwidth (amplitude modulated in the case of PAL and NTSC standards, and frequency modulated in the case of SECAM). This chrominance information is combined with the Y or luminance signal to give encoded video.

The coding process places both chrominance and luminance signals with overlapping frequency spectra in the same time / frequency domain. Take any element of an encoded picture and that element contains both luminance and chrominance information.

The function of a decoder is to achieve a clean Y or luminance signal with broad bandwidth and no chrominance information remaining on it, and clean R-Y and B-Y chrominance signals with as little interference between the three component signals as possible. From these signals, the red, green and blue dematrixed signal can be achieved.

The most important aspect of a decoder is not its technical specification, but the resulting picture. Pictures from one decoder may "look" better than those from another but the specifications may appear to be very similar. In general, clean luminance (with very little residual chrominance signal) and good transient response are more important than bandwidth. Thus, a delay line / deluxe decoder with these features but only 3MHz bandwidth for the luminance may well look better than a comb filter decoder which has poorer chrominance rejection.

The simple decoder employs band-stop and band-pass filters to separate the Y and C signals. In PAL, the subcarrier frequency is 4.43MHz with a modulation band of 1.5MHz. By filtering out the chrominance signal using a notch filter, little subcarrier remains on the Y signal. However, the bandwidth of the luminance signal is reduced to some 3MHz (everything above 3MHz is filtered out together with the subcarrier 4.43 ±1.5MHz), giving a lack of horizontal resolution. Phase errors result in "Hanover Bars" and there is often fine and coarse cross-colour.

The delay line of deluxe decoder (such as the Eurogold MD-100) is a development of the simple decoder whereby the chrominance signals are averaged over two adjacent lines. This eliminates the Hanover Bars (the result is a decrease in chroma saturation) and reduces the fine cross-colour. This type of decoder has a low component count and is employed in most commercial television receivers. By careful design of the delay line decoder, use of top quality delay lines, sharp filters with overshoot suppression and variable filtering level depending on luminance level, very satisfactory results can be achieved for most applications.

Comb filter decoders employ a comb filter to separate the luminance and chrominance signals. The basic comb filter relies on there being no significant vertical changes in the picture content. The most common method employs a 3-line comb in which the composite video signal in the preceding and succeeding lines are combined with the line being decoded to yield a luminance-only signal by cancellation of the subcarrier. Any changes to picture content from line to line (i.e. vertical changes) lead to incomplete cancellation, resulting in this method being unsuitable on its own. However, the advantages are good horizontal luminance resolution (i.e. bandwidth) and good subcarrier rejection where the signal is line-repetitive.

Adaptive comb filter decoders combine the straightforward comb filter with the simple or delay line decoder. In effect, they switch between the two methods of decoding depending upon picture content, giving the best of both types of decoder. The picture is analyzed for vertical changes in picture content, usually on a line by line basis. This gives good luminance bandwidth (right through the subcarrier frequency band) and good subcarrier rejection. But there is some cross-colour, especially on vertical changes in picture content. Further, the adaptation switch (the change from one method of decoding to the other) can introduce spurious effects, especially on moving pictures.

Variable adaptive comb filter decoders (such as the DPC-100) offer a further refinement of the adaptive decoder, overcoming its problems by analyzing the picture content on a picture element basis and applying different levels of either comb filter chrominance filtering or standard notch filtering. Each element is analyzed to determine the most significant picture content, such as vertical picture changes, diagonal luminance detail, low luminance level, large areas of colour, horizontal or vertical coloured edges. The filtering characteristics are then adjusted to suit. This gives the best overall results available without going to the expense of a digital decoder (which has another set of disadvantages). The luminance and chrominance filters are individually optimised, there is broad luminance bandwidth and good subcarrier rejection. Such decoders are also easy to adjust. Their main use is in graphics and other high-resolution and critical applications. In other cases, a good delay line decoder will often suffice.
 

Last Updated