Electronicdesign 5510 Billlaumeister 595x335

Image Conscious

April 1, 2005
Consumers expect constant performance enhancement from their set-top boxes and video filtering. Bill Laumeister views the latest developments.

Advancing technology continues to advance the quality of the television pictures in our living rooms. Consumers see the quality of DVD and HDTV pictures and they demand better pictures from all video providers. Set-top box (STB) manufacturers have met the challenge, with today's digital products producing better-than-ever images. Two key areas of improvement involve anti-alias and reconstruction video filters. Microelectronics has replaced bulky expensive discrete component filters. To appreciate the improvement, it must be remembered how many times a video signal is processed during its travel to the home.

Analogue light is digitised by the CCD camera and recorded. In the editing and post-production process, the image may be converted repeatedly between analogue and digital. The digital signal is then sent via satellite to the cable head end, and in turn the cable signal is decoded by the STB and sent to the TV set. The TV set may re-digitise the analogue signal for scan conversion and presentation on a CRT, LCD or plasma display.

In an ideal video world, we would convert analogue light to digital video once for transmission before returning to analogue light for our eyes. Each conversion requires care and gives opportunity for garbage and artefacts to spoil our video. To provide a rule-of-thumb for compression ratios, professional recording and editing systems compress at an average range of 2 to 1 to about 10 to 1. MPEG to the home compresses on average 75 to 1 to 100 to 1. Teleconferencing may be compressed 200 to 1 or more.

These are broad averages to illustrate that the professional post-production is done before the heavy compression. Motion pictures may be converted from film at even higher resolutions to ensure the highest quality video for High Definition (HD) and Standard Definition (SD) TV.

At each analogue to digital (A/D) and digital to analogue (D/A) conversion, we must obey "Nyquist." In 1927, Harry Nyquist of AT&T and Bell Labs determined that an analogue signal should be sampled at twice the frequency of its highest-frequency component in order to be converted into an adequate representation of the signal in digital form. If the signal is sampled slower than the Nyquist limit, frequencies in the original signal that are above half the sampling rate will be "aliased." They will appear in the resulting signal as lower frequencies. An analogue low-pass filter is used before sampling to ensure that no components with frequencies greater than half the sample frequency remain. This is called an "anti-aliasing filter".

After a D/A conversion, a reconstruction low-pass filter is used to remove frequencies above our band of interest. The clock at 27MHz is typical of a MPEG2 coded signal. In addition to the wanted baseband video signal, there is the 27MHz clock and the side bands above and below it. A good filter like the dotted line in Figure 2a removes the high frequency clock and side bands, leaving pristine video. However, if someone skimps on the reconstruction filter we have a disaster. Any analogue non-linearity and, in fact, the next A/D without an adequate anti-aliasing filter will cause the high-frequency clock and sidebands to be transformed to low-frequency interference in the video. In addition, in this example, this consumer has the misfortune of a loose cable ground between the STB and TV set. This allows the neighbour's CB Radio signal to enter. Instead of a beautiful picture, our consumer sees trash. The horizontal bands coming, going and moving around are from the CB Radio leakage. Alias errors are harder to spot. The side bands only exist while a high frequency exists in the baseband video. During the edge transition, the unwanted side band energy has an unknown phase and sometimes adds or subtracts.

The effect of aliases on video compression deserves mention. All compression systems, no matter what mathematical transform is used, work in a similar manner. Two basic techniques are used because the human is relatively insensitive to them. First, compression selectively throws away high-frequency information because preserving low-frequency detail is more important than high-frequency detail. Second, compression removes redundancy from a series of pictures. If a background is stationary, it only needs to be transmitted once. Then the available bandwidth can be used for more detail in the foreground objects.

Aliases make compression difficult, since the added high-frequency "busy-ness" confuses the compression system, resulting in poorer images. A good anti-aliasing filter must be used. The filter's state of the art has changed remarkably in the last decade. Expensive and bulky video filters were once discrete inductors and capacitors for video frequencies. As high-frequency op amps became available, active filters once used only for audio became practical for video frequencies. Fairchild Semiconductor acquired the technology to build continuous-time video filters using tiny capacitors (in the femtofarad range, 10—15 farads) as monolithic integrated circuits. The transistors, capacitors, and resistors are all formed together on silicon. The result is a small, inexpensive, reliable, pre-tested filter with a coax driver. The filters are in full production and used by the millions in video products worldwide.

Filters are available in various bandwidths to support Standard Definition, NTSC and PAL, Progressive Scan 480p, and High Definition 1080I and 720p. Composite (NTSC, PAL, SECAM), S-Video (YC), Component (Y Pr Pb and RGB, HD, SD and PS) formats are supported. Advanced video applications include cable, satellite and Internet set-top boxes, DVD recorders and players, PVRs (Personal Video Recorders), progressive scan converters, HDTV, VOD (video on demand), and TV sets. One integrated circuit contains up to six filter channels with improved performance, small size, flexibility and lower part count for consumer products. Built in two-to-one multiplex switches on many inputs enhances the versatility of some filters.

Also available is a specialised video filter that provides a pre-correction just like an over-the-air television broadcast transmitter. When television was invented in the 1920s and 1930s, TV sets were made as inexpensive as possible. Engineers found a clever way to put one expensive filter corrector at the transmitter rather than putting correction into every TV set. This pre-correction is mandated in FCC regulations. It corrects for the intermediate frequency (IF) band-pass filter in the TV set, resulting in clean video transitions and properly timed colour information.

In STBs, this filter pre-correction is placed just before the video modulator. In addition, this filter cleans a space in the video where the modulator will place the audio carrier. Thus, when the consumer connects the Channel 3 or 4 RF signal to their antenna terminal, they receive the best possible picture. The picture has cleaner edges, the chroma is properly timed, and the video doesn't interfere with the audio. As a television transmitter, this pre-correction filter will run into the thousands of dollars.

In short, the STB industry has used the video state-of-the-art semiconductors to continue increasing video quality while maintaining an affordable product. Images used with this article were obtained from IMSI's Master Photos Collection, 1895 Francisco Blvd. East, San Rafael, CA 94901-5506 USA.

Sponsored Recommendations


To join the conversation, and become an exclusive member of Electronic Design, create an account today!