[opendtv] Re: PAL; 405 line pictures

  • From: S J Birkill <sjb@xxxxxxxxx>
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Fri, 15 Dec 2006 10:33:20 +0000

Walter Bruch and his PAL team at Telefunken tackled the major shortcoming of NTSC: hue errors due to distortion in video amplifiers and modulators. It was the cumulative phase errors through dozens of amplifiers in a TV studio complex, hundreds in a terrestrial distribution chain, plus transmitter modulators and receiver demods, essentially non-linear circuits, that spoiled NTSC. It's easy to forget that before satellite distribution in the 1970s, Americans seldom saw accurate color on their screens.

Differential phase was the killer. Those non-linear stages systematically shifted chroma phase according to the instantaneous luminance level (above black, burst reference) the chroma was riding on: flesh tones took the brunt, yielding NTSC's traditional green, orange or purple faces; hence the system's need for a 'hue' control. PAL eliminated this, converting phase error to a small amplitude error -- diff phase distortion looked just like diff gain, introducing only a slight desaturation at the higher luminance levels. One less knob on the set.

PAL's line-wise alternation of the V (R-Y, give or take small differences in scaling, matrixing and gamma-correction) chroma component's modulation phase, serves to correct chroma phase errors by optical (simple PAL, long obsolete) or electrical (delay-line PAL) averaging. The delay-line method halves vertical color resolution, which then matches (again approximately, so don't take me to task on the numbers -- this is from memory) its band-limited horizontal resolution. NTSC retains the higher vertical color resolution -- that's the *only* loss from PAL coding. In fact there's a slight collateral gain: the phase switching dithers the CSC pattern, making for a less visible diagonal 'dot crawl' in saturated picture areas; CSC notch filtering in the receiver could be relaxed a little.

['Composite video over telephone lines' is misleading: 625-line video distribution in the UK (for instance) was from its very beginning (1964) via 5.8-MHz equalized co-axial cable within studio centers and across cities, and via 2GHz or 7GHz FM microwave links between studios and transmitters. Even in the 405-line days when video bandwidth was 3MHz, STLs on the main distribution backbone were in Post Office co-axes using an HF carrier system, with a combination of rebroadcast and microwave radio links for the more remote stations.]

Hue error correction is just one component of the reason the PAL pictures looked so much better than NTSC in the early days. There *were* different production values: European lighting and set designers avoided the gaudy sets used in US studio productions to 'show off' color TV, instead letting natural color prevail (alas those values have largely been lost!) The 576 active lines (versus 480) plus the increased video bandwidth (by about the same ratio) helped a lot. In PAL-I countries at least, the bandwidth could be better exploited, as there was space for HF luminance components above the 4.43MHz CSC notch, before cut-off at about 5.5MHz and the receiver's 6MHz sound (aural) notch; things were much tighter between 3.58 and 4.2MHz in NTSC land. And of course for PAL the distribution was shorter: rather than syndicates 'out west', BBC TV's furthest outpost was Shetland, a mere 600 miles from London.

I don't feel the black level/blanking level difference was noticeable. IIRC it was designed to minimise the visibility of sync pulse overshoots during the line flyback (horizontal retrace) period, but with improved blanking (suppression) in color TVs it became unnecessary. In any case it was accepted practice even in PAL studios to apply a set-up ('lift') of between 2 and 5% to the picture, which amounted to much the same thing. The reduction of video dynamic range (an SNR delta of <1dB) due to the IRE blanking level was surely insignificant in analog days.

405-line pictures of course looked sharp because there *was* no CSC. But still it needed good STLs, careful transmitter alignment and a good TV to realise the system's potential 3MHz video bandwidth -- most of the time you'd be lucky to see 2.5MHz in the home. 377 active lines was only three-quarters of what US viewers were accustomed to, and it showed. Don't forget also that sync separators were more art than science in the pre-color days: TV sets frequently lost interlace totally, giving the effect of seeing the world through a stationary 188-line raster! And let's not forget the 10.125kHz line whistle and the 50Hz field flicker on those fast-phosphor B&W tubes. That delegation must have been used to some pretty rough 525-line stuff!


You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.

Other related posts: