Colour TV
It is traditional to give facetious translations for the colour
system acronyms:
- "Never Twice the Same Color".
- "Pictures At Last" (refers to
the amount of time it took for Europe to get colour TV).
- "Pay for Added Luxury"
- "System Essentially (Entirely?) Contrary
to the American Method". |
NTSC
All analog colour TV systems are based on the NTSC system, which
for its day, was a brilliant feat of engineering. The idea behind
it was that to transmit color TV, it wasn't necessary to transmit
separate channels of Red, Green, and Blue. Three channels are
necessary, because our eyes have three types of colour receptor,
but these can be made up of combinations of R G and B, such that
one channel corresponds to a monochrome picture, and the other
channels tell the TV how to deviate from monochrome in order
to recreate colour. We take it for granted now that a black and
white TV can tune to a colour signal, but the systems proposed
before this invention were not black and white compatible.
The signals used are called Luminance (Y), (where Y= 0.59R
+ 0.3G + 0.11B, the combination which simulates white light),
and colour difference R-Y and B-Y. The colour difference
signals together are known as Chrominance or Chroma.
The Y-channel is simply modulated onto a monochrome TV waveform,
to provide a black and white compatible TV signal, and the chroma
has to go somewhere else. The obvious thing to do with the chroma,
is to put it into a couple of spare TV channels next to the monochrome
one, but a few clever tricks make such a bandwidth-wasteful approach
unnecessary. The first observation is that the eye is less sensitive
to detail in colour (cone cell vision) than it is in monochrone
(rod cell vision), so the two colour difference signals can be
reduced in bandwidth so that they each need only half the width
of a TV channel (or less). So now, we've fitted the colour signal
into only two monochrome TV channels, but better still; two signals
can be squashed into the space of one by using a scheme called
Quadrature Amplitude Modulation (QAM) - the method used nowadays
to get large(ish) amounts of data to travel through telephone
lines. So now we're down to one and a half TV channels, but then
the ghost of old Jean Baptiste Fourier comes to show how to get
it all into one.
If you look at the spectrum of a TV signal, you find that it
is not continuous, but is made up of spikes, at multiples of
the line scanning frequency. Each of the spikes has sidebands
on it, spaced off at multiples of the frame rate, but the frame
sidebands are weak compared to the line harmonics; and so the
spectrum essentially has hundreds of large gaps in it, each easily
big enough to fit an AM radio station. Chroma signals are also
TV signals, so they have the same type of spectrum, so the trick
is to modulate the chroma onto a subcarrier which sits exactly
half way between two of the spikes of the luminance spectrum.
In this way, the chroma signal can be placed right in the video
band, and all of the spikes of its spectrum fit between the luminance
spikes. This technique is called interleaving and, in
combination with interlacing, results in a TV waveform
in which the subcarrier phase gets back to its starting point
after four fields (2 frames). Modern video-tape recording was
in the process of being invented at the time by one A.
M. Poiniatoff (par EXcellance)*, with the financial backing of Bing
Crosby. The RCA engineers may not have known it, but NTSC would
turn out to be 'electronic editing friendly'.
To get the chrominance signals back out of the TV signal, the
system designers resorted to a trick called synchronous demodulation,
- a method used by spectoscopists and others to recover signals
buried in noise. There was one small problem however, which was
that the subcarrier was visible as tiny dots on the screen, and
while colour tubes of the day were too crude to reproduce them,
they could be seen on black and white sets. The solution was
to use a trick developed for short-wave radio communication,
which was to use suppressed-carrier amplitude-modulation for
the chroma. It may sound surprising, but the carrier signal of
an AM radio transmission carries no information. A lot of transmitter
power can be saved by leaving it out, as long as it is re-inserted
in the receiver prior to demodulation. A short-wave radio had
a carrier insertion oscillator or beat-frequency oscillator
(BFO) for this purpose. The operator simply tweaks the tuning
to get the oscillator in the right place, et voila - the signal
becomes intelligible. For a QAM signal however, and for synchronous
detection, both the frequency and phase of the carrier
must be re-established, so in NTSC, a small reference burst of
carrier is sent just before the beginning of each line. Using
suppressed-carrier QAM was a brilliant idea, because it
meant that in areas of the picture where there was no colour,
there was no colour signal either. The dot interference was thus
greatly reduced, and effectively confined to areas of high colour
saturation only.
NTSC achieved a respectable 3:1 bandwidth compression, in an
age when valves (vacuum tubes) were the dominant technology,
and no one had yet made an integrated circuit, let alone a DSP
chip. It was also very daring, using every analog signal processing
trick in the book; and to cap it all, it worked. It is not perfect
however, and suffers from two noticeable defects:
1) When the video signal is rich in frequencies which lie in
the colour channel, luminance leaks into the colour decoder and
gives rise to psychedelic effects. For this reason, checks and
pin-stripes will always be out of fashion in NTSC TV studios
(and PAL is inherently worse). The effect is called 'color
fire' or 'cross color', and can be eliminated by modern
signal processing techniques. The TV set has a 'color killer'
circuit, to prevent cross-colour from appearing on monochrome
pictures, although nowadays TV companies tend to sabotage black
and white films by leaving the colour burst switched on..
2) When the composite NTSC signal suffers from distortion in
the transmission chain, the QAM signal is skewed in phase, and
hue shifts occur. An NTSC TV needs to have a Hue control,
to get flesh tones looking right, but even this cannot fix the
brightness dependent differential phase distortion which
sometimes occurs. The NTSC knew about this, and an alternative
scheme called Chroma Phase Alternation (CPA) was suggested
as a solution. CPA was based on the observation that if one of
the colour difference signals (e.g., R-Y) was inverted on alternate
fields, then any hue errors on alternate lines of an interlaced
frame would be equal and opposite, and if you stood back from
the screen, pairs of incorrectly coloured lines would average
to the correct colour. The problem was, that if phase errors
were bad, they gave the picture a flickering 'Venetian blind'
effect, which could look a lot nastier than a straightforward
hue error. The NTSC decided that the marginal benefit of CPA
did not warrant the added complexity. |
PAL
As the American NTSC system reached the marketplace, other countries,
notably in Europe, were working on their systems. In particular,
a team at Telefunken, under the direction of Dr Walter Brüch,
was working on an ingenious modification to the NTSC system which
involved inverting the phase of one of the colour difference
signals on alternate lines. They called the system PAL, which
stood for Phase Alternating Line, or something like that.
The problem with the PAL method was that, if the chroma phase
errors were bad, they gave the picture a revolting 'Venetian
Blind' effect, which they called 'Hanover Bars', after
the town in which the effect was 'first' discovered (The NTSC
almost certainly considered both line and field CPA - but would
have rejected the line version on the grounds that, over a whole
frame, it exacerbates the Venetian blind effect by producing
a pair of lines of one hue followed by a pair of lines of another).
The solution was to average the hue errors electronically, by
taking the TV line coming in off air and combining it with the
previous line stored in an analog memory (un memoire). The original
memory was a device called a 'delay line' (line as in wire,
or cable, not TV line, even though it stored almost exactly
one TV line), a cumbersome and lossy collection of coils and
capacitors designed to simulate the time delay of a very long
cable. This was soon replaced by a small block of glass with
two piezo-transducers glued to it - an ultrasonic delay
line.
The PAL variant of NTSC needed a few tweaks to turn it into a
viable standard. In particular, the dot interference with a half-line
colour subcarrier offset was exacerbated by the phase inversion
process, which caused the dots to line-up vertically. The solution
was to move the subcarrier to a position a quarter of the line
frequency away from one of the line harmonics (actually 15625
x 283.75 + 25 Hz = 4.43361875 MHz). This is something of a compromise,
because the interleaving is not so good. This reduces the signal
to noise ratio of the synchronous demodulation process, exacerbates
colour fire, and gives highly saturated parts of the picture
a crawling appearance. The quarter-line offset, with interlacing,
also results in a subcarrier which returns to its original phase
after 8 fields (4 frames), which precludes precise electronic
editing. This was a small price to pay however, for the opportunity
to take out patents on top of the NTSC system and use them to
control the European marketplace. The point was not to patent
the transmission standard however, which was in any case just
NTSC-CPA-H, but to patent the technology used in the receiver.
The Telefunken team described three decoding methods for HCPA
(sorry, PAL), which they called PAL-S, PAL-D, and PAL-N (the
N in this case stands for 'new' and is nothing to do with TV
system N used in South America). PAL-S (simple PAL), was the
"let them stand back until the Hanover bars aren't noticeable"
approach, which couldn't be patented because of the NTSC prior
art. PAL-D was the basic delay-line method, and PAL-N or 'Chrominance
Lock', was a more sophisticated delay-line method which could
track and cancel differential phase distortion, without the loss
of colour saturation which occurs with the basic D method. Telefunken
patented the delay-line methods, and used these patents vigorously
in an attempt to exclude Japanese TV manufacturers from the European
marketplace. Consequently, until the PAL patents expired in the
mid 1970s, all Japanese TV sets in Europe either used the disgusting
PAL-S, or were made by Sony.
In the early 1970s, Sony introduced a range of PAL Trinitron
TV sets, which had a Hue control like an NTSC set. These were
a breath of fresh air in comparison to the dreadful Shadow-Mask
TVs of the day, and it was quite a status symbol to own one.
The colour decoder contained a delay line. Telefunken sued -
and lost. The dreaded Japanese had hit upon a third delay-line
method, which was so devilishly simple that only someone whose
brain was not saturated with pro-PAL propaganda could see it.
Sony used the memoire to store a line so that it could throw
away alternate lines and treat the signal as though it was NTSC*.
If NTSC was as bad as it was claimed to be, Sony should have
been inundated with complaints; but as it was, if you owned a
Trinitron set in those days, people came round to your house
to watch it with you, and the TV companies adopted the video-monitor
versions as studio monitors (despite the EIA tube phosphors -
it was the brightness they wanted). The irony was that
the most discerning TV owners were watching PAL as NTSC. Sony
changed to the PAL-D method when the Telefunken patents expired,
and felt obliged to devise a hue control for that, to keep up
the tradition. The control didn't do anything useful, it basically
gave the user the choice of whether or not to have Hanover bars,
and they dropped the idea fairly quickly.
* notes on
the Sony system |
SECAM
The French system results from a highly pertinent observation,
by Henri de France, its inventor; that if you're going to use
an expensive memoire to decode the signal, then you might as
well dispense with the troublesome QAM and simply send R-Y and
B-Y on alternate lines. He thus came close to a scheme which
might have given a pronounced improvement in any environment
(studio, editing, and transmission), but the devil is always
in the details. There were two technically feasible methods,
at the time, for extracting signals buried beneath other signals:
one was synchronous demodulation, and the other was the FM capture
effect. It is well known that FM radio stations are immune to
impulse interference, and the idea was to use this trick to make
the colour channel immune to the luminance channel. So much for
cross colour, but unfortunately, the immunity is not reciprocal.
You can't suppress an FM carrier, so an FM-SECAM system has dots
in parts of the picture where there is no colour, and the dots,
are not related to the line structure. Consequently, a SECAM
signal makes for very poor viewing on a black-and-white TV set
(some would say flatly that it is not black-and-white compatible),
and there are further problems in processing the signal in the
studio.
Studios working in NTSC or PAL can lock all their cameras to
a subcarrier reference. PAL studios also need to lock their cameras
to a line identification reference, so that they all produce
+(R-Y) or -(R-Y) lines at the same time. When this is done, it
is possible to cross-fade between different sources almost as
easily as if they were monochrome. This is fundamental studio
practice, but it can't be done if the subcarriers are FM. If
you mix two FM signals together, you get horrible interference.
The obvious solution was to work with a separate baseband chrominance
channel (you only need one with SECAM), but the pragmatic
solution adopted by many TV companies was to buy PAL equipment,
and transcode to SECAM for final distribution. This is not the
cop-out that it might seem however, because SECAM signals are
very robust in transmission. (Many TV companies, of course, now
use digital systems internally.)
Both the PAL and SECAM systems need to transmit a line identification
reference, to tell the TV what type of chrominance information
is coming next. In the PAL case, this is done by shifting the
phase of the subcarrier reference burst. In the SECAM case, this
is done by sending a reference burst in the vertical blanking
interval (SECAM-V) or in the horizontal blanking interval (SECAM-H).
SECAM-V is the older of the two systems, and the signal can carry
both types of line ident for transitional compatibility with
older sets. The V-ident signal has to go however, if the TV station
wants to transmit subtitles or Teletext. |
S-Video
The point about all of the colour TV standards is that they were
actually conceived as transmission standards. When you
add the colour information to the TV signal, it always degrades
the quality of the basic monochrome picture, so there is really
no need to do it unless you have to send the signal by radio.
It took the video equipment manufacturers a while to grasp this
point, but when they did, they came up with S-Video. Prior
to that, we had to work with composite CVBS (Chroma, Video, Blanking,
and Sync)*, or separate RGB. S-Video (Separated) is just the
C and the VBS in separate cables, but otherwise exactly as they
would have been in composite form. If you want to use a monochrome
video monitor with a colour camera, feed it with the VBS part
of the S-Video, rather than composite, and you will get a picture
free from subcarrier dots.
* CVBS originally stood for 'Composite Video
Blanking and Sync.', but the C came to stand for Chroma by consensus
at some point. |
VHS Video Recording.
If you are exchanging domestic-format videotapes with people
in other countries, the platform of interest is almost certainly
VHS. The following points are therefore pertinent:
1) All 525 line NTSC machines use the same recording format.
2) All 625 line PAL machines use the same recording format.
3) 525 line and 625 line VHS machines use the same scanning geometry,
they just rotate the heads and feed the tape at different speeds;
so they can be made to play back alien tapes if the manufacturer
decides to include the facility. This has led to the development
of special hybrid colour signals (see next section) which can
fool a TV into working at the wrong line standard.
4) All SECAM recordings are not the same.
Video recorders convert the luminance signal into FM, and record
it as diagonal stripes on the tape, one field at a time. The
amount of tape which is fed forward as each stripe is written
depends on the thickness of the head and the speed at which the
drum rotates, which is why 625 (E) and 525 (T) cassettes have
different lengths of tape for a given time duration. The chrominance
is separated off for recording, and is moved to a sub-band below
the luminance, at around 650KHz. PAL and NTSC recorders use a
straightforward heterodyning system to shift the chroma down,
and shift it back on playback by using a fast VCO (voltage controlled
oscillator), which is adjusted by comparing the burst signals
against a local 3.58 or 4.43MHz reference. The VCO system thus
gives time-base correction to the chroma, and protects
the delicate phase information against the vagaries of a mechanical
system (i.e., wow and flutter). There is usually no corresponding
timebase correction for the luminance however, and so diagonal
recordings always have slightly wobbly edges on any verticals
in the picture. This problem can be cured by feeding the video
signal through a box called a Timebase Corrector (TBC).
Some up-market S-VHS players have a TBC built in.
SECAM can be recorded onto standard VHS in one of two ways. It
can either be heterodyned down and back; or since it is FM, it
can be treated as a string of pulses, divided by four to get
it down to the sub-band, and multiplied by four to get it back.
The divide-by-four method is most common. The heterodyne method
is called MESECAM (which I think stands for 'Middle-East'). S-VHS
recorders don't use either of these methods however; they transcode
to PAL for recording, and transcode back to SECAM for playback;
which means that S-VHS is compatible across all 625-line
PAL and SECAM countries (but unfortunately not well established
as a domestic VTR format). |
Hybrid Playback Standards.
NTSC-4.43, PAL-525, and NTSC 625.
These are not transmission standards, although they do come out
of RF modulators. They are used to enable some VCRs to play back
tapes with the wrong line standard. They all exploit the fact
that the 625 and 525 line systems have similar line frequencies
(15625 vs 15734Hz) Thus a monitor or TV can usually sync to either,
with a small tweak of the vertical hold to make up the difference
between 50 and 59.94Hz. The purpose of the hybrid standard is
to get the colour to work as well.
NTSC-4.43 appeared in the 1970s, as a way of enabling
Sony U-Matic PAL VCRs to play back 525 line NTSC tapes. The reproduction
quality is excellent, but the system requires a special type
of monitor.
PAL-525 (Mitsubishi & others), involves recoding the
NTSC signal as PAL, on a 4.43MHz subcarrier. This works with
almost any 625 line monitor or TV, but the decoder delay line
is 0.44 microseconds longer than the actual lines, and this causes
decoding errors at colour boundaries in the picture. The results
are generally acceptable nonetheless.
NTSC-625 is a simple matter of unscrambling the PAL signals
and re-coding as NTSC-3.58. There are no inherent problems other
than that the chroma interleaving is not optimal - which doesn't
matter at all provided that S-Video is used for the link between
the VTR and the monitor.. |
Back to TV standards.
|
