Basic Monochrome Video Standards.
|
Standard: |
EIA |
CCIR |
|
Lines / field |
262.5 |
312.5 |
|
fields / sec |
59.94 |
50 |
|
Line frequency |
15734 Hz |
15625 Hz |
|
Line period |
63.55ms |
64ms |
|
Active line period |
52.6ms |
52ms |
|
Active lines / field |
241 |
287 |
|
Active lines / frame |
482 |
574 |
|
Aspect ratio w : h |
4:3 |
4:3 |
|
Optimum bandwidth |
6.11 MHz |
7.36 MHz |
|
Transmitted bandwidth |
4.2 MHz |
~5 MHz |
Active periods: Only part of each line carries picture
information, and not all of the lines are used for picture information.
The dead-times are called the blanking intervals, and are used
to transmit synchronising pulses, black level reference, colour
reference burst, and (nowadays) test signals and data. Thus a
"625" line picture really only has 574 lines, and "525"
has 482.
Video bandwidth: There are no hard and fast rules regarding
video bandwidth but, for a given line standard, there is a point
above which the law of diminishing returns sets in, and below
which horizontal blurring starts to occur. Nowadays we might
call this point the 'square pixel bandwidth', and work it out
by determining what is needed to display the same number of pixels
per unit length in both the horizontal and vertical directions.
Taking the CCIR system as an example, there are 574 lines, and
the aspect ratio is 4:3, which gives equal H and V resolution
if the number of horizontal pixels is 765.3. Those 765 pixels
must be transmitted in 52ms, and the
highest frequency video component corresponds to the case when
adjacent pixels are turned on and off alternately. Thus the worst
case signal must change polarity 302.7 times in 52ms,
i.e., it is 7.36 MHz.
Note that in the EIA case, the 'ideal' number of pixels is 643
x 482. In practice, no information is carried right at the edges
of the picture, so this reduces to a practical 640 x 480, which
you may recognise as the basic resolution of a Personal Computer.
TV resolution is sometimes quoted in lines. To display vertical
lines, you must turn adjacent pixels on and off, so 765 pixels
gives 383 lines of horizontal resolution, 574 TV lines gives
apparent sharpness equivalent to 287 lines of vertical resolution
(neglecting aliasing effects).
All of the broadcast TV standards have slightly sub-optimal video
bandwidth, because the founding fathers were worried about the
visibility of the TV line structure and the problem of aliasing.
The point is that for a given bandwidth, you can trade horizontal
resolution for vertical resolution (horizontal resolution
is the ability to display vertical lines, and vice versa).The
problem with TV is that the horizontal and vertical sampling
methods are not equivalent; i.e., you can put a change in brightness
anywhere you like as you move horizontally in the picture, but
as you move vertically, the positions where brightness changes
can occur are constrained by the line structure. Thus, if you
put up a test card of horizontal lines of spacing close to the
system resolution, the lines do not display properly, and have
superimposed on them an undulation in brightness at the difference
in spatial frequency between the test card and the TV raster.
The problem is called aliasing, and is exactly the same as that
encountered in digital sampling. Nowadays we would say that the
TV has no protection against vertical spatial frequencies which
exceed the Nyquist limit, and the pragmatic solution is to use
more lines than the available bandwidth would appear to merit.
By further convoluted argument, we get to the fact that the broadcast
video bandwidth used is approximately equal to the 'square pixel'
bandwidth divided by Ö2 (i.e.,
5.2MHz for 625/50, 4.3MHz for 525/60) . For closed circuit video
work, there is usually not much point in increasing the video
bandwidth beyond the square pixel bandwidth multiplied
by Ö2. |
Why the peculiar numbers of lines?
An interlaced picture must always have an odd number of lines
per frame, because a field must have a half-integer number of
lines in order for the lines of one field to lie halfway between
the lines of the next. It also follows that the line and field
frequencies must be derived from a common reference in order
for this exact relationship to be maintained. The common reference
is twice the line frequency (2fH) which
is divided by 2 to get the line sync, and by the line number
(N) to get the field sync. The development work on electronic
TV was done in the 1930s (interlacing was invented in 1932),
and the circuits used had to be as simple as possible to avoid
the need for enormous numbers of valves (vacuum tubes). Division
by 2 was reliably accomplished by using the Eccles-Jordan circuit
(flip-flop) as it is today, but division by an arbitrary integer
had to be done by means of a critically adjusted non-retriggerable
monostable multivibrator. If the monostable's operating parameters
drifted, the circuit was likely to jump to a new division ratio,
so the early system designers liked to stick to schemes which
depended on divisions by low numbers, i.e., 3, 5, or 7, which
allowed for considerable drift without risk of jumping. Thus
the preferred TV standards (in approximate historical order)
came out like this: |
|
Line number |
Division scheme |
|
|
243 |
3 x 3 x 3 x 3 x 3 |
UK Experimental 1936 |
|
405 |
3 x 3 x 3 x 3 x 5 |
UK Experimental 1936. UK, Eire, 1939 - 1984. |
|
441 |
3 x 3 x 7 x 7 |
Germany 1939, USA (Empire state) 1939. |
|
525 |
3 x 5 x 5 x 7 |
USA |
|
625 |
5 x 5 x 5 x 5 |
Europe |
|
819 |
7 x 9 x 13 |
France (now obsolete) |
|
Field rate: Interlaced TV systems were never locked to
the mains, as is claimed by numerous sources. Field rates were
however chosen to be the same as the nominal mains frequency
so that any 'hum bar' on the picture (due to bad power supply
smoothing) remained approximately stationary. This, unfortunately,
results in a rather low overall refresh rate in 50Hz countries,
and forces flicker sensitive individuals to stare at the screen
to avoid being irritated by it. A solution to this problem has
only recently arrived; in the form of TV sets which perform an
internal standards conversion to 100 fields/sec. |
back to main text.
|
