Why is video inverted for transmission?

[Arny Krueger]

I can top that. I helped to maintain a 1500-tube (all but
three of which were dual or triple purpose) analog
computer back a little over 50 years ago. It was part of
a Nike Surface-to-air missile system. The standard first
try repair was to kick it in the area where we suspected
the problem lay, which was effective more often than not.
One day, though, a general was making an announced
inspection tour; on that morning, of course, the computer
failed. We applied the standard fix, but this time the
fixer kicked too hard and caved in one of the doors. The
general came in, looked around and asked "Did that fix
it?" We told that it had, and he said "good!" and walked
out. We hadn't even told him what had happened. And,
incidentally, we called the van it was in the "pizza
oven". You can't believe how much heat 1500 tubes put out
unless you've been there. Viva la Solid State. ]

My only comeback is that I had two of these 400 tube monsters and another
smaller one with only about 200 tubes. Even so my total of 1,000 tubes falls
short of 1,500. BTW, this was the Hawk system.

Do I get bonus points if my monsters were in the open air (sun, rain and
hurricane-force winds), on the top of 60-80 foot towers in the Everglades,
complete with alligator-infested scenery that was mostly under water?

Once I got *tired* of that, they moved me to Germany, where the radars were
still in the open air (sun, sleet, rain, waist-deep snow and subzero temps),
on individualized hilltops. The scenery in Germany included a 270 degree
vista of a large valley, complete with farms, little towns, and a castle or
two.

 

[Don Bowey]

In less than 50 years ignorant people will be commenting on how primitive
things were in the early 2000s. Better educated people will recognize that
they are fortunate that engineers in early electronics history created the
basis for everything they enjoy.

I can top that. I helped to maintain a 1500-tube (all but three of which
were dual or triple purpose) analog computer back a little over 50 years
ago. It was part of a Nike Surface-to-air missile system. The standard
first try repair was to kick it in the area where we suspected the
problem lay, which was effective more often than not. One day, though, a
general was making an announced inspection tour; on that morning, of
course, the computer failed. We applied the standard fix, but this time
the fixer kicked too hard and caved in one of the doors. The general
came in, looked around and asked "Did that fix it?" We told that it had,
and he said "good!" and walked out. We hadn't even told him what had
happened. And, incidentally, we called the van it was in the "pizza
oven". You can't believe how much heat 1500 tubes put out unless you've
been there. Viva la Solid State.
Allen

Our first test was always to darken the room and look for dead tubes.

 

[Richard Crowley]

"Richard Fry" wrote ...

_________

Accurate colour reproduction requires accurate transmission of the
luminance (brightness) value of each colour, which is set by a
specific, DC-coupled voltage.

Well, the luminance ("Y") is defined by the instantaneous
DC voltage of the signal, but none of the color information
is carried there. The color information comes through on
a 3.579545MHz subcarrier and the color hue is defined by
the phase of the signal, while the saturation of the color is
defined by the amplitude of that same signal. A black, white,
or gray area has no associated subcarrier amplitude.

If the video signal was transmitted using AC coupling,
then luminance values would be a function of the
average voltage of the video waveform.

When it comes right down to it, ALL analog video is AC
coupled. Through many stages of electronics, to be sure,
but even if all the electronics were DC coupled, the trip
through the air (or written to and read back from a tape)
is AC coupled. The sync parts of the signal (particularly
the blanking/ "back porch", etc.) make convienent reference
points to "clamp" the signal and perform "DC restoration".

 

[Don Pearce]

Noise in the black portion shows up more than in the white. Black is
transmitted at higher power, more received signal. less noise.

Huh? This is just nonsense. The noise doesn't live at the bottom of
the signal where the white is. The noise accompanies the signal
whatever level it is at. The black will have just as much noise in it
as the white. The reason why small signals appear noisier is that you
have to apply gain to make them bigger - making the noise bigger too.
This is not the case with a TV signal - you don't apply more gain when
the signal drops into the white.

d

 

[Allen]

Arny Krueger wrote:

Once I got *tired* of that, they moved me to Germany, where the radars were
still in the open air (sun, sleet, rain, waist-deep snow and subzero temps),
on individualized hilltops. The scenery in Germany included a 270 degree
vista of a large valley, complete with farms, little towns, and a castle or
two.

You got to see more of the world than I did. My time was spent keeping
the North Korean air force from destroying Pittsburgh.
Allen

 

[Joerg]

_________

Accurate colour reproduction requires accurate transmission of the
luminance (brightness) value of each colour, which is set by a
specific, DC-coupled voltage. If the video signal was transmitted
using AC coupling, then luminance values would be a function of the
average voltage of the video waveform.

For that matter accurate monochome reproduction also requires DC
coupling, but it is not as objectionable if not used (cheap TV set,
etc).

Also - the purpose of inverting video for transmission is to transmit
the peaks of sync pulses at +100% modulation, which allows TV
receivers to show the most stable picture in the presence of noise
(eg, fringes of the coverage area of the TV station).

RF
RCA Broadcast Field Engineer, retired

I know how it works, Richard. My point was, why restore the clamp level
or any other level if the contents of the transmission itself ain't
worth watching? It they did more re-runs of Andy Griffith or Bonanza or
whatever, ok, but not with the average programming these days.

 

[Richard Fry]

Well, the luminance ("Y") is defined by the instantaneous

DC voltage of the signal, but none of the color information
is carried there. The color information comes through on
a 3.579545MHz subcarrier ...

______

Luminance, hue and saturation information are required to define every
color, so it isn't strictly accurate to write that "color information
comes through on a 3.58.. MHz subcarrier." Chrominance information
does, but luminance information is contained in the standard
monochrome video spectrum from about 60 Hz to about 4.2 MHz (NTSC). It
takes both the luminance and the chrominance signals to convey
accurate color.

For example, if the luminance information is removed from a standard
color bar test signal, the color bars will not look the same as before.

 

[Bob Myers]

Huh? This is just nonsense. The noise doesn't live at the bottom of
the signal where the white is. The noise accompanies the signal
whatever level it is at. The black will have just as much noise in it
as the white. The reason why small signals appear noisier is that you
have to apply gain to make them bigger - making the noise bigger too.

Not in the case of video, though - the reason that the blacks
are "more sensitive to noise" than the white is actually because
our eyes are more sensitive to small changes in dark regions
than in light regions. As you noted, though, the noise is going to
hit the blacks just as much as the whites within the signal - but
as others noted, it's better that a brief high-amplitude noise spike
make a "black dot" in a white region than the other way around.
The only part of the standard I'm aware of that was specifically
put in to handle the increased sensitivty to black-region noise
was a modification to the assumed standard response ("gamma")
curve.

And, of course, the other (and possibly the primary) reason for
the inverted transmission was to put the most "oomph" behind
the sync pulses - which means that even those in the fringe areas
can get a stable, albeit snowy, picture.

Bob M.

 

[Gene E. Bloch]

Arny Krueger wrote:

You got to see more of the world than I did. My time was spent keeping the
North Korean air force from destroying Pittsburgh.
Allen

OT (I guess)

My first thought at the start of this thread was "OMG, another radium
thread", but I read a few posts anyway. So:

1. I'm learning something about why black is full signal - actually a
meaningful question after all.

2. I'm enjoying the tales in this part of the thread about pizza ovens
and so forth. I even had a couple of LOLs...

 

[Richard Crowley]

"Richard Fry" wrote ...

______

Luminance, hue and saturation information are required to define every
color, so it isn't strictly accurate to write that "color information
comes through on a 3.58.. MHz subcarrier." Chrominance information
does, but luminance information is contained in the standard
monochrome video spectrum from about 60 Hz to about 4.2 MHz (NTSC). It
takes both the luminance and the chrominance signals to convey
accurate color.

For example, if the luminance information is removed from a standard
color bar test signal, the color bars will not look the same as before.

Yes, you are correct, one cannot decode a full NTSC composite
color signal without the "Y" luminance part. OTOH, the backwards-
compatibility feature of NTSC guarantees that the "Y" luminance
part will stand on its own without the "C" chrominance portion.

 

[Mark]

Other than brief portions of the evening news the question arises:
What's the whole point in restoration these days? If OTA-TV really goes
digital some distant day we might not even bother buying a new set.

Some distant day in the US is Feb 17, 2009.

Mark

 

[Richard Crowley]

"glen herrmannsfeldt" wrote ...

Richard Crowley wrote:

(snip)


On tape, luminance is usually an FM signal, which could be
considered to preserve DC. (If you are careful with the
calibration.)

Right. But even professional equipment relied on downstream
DC restoration and/or clamping rather than absolute accuracy
of the FM calibration.

 

[glen herrmannsfeldt]

Luminance, hue and saturation information are required to define every
color, so it isn't strictly accurate to write that "color information
comes through on a 3.58.. MHz subcarrier." Chrominance information
does, but luminance information is contained in the standard
monochrome video spectrum from about 60 Hz to about 4.2 MHz (NTSC). It
takes both the luminance and the chrominance signals to convey
accurate color.

For example, if the luminance information is removed from a standard
color bar test signal, the color bars will not look the same as before.

If luminance is removed from a monochrome picture it won't look
the same as before, either.

If chrominance is removed from a color picture it should form
a good monochrome picture.

Since gamma correction is done differently for color, the
resulting luminance isn't exactly correct. Given that,
I would have to agree that some chrominance information is
in the luminance signal, but it is relatively small.

-- glen

 

[Joel Kolstad]

Some distant day in the US is Feb 17, 2009.

Looks like it's really going to happen this time too -- all the analog TV
boxes now have the, "Warning! This TV will stop receiving over-the-air
broadcast in 2009!" stickers on them these days.

 

[Bob Myers]

Since gamma correction is done differently for color, the
resulting luminance isn't exactly correct. Given that,
I would have to agree that some chrominance information is
in the luminance signal, but it is relatively small.

The original definitions of the three signals in the NTSC
standard were as follows (they've since changed slightly,
to harmonize with the European-standard "U" and "V"
signals in place of the original "I" and "Q"):

Y = 0.587G + 0.114B + 0.299R
I = -0.274G - 0.321B + 0.596R
Q = -0.532G + 0.311B + 0.211R

Now, it so happens that with this definition, the "Y"
signal is something that works very well for a visually-
acceptable luminance-only ("black and white") image,
but which of these three would you say carries the
most or the least color information? You can't
recover the needed RGB signals without all three.

Bob M.

 

[glen herrmannsfeldt]

Richard Crowley wrote:

(snip)

When it comes right down to it, ALL analog video is AC
coupled. Through many stages of electronics, to be sure,
but even if all the electronics were DC coupled, the trip
through the air (or written to and read back from a tape)
is AC coupled.

On tape, luminance is usually an FM signal, which could be
considered to preserve DC. (If you are careful with the
calibration.)

Otherwise, yes.

-- glen

 

[Joerg]

Looks like it's really going to happen this time too -- all the analog TV
boxes now have the, "Warning! This TV will stop receiving over-the-air
broadcast in 2009!" stickers on them these days.

Yeah but it has been postponed how many times now? I guess this deadline
is more serious though as some spectra have supposedly been auctioned
off. And at least out here no converters in sight. Oh man, I do not want
to be in the shoes of a local politician on Feb-18, 2009.

 

[Jerry G.]

Reverse video to give higher transmission power during black was chosen,
because noise is more visible to the eye in the black levels of video.
With the whites, noise is less visible to the eye. Therefore the higher
transmission power levels in black levels allow for less visible noise
due to any RF carrier signal loss.

It has nothing to do with sync pulses. The sync occurs during the
blanking time at the beginning of each picture frame. The total
horizontal blanking is approximately 10.6 to 10.8 usec. The sync pulse
approximately 4.2 to 4.8 usec wide. The colour colour burst pulse sits
on the back-porch of the blanking.

Excellent web page about the structure of a video signal:
http://zone.ni.com/devzone/cda/tut/p/id/4750


--

Jerry G.

Prolonged blacks can damage television transmitters, however (video
is inverted for transmission, so black requires full power from the
transmitter).

Why is video inverted for transmission?

 

[Randy Yates]

Reverse video to give higher transmission power during black was chosen,
because noise is more visible to the eye in the black levels of video.
With the whites, noise is less visible to the eye. Therefore the higher
transmission power levels in black levels allow for less visible noise
due to any RF carrier signal loss.

It has nothing to do with sync pulses. The sync occurs during the
blanking time at the beginning of each picture frame. The total
horizontal blanking is approximately 10.6 to 10.8 usec. The sync pulse
approximately 4.2 to 4.8 usec wide. The colour colour burst pulse sits
on the back-porch of the blanking.

Excellent web page about the structure of a video signal:
http://zone.ni.com/devzone/cda/tut/p/id/4750

While we're on the subject of TVs, here's a question that had me
going for awhile several years back.

In the old days the television screen, when turned off, was a dark
greenish color. Why is it (or was it), then, that areas of "black" in
a picture look(ed) black and not green?
--
% Randy Yates % "My Shangri-la has gone away, fading like
%% Fuquay-Varina, NC % the Beatles on 'Hey Jude'"
%%% 919-577-9882 %
%%%% <[email protected]> % 'Shangri-La', *A New World Record*, ELO
http://home.earthlink.net/~yatescr

 

[glen herrmannsfeldt]

Bob Myers wrote:
(snip)

The original definitions of the three signals in the NTSC
standard were as follows (they've since changed slightly,
to harmonize with the European-standard "U" and "V"
signals in place of the original "I" and "Q"):

Y = 0.587G + 0.114B + 0.299R
I = -0.274G - 0.321B + 0.596R
Q = -0.532G + 0.311B + 0.211R

Now, it so happens that with this definition, the "Y"
signal is something that works very well for a visually-
acceptable luminance-only ("black and white") image,
but which of these three would you say carries the
most or the least color information? You can't
recover the needed RGB signals without all three.

No, you can't recover RGB without all three.
Yes, Y is a good "black and white" image signal, but
it isn't the correct one. The R, G, and B in the
equations above have been gamma corrected.
If R=G=B then Y is appropriately gamma corrected,
otherwise it isn't the exact gamma corrected Y.
(Hint: (X+Y)**2 is not equal to X**2+Y**2.)

Since RGB has three degrees of freedom, and Y has one
(luminance) that leaves two for chrominance.
That is, chrominance is the amount of information that,
when combined with luminance gives the appropriate color
signal.

(I believe it is usual to use R' G' B' and R G B
gamma corrected and uncorrected signals, though I forget
which one gets the prime and which doesn't.)

-- glen