Questions about structure of NTSC TV video signal

[Videogamer555]

I'm talking about the baseband signal (aka "composite video") here, not the RF broadcast. Specifically, I need to know some info about the VBI, during which their are "equalizing" and "v-sync" pulses (and after these data can be embedded before the first image line).

What I know regarding the VBI portion of a video signal:
There are 3 lines of equalizing pulses.
Then 3 lines of v-sync pulses.
Then 3 more lines of equalizing pulses.
There are 2 equalizing pulses on each line that has them.
There are 2 v-sync pulses on each line that has them.
Each pulse has a high (0 volts) state, and a low (-0.2857 volts) state.
In all of these pulses, the high state comes after the low state.
Equalizing pulses have a short low state, and a long high state.
Sync pulses have a long low state, and a short high state.
The short low state of an equalizing pulse is not the same length as the short high state of a sync pulse.
The long high state of an equalizing pulse is not the same length as the long low state of a sync pulse.

But I'm left with four questions:
How long (accurate to 1/10th of a microsecond) is the duration of the low state of an equalizing pulse?
How long (accurate to 1/10th of a microsecond) is the duration of the high state of an equalizing pulse?
How long (accurate to 1/10th of a microsecond) is the duration of the low state of a v-sync pulse?
How long (accurate to 1/10th of a microsecond) is the duration of the high state of a v-sync pulse?

Please reply and answer these questions. They are key to me completing the video signal generator project I'm working on. I've got the timing down for all of the other parts of an NTSC analog video signal, but I'm missing those four pieces of information. They will be crucial to me completing my project.

[betwixt]

Take a look at this page and the links at the end of it. It is written for PAL but also contians NTSC timings. I found it very useful when designing my own video generator.
http://www.batsocks.co.uk/readme/video_timing.htm

Brian.

[JVRaines]

I think this diagram (which I found here) will answer your questions. "H" is the duration of one horizontal scan line.

[Videogamer555]

I think this diagram (which I found here) will answer your questions. "H" is the duration of one horizontal scan line.


[Attachment 36160 - Click to enlarge]

That diagram appears to be upside down. Equalizing and sync pulses isn't aren't positive-going pulses, but rather negative going. In fact, it's "blacker" than the blanking level, which is itself "blacker than black". Darker parts of an image are represented by lower voltages (but greater than 0 volts), and brighter parts by higher voltages, and the synchronization and blanking signals are negative voltages, or at most 0 volts. I have no idea why this graph appears to show the equalizing pulse and sync pulse being positive.

[Videogamer555]

And I've got another question.

Now regarding VBI line count. How many lines are in the vertical blanking interval?
According this http://martin.hinner.info/vga/ntsc.gif there are 19 VBI lines in each field (though in the 2nd field it's 18 complete lines with half a line added at the beginning and half a line added at the ending of the VBI).

According to http://www.ntsc-tv.com/images/tv/RS-170A.gif there's 20 (not 19) lines in each field (with the second field it's 19 complete lines, and half a line added to the beginning and ending of the VBI)

So which is it? 19 or 20 lines in the VBI interval of a field? I often hear about closed caption text data being added to line 21, but if there's only 20 lines (at most) in the VBI portion of a field, where do they get VBI line 21 from? Do they actually write the closed caption data to the first line of the video image, and just hope you'll have your V-Size setting on your TV such that you won't see the first line?

[Cornucopia]

I'm pretty sure: for NTSC SD, 525 lines= total of analog picture+VBI, 480=total of digital picture 525-480=45. Since interlacing slices the picture in 1/2, 45 / 2 = 22.5 (yes, there is a 1/2 a line in each field). But if you consider some pro A->D converters consider digital active frame to be 486 lines, that would only leave 39 lines (or 19.5 per field). But many also consider the 486 to be capturing some of the non-active picture area, so it depends on who you talk to.

Regardless, line 20 is common for WSS (wide-screen signalling) and other source ID info, Line 21 & 22 is common for eia-608 CC. VITC can be set to a few different lines, but seems to commonly be on 10 and again on 13 or 16 or 18.

This might help: http://www.dvmp.co.uk/digital-video.htm

Scott

[JVRaines]

That diagram appears to be upside down.

It is if you assume it's baseband video. But this is the NTSC standard for radio broadcasting. Up is in the direction of increasing modulation. NTSC was negatively modulated so that brighter pictures, assumed to be more common than dark pictures, require less power at the transmitter.

Now regarding VBI line count. How many lines are in the vertical blanking interval?

There are 19 lines in the VBI. Lines 20 and 21 can be used for nonpicture information because they are in the overscan area.

[Videogamer555]

[QUOTE=JVRaines;2437152]

Now regarding VBI line count. How many lines are in the vertical blanking interval?

There are 19 lines in the VBI. Lines 20 and 21 can be used for nonpicture information because they are in the overscan area.

So are those lines 20 and 21 for the first field or the second field? Or are you talking about lines 20 and 21 for the entire frame after the 2 fields have been combined? How many overscan lines are in the first field? How many are in the second field?
Also if the overscan region doesn't contain image data, nor any other data (such as data bits for closed caption text), is it usually filled out with blanking-level voltage (0 volts), or black-level voltage (0.07 volts)? And for a typical image that has 480 scan-lines, does 480 include the overscan lines, or are there more than 480 lines if you count the overscan lines?

And if you have an even number of lines like 480, how does that work out when you have an extra half-line at the end of the first field and an extra half-line and the start of the second field? Simple arithmetic says that 240.5 *2 = 481, not 480. So how where does this extra line come from? And how does a screen (and video camera) handle increments of "half line" (since each field contains an extra half line)? A screen contains an integer number of lines, when displaying an image, and a camera contains an integer number of lines when capturing an image. So how do these half-lines actually work? I'd think this would cause some major difficulty in both capturing and displaying an image.

[Cornucopia]

In the analog world (525, 625), you have an extra 1/2 line at the start of field 1 and an extra 1/2 line at the end of field 2. But those are part of the blanking, so the active picture is a straightforward even number of whole lines (480, 486, 576).

Those lines were inherent back in the world of constantly-scanning CRTs & Plumbicon tube cameras.

It makes much sense, but is only confusing to those who weren't around it all their lives. In the digital world, there is still VBI, but end-users never have to worry about that.

Look at a past post of mine where I described & showed a Lissajous curve estimation of how these analog devices worked. Remember, they were constantly cycling back around. It just didn't look like that because during blanking the trail was visibly turned off/blank/black.

************************
Overscan is a completely different matter.
Remember, whether in Analog or Digital realms, there is Active Picture & Blanking (for reset/sync). Overscan is a portion of the active picture that is/was considered expendable, because it likely had wavy/garbled/corrupted image in it (due to loose tolerances in electronics/frequencies-used/tape-reproduction mechanics/film-transfers). That is why TVs had a bezel and cinemas/telecines had curtains, to cut off the unsightly outer edges. Pretty unnecessary for most modern, well-produced digital material.

Blanking is/was used only for syncing & resetting (timing & placement) the image for the next line, next frame. And for adding ancillary data (as mentioned before). CC, VITC, IDs, WSS, Colorburst...
Blanking is NOT overscan. Overscan is NOT blanking. Blanking is (predominantly) blank. It is either 0IRE or Negative IRE. The black part of active picture is either 0IRE (PAL, NTSC-J) or +7.5IRE (NTSC-N.Am.).

Scott

[pandy]

There is no such thing as negative voltage in video - symmetrical supply is very rarely used in real video circuits.
Answers for all you questions are provided in proper standard and for NTSC it will be: https://www.itu.int/rec/R-REC-BT.1700-0-200502-I/en

Also if the overscan region doesn't contain image data, nor any other data (such as data bits for closed caption text), is it usually filled out with blanking-level voltage (0 volts), or black-level voltage (0.07 volts)? And for a typical image that has 480 scan-lines, does 480 include the overscan lines, or are there more than 480 lines if you count the overscan lines?

Overscan https://en.wikipedia.org/wiki/Overscan is normal video which may be not visible on display but it is integral part active video i.e. it is same video like any other.

And if you have an even number of lines like 480, how does that work out when you have an extra half-line at the end of the first field and an extra half-line and the start of the second field? Simple arithmetic says that 240.5 *2 = 481, not 480. So how where does this extra line come from? And how does a screen (and video camera) handle increments of "half line" (since each field contains an extra half line)? A screen contains an integer number of lines, when displaying an image, and a camera contains an integer number of lines when capturing an image. So how do these half-lines actually work? I'd think this would cause some major difficulty in both capturing and displaying an image.

This is to create proper interlace - usually digital systems are not capable to create half line and as such one field is longer by 1 line than other (one field has odd number of lines and other field even - long and short).
Interline offset was produced on CRT by capacitor averaging serration/equalization pulses - nowadays it is created artificially by shifting odd field by half in one direction and in complementary field in other direction (that's why real progressive video may jitter vertically after bob on progressive display) http://avisynth.nl/index.php/Bob

[JVRaines]

Let's see if I can add to what the other posters have said.

So are those lines 20 and 21 for the first field or the second field? Or are you talking about lines 20 and 21 for the entire frame after the 2 fields have been combined?

Every line in the interlaced scan has its own number ranging from 1 to 525. The Field 1 image begins on Line 20 and the Field 2 image begins halfway through Line 282. There is no literal "frame" in NTSC, only alternating fields.

How many overscan lines are in the first field? How many are in the second field?

Overscan is not part of the NTSC signal. It was an industry practice adopted by TV set manufacturers. If the picture isn't lined up properly, the viewer will only know when an edge becomes visible. So the CRT was adjusted to scan past the bezel to cover up misalignments. Broadcasters counted on up to 10 percent of the picture being invisible this way; it relieved them from making constant, fine timing adjustments and allowed them to send data on the hidden image lines.

And for a typical image that has 480 scan-lines, does 480 include the overscan lines, or are there more than 480 lines if you count the overscan lines?

NTSC has 487 image lines according to the standard. One of them is actually two half-lines, which (as noted above) is rather awkward to deal with digitally. Professional digital video samples 486 lines. DV reduced the number to 480 because it's friendlier to computer processing.

[Videogamer555]

Let's see if I can add to what the other posters have said.

So are those lines 20 and 21 for the first field or the second field? Or are you talking about lines 20 and 21 for the entire frame after the 2 fields have been combined?

Every line in the interlaced scan has its own number ranging from 1 to 525. The Field 1 image begins on Line 20 and the Field 2 image begins halfway through Line 282. There is no literal "frame" in NTSC, only alternating fields.

How many overscan lines are in the first field? How many are in the second field?

Overscan is not part of the NTSC signal. It was an industry practice adopted by TV set manufacturers. If the picture isn't lined up properly, the viewer will only know when an edge becomes visible. So the CRT was adjusted to scan past the bezel to cover up misalignments. Broadcasters counted on up to 10 percent of the picture being invisible this way; it relieved them from making constant, fine timing adjustments and allowed them to send data on the hidden image lines.

And for a typical image that has 480 scan-lines, does 480 include the overscan lines, or are there more than 480 lines if you count the overscan lines?

NTSC has 487 image lines according to the standard. One of them is actually two half-lines, which (as noted above) is rather awkward to deal with digitally. Professional digital video samples 486 lines. DV reduced the number to 480 because it's friendlier to computer processing.

If the input video contains 486 lines, are those extra 6 lines considered part of the overscan? And how are these 6 lines divided between the 2 fields? In generating a video signal with digital equipment, if it only handles 480 lines, how does analog equipment play it correctly if it's missing those extra 6 or 7 lines? And if the digital equipment is taking in 486 or 487 lines from an analog source, does it just drop the extra 6 or 7 lines?

[Videogamer555]

This is to create proper interlace - usually digital systems are not capable to create half line and as such one field is longer by 1 line than other (one field has odd number of lines and other field even - long and short).
Interline offset was produced on CRT by capacitor averaging serration/equalization pulses - nowadays it is created artificially by shifting odd field by half in one direction and in complementary field in other direction (that's why real progressive video may jitter vertically after bob on progressive display) http://avisynth.nl/index.php/Bob

So if digital systems don't create an extra half line, how do they handle generating lines that are supposed to contain half-picture and half-blanking? Is that line usually set to all blanking (and synchronization pulse if it would normally contain that)? Or is it set entirely to active picture? Or do they just skip generating these lines, so that the total number of lines generated by a digital system is actually less than 525?

[pandy]

This is to create proper interlace - usually digital systems are not capable to create half line and as such one field is longer by 1 line than other (one field has odd number of lines and other field even - long and short).
Interline offset was produced on CRT by capacitor averaging serration/equalization pulses - nowadays it is created artificially by shifting odd field by half in one direction and in complementary field in other direction (that's why real progressive video may jitter vertically after bob on progressive display) http://avisynth.nl/index.php/Bob

So if digital systems don't create an extra half line, how do they handle generating lines that are supposed to contain half-picture and half-blanking? Is that line usually set to all blanking (and synchronization pulse if it would normally contain that)? Or is it set entirely to active picture? Or do they just skip generating these lines, so that the total number of lines generated by a digital system is actually less than 525?

Instead creating two fields with 262.5 line, one field have 262 lines (short) and second field has 263 lines (long), analog equipment average over time video as such proper 525 lines frame is retained. Block responsible for synchronization at receiver side is called sync slicer/sync separator/sync processor/sync stripper etc and it is capable to distinguish between odd and even fields and as such return proper time structure of video.

Half line offset is created in are something called serration/equalization pulses http://www.ntsc-tv.com/ntsc-index-02.htm look at those 5 diagrams - notice subtle duty ratio differences between each field (there are 4 different fields in fact - this is important for colour processing and frequently is called 4 field sequence - as a curiosity - there is even MPEG official syntax to mark every field phase inside codec)






Blanking is every signal/time that don't contain active video signal (itself active video signal may have normally visible and normally invisible part - normally invisible part is so called overscan and as already explain this normally not visible but active video purposed is to relax various devices timing requirements - this is valid for broadcast/production studio and for TV receivers) , some of blanking lines are used to carry additional information - as overall it is more correct to use blanking period as it cover both line and sync pulses - remember there is 525 Horizontal sync pulses per picture and there is few Vertical sync pulses (as such those lines used for Vertical sync can't be used to carry any other information than synchronization).

[Brad]

In the analog world (525, 625), you have an extra 1/2 line at the start of field 1 and an extra 1/2 line at the end of field 2. But those are part of the blanking, so the active picture is a straightforward even number of whole lines (480, 486, 576).

I don't believe I've ever been in a position to correct you before.

The "active" picture area begins and ends on the half-lines. PAL is 287.5 * 2 which becomes 288 * 2 for digital. NTSC is a lot weirder, but when there are sufficient active lines from an analog source, you can see the same idea. When the source is 480-line digital, the padded half-lines are just fully "black" all the way across, so in that case I guess you could say they are part of blanking.

PAL DVD. Note the blackness at top left and bottom right.


486-line LaserDisc capture. One row of codes, one row of blackness, one half-line, 482 full rows, final half-line.


Possibly of interest to some reading this thread: https://forum.videohelp.com/threads/359690-Geek-fun-with-VBI-Macrovision-DVD-recorders-...TSC-scan-lines

[Cornucopia]

I should have said, "for NTSC analog video, the extra 1/2 line starts at the beginning of the even field in the top center and the other 1/2 line ends at the end of the odd field at the bottom center". Depending on whom you believe, active picture for analog NTSC could have been anywhere from 480 WHOLE lines (with 22.5 blanking lines per field) through 487 lines (with 19 blanking lines per field). Supposedly valid equipment sources, including manufacturers Techtronix and National Instruments, still vary in their listing, with 483 and 485 being common.
But that was the nature of the analog beast - there was a little room for error/variation.

Digital changed that: Active picture is all whole-line interlaced or progressive fields, either 486 or 480, depending upon the spec.

Wasn't intending to refer to PAL at all (OP was asking about NTSC).

This is a good resource with lots of (redundant) visuals: http://www.infocellar.com/television/scanning.htm

Scott

[JVRaines]

If the input video contains 486 lines, are those extra 6 lines considered part of the overscan? And how are these 6 lines divided between the 2 fields? In generating a video signal with digital equipment, if it only handles 480 lines, how does analog equipment play it correctly if it's missing those extra 6 or 7 lines? And if the digital equipment is taking in 486 or 487 lines from an analog source, does it just drop the extra 6 or 7 lines?

They aren't "extra" lines. They are a required part of the NTSC signal. It's only certain samplers/formats which drop them because 480 (being divisible by 8 and 16) is more efficient in the microprocessor world. I'm sure the half-lines are always dropped, but you'd have to examine the A/D code to find out which 6 full lines are thrown away. They could be all at the top, or all at the bottom, or divided up. Whichever way, they certainly fall into the nominal overscan area.

D/A converters must create a valid signal, so they simply output black on the missing lines.

[JVRaines]

This is a good resource with lots of (redundant) visuals: http://www.infocellar.com/television/scanning.htm

And here is a good primer from National Instruments.

[pandy]

Yep, for today there are only 2 consumer SD video standards 525 lines and 625 lines. 480 and 576 lines is only for digital video as 480 can be divided by 16 and 486 not and based on latest standard seem that 525 line video has 483 active video lines not 486.

[JVRaines]

To be perfectly clear, NTSC has 486 full lines of video. Two or three of them are used, by convention, for data services such as closed captioning and teletext. But these are SMPTE standards, not NTSC. Those lines could be used for video instead. They are not part of the vertical blanking interval, no matter what Wikipedia says.

[pandy]

To be perfectly clear, NTSC has 486 full lines of video. Two or three of them are used, by convention, for data services such as closed captioning and teletext. But these are SMPTE standards, not NTSC. Those lines could be used for video instead. They are not part of the vertical blanking interval, no matter what Wikipedia says.

SMPTE 170M-2004:

Vertical blanking
(63.556... µs × 20 lines
+ 1.5 µs = 1272.62... µs
[see notes]
20 lines
plus
1.5 µs 1)
± 0.1
lines
µs

Some component equipment may not blank lines 20 and 282, resulting in 19-line (+ 1.5 µs) vertical blanking if signals
from such equipment are directly encoded into NTSC.

ITU-R BT.1700:
525 line system (not NTSC but PAL M) Number of active lines 483.