NTSC Modulation Standard

━━━━━━━━━━━━━━━━━━━━━━━━ This document is very popular.

The Impressionistic Era of TV. Please send comments to:
It's Never The Same Color! jsgil AT amstzone DOT org

The first analog Color TV system realized which is backward compatible with the existing B & W signal. To combine a
Chroma signal with the existing Luma(Y)signal a quadrature sub­carrier Chroma signal is used. On the Cartesian grid
the x & y axes are defined with B−Y & R−Y respectively. When transmitted along with the Luma(Y) G−Y signal can be
recovered from the B−Y & R−Y signals.

Matrixing
━━━━━━━━━

Let:

R = Red \
G = Green Each range from 0 to 1.
B = Blue /

Y = Matrixed B & W Luma sub-channel.

U = Matrixed Blue Chroma sub-channel. U #2900FC 249.76° −U #D3FC00 69.76°
V = Matrixed Red Chroma sub-channel. V #FF0056 339.76° −V #00FFA9 159.76°
W = Matrixed Green Chroma sub-channel. W #1BFA00 113.52° −W #DF00FA 293.52°
HSV HSV
Enhanced channels: Hue Hue
I = Matrixed Skin Chroma sub-channel. I #FC6600 24.29° −I #0096FC 204.29°
Q = Matrixed Purple Chroma sub-channel. Q #8900FE 272.36° −Q #75FE00 92.36°

We have:
Y = 0.299 × R + 0.587 × G + 0.114 × B
B − Y = −0.299 × R − 0.587 × G + 0.886 × B
R − Y = 0.701 × R − 0.587 × G − 0.114 × B

G − Y = −0.299 × R + 0.413 × G − 0.114 × B

= −0.194208 × (B − Y) −0.509370 × (R − Y) (−0.1942078377, −0.5093696834)
Encode:
If: U[x] = 0.492111 × ( B − Y ) × 0° ┐ Quadrature (0.4921110411)
V[y] = 0.877283 × ( R − Y ) × 90° ┘ Sub-Carrier (0.8772832199)
Then: W = 1.424415 × ( G − Y ) @ 235.796°
Chroma Vector = √ U² + V²
Chroma Hue θ = aTan2(V,U) [Radians]
If θ < 0 then add 2π.[360°]
Decode: SyncDet
U: B − Y = -┼- @ 0.000° ÷ 0.492111
V: R − Y = -┼- @ 90.000° ÷ 0.877283
W: G − Y = -┼- @ 235.796° ÷ 1.424415 (1.4244145537, 235.79647610°)
or
G − Y = −0.394642 × (B − Y) − 0.580622 × (R − Y) (−0.3946423068, −0.5806217020)

These scaling factors are for the quadrature Chroma signal before the
0.492111 & 0.877283 unscaling factors are applied to the B−Y & R−Y axes
respectively.
 100% Color Bars Composite Luma & Chroma
━━━━━━━━━━━━━━━━━ 0.492111×(B−Y) & 0.877283×(R−Y)

1⅓ ▶┌──┬──┐
┌──┐◀ 1.000 ┌──┤ ┊ ├──┐◀ 1.177453
│ └──┐◀ 0.886 │ ├──┤ ┊ ├──┐◀ 1.003453
│ └──┐◀ 0.701 │ ┊ ├──┤ ┊ ├──┐◀ 0.931333
│ └──┐◀ 0.587 0.438667 ▶└──┤ ├──┤ ┊ │
Y │ └──┐◀ 0.413 │ │ ┊ ├──┤ ├──┐◀ 0.561333
│ Luma └──┐◀ 0.299 0.068667 ▶└──┤ ┊ ├──┤ ┊
│ └──┐◀ 0.114 │ ┏▶└──┤ ┊ ├──┤
──┘ └─── ◀ 0.000 ▶ ─┘ ┃ ┏▶└──┤ ┊ ├───
Wh Yl Cy Gr Mg Rd Bl Bk −0.003453 ┃ └──┴──┘◀ −⅓
−0.177453
0.886 ▶┌──┐
0.587 ▶┌──┐ │ │ ┌──┐ ┌──┐ ┌──┐ ┌──┐ ◀ 1
0.299 ▶┌──┐ │ │ │ │ │ │ │ │ │ │ │ │
B − Y ───┐ │ │ │ │ │ └───◀ 0.0 B │ │ │ │ │ │ │ │
(U) │ │ │ │ └──┘◀ −0.299 │ │ │ │ │ │ │ │
│ │ └──┘◀ −0.587 ┘ └──┘ └──┘ └──┘ └──◀ 0
└──┘◀ −0.886

┌──┐◀ 0.701
0.587 ▶┌──┘ │ ┌─────┐ ┌─────┐ ◀ 1
0.114 ▶┌──┐ │ │ │ │ │ │
R − Y ───┘ │ │ │ ┌───◀ 0.0 R │ │ │ │
(V) │ │ └──┘◀ −0.114 │ │ │ │
│ ┌──┘◀ −0.587 ┘ └─────┘ └─────◀ 0
−0.701 ▶└──┘

Wh Yl Cy Gr Mg Rd Bl Bk Wh Yl Cy Gr Mg Rd Bl Bk

0.413 ▶┌──┐
0.299 ▶┌──┘ │ ┌───────────┐ ◀ 1
0.114 ▶┌──┘ │ │ │
G − Y ───┘ │ ┌───◀ 0.0 G │ │
(W) │ ┌──┘◀ −0.114 │ │
│ ┌──┘◀ −0.299 ┘ └───────────◀ 0
└──┘◀ −0.413

Color Luma Chroma Levels Chroma Chroma

Bar Level 0.492111×U 0.877283×V Hue θ Peak Level
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
White 100.0% N/A N/A N/A N/A
Yellow 88.6% −0.436010 +0.100010 167.08° 0.447333
Cyan 70.1% +0.147141 −0.614976 283.46° 0.632333
Green 58.7% −0.288869 −0.514965 240.71° 0.590453
Magenta 41.3% +0.288869 +0.514965 60.71° 0.590453
Red 29.9% −0.147141 +0.614976 103.46° 0.632333
Blue 11.4% +0.436010 −0.100010 347.08° 0.447333
Black 0.0% N/A N/A N/A N/A

The Chroma scaling for the colors with full saturation produces a minimum peak level of 0.4473 for the Yellow−
Blue axis and a maximum peak level of 0.6323 for the Cyan−Red axis while the Green−Magenta axis is in the
middle with 0.5904. When modulated the p­p levels are 0.8947, 1.2647, & 1.1809 respectively. When
combined with Luma the Luma + Chroma peak for Yellow & Cyan is at +133⅓% and Red & Blue is at -33⅓%.
After scaling the degree of separation between the MRYGCB color axes and their amplitudes is made even more unequal
as shown in the vector image on page 4.

When the B−Y axis portion is added to the Luma the Yellow positive peak produced peak levels exceeding maximum
signal levels and the negative peak levels for Blue exceeded sync levels thus interfering with syncing so this axis has
been reduced by a factor of 0.492111. This greater level of reduction compared to R−Y is needed due to a value of only
0.114 of the Blue signal used to create the Luma signal. This has a double impact in that the Blue percentage only
subtracts 0.114 from the Luma level of 1 placing the Luma level at 0.886 for the Yellow portion of the Chroma
sub­carrier to be biased with and for the Blue portion only adds 0.114 to the black level to be biased with. Also when
B−Y is generated the low percentage of Blue within the Luma does not reduce Y by much for Yellow & Blue peak
modulations thus making it larger in amplitude compared to R−Y.

The same holds true for the Cyan−Red axis but to a lesser extent. For Cyan 0.299 is subtracted from the Luma and for
Red 0.701 is subtracted leaving Luma signal levels for Cyan & Red at 0.701 & 0.299 respectively for biasing
requiring only a 0.877283 reduction for R−Y. This puts the Cyan−Red axis peak levels at the same peak levels as the
Yellow−Blue axis in the composite signal as seen in the composite image.

After the B−Y & R−Y axes scaling the Green−Magenta axis levels produced within the quadrature Chroma sub­carrier
are somewhere in between the Yellow−Blue and Cyan−Red axes levels. The Luma levels for Green & Magenta are
centered around 50% of the Luma at 0.587 & 0.413 respectively for biasing and does not produce any peak levels
exceeding maximum signal level modulation so no adjustment is needed.

Since NTSC is required to be compatible with the existing B & W receivers and fit within the 6 mHz channel allocation
this did not leave much bandwidth available for the Chroma signal so maximizing signal quality is greatly needed. It was
discovered that vision of the eye is less sensitive to color changes than it is to brightness changes thus allowing a lower
fidelity color signal transmitted in relation to the B & W signal without being noticed. The B & W portion would have a
maximum bandwidth of 4.2 MHz while the highest color fidelity would be 35% of that at 1½ MHz. The eye is also more
sensitive to the flesh tones than to the other colors so the I & Q method, In phase and Quadrature alignment, was devised
where the I channel would carry the oranges where the flesh tones are and would have a higher bandwidth for the lower
sideband at 1.5 MHz and the upper sideband would be vestigial with a 500 kHz bandwidth. The Q channel where the
purples are would have both its upper and lower sidebands limited to a 500 kHz bandwidth. The total bandwidth of the
Chroma signal is 2 MHz. The I & Q channels are usually matrixed directly from the Red, Green & Blue signals for
transmission and band limited to 1½ MHz & 500 kHz respectively before being sent to the quadrature modulators. A
ColorBurst signal is added that is 57° away from the I channel at 180°. I & Q can also be obtained from the U & V
signals which represent the B−Y & R−Y signals respectively with the following formulas:
Skin (I) 123° ( U × Cos(123°) + V × Sin(123°) )
Purple (Q) 33° ( U × Cos( 33°) + V × Sin( 33°) )
To derive I & Q directly from Red, Green, Blue, and since Y = 0.299×R + 0.587×G + 0.114×B ,
substituting Y with the scaled Red, Green, Blue, values into 0.492111×(B−Y) & 0.877283×(R−Y) and
substituting these into the equations above and solving for Red, Green, Blue,will give the scaling factors for each
color.

U = 0.492111 × (B − Y) 'x' V = 0.877283 × (R − Y) 'y'

= 0.492111 × [−0.299 −0.587 +0.886] = 0.877283 × [+0.701 −0.587 −0.114]

I = 0.492111 × Cos(123°) × (B − Y) Q = 0.492111 × Cos(33°) × (B − Y)

+0.877283 × Sin(123°) × (R − Y) +0.877283 × Sin(33°) × (R − Y)

I = −0.268023 × [−0.299 −0.587 +0.886] 'Ux' Q = 0.412719 × [−0.299 −0.587 +0.886] 'Ux'
+0.735751 × [+0.701 −0.587 −0.114] 'Vy' +0.477803 × [+0.701 −0.587 −0.114] 'Vy'
(0.5959007249 −0.2745567667 −0.3213439582) (0.2115366883 −0.5227362571 0.3111995688)
I = 0.595901×Rd −0.274557×Gr −0.321344×Bl Q = 0.211537×Rd −0.522736×Gr +0.311200×Bl
In the vector image below it can be seen that the B−Y axis is compressed in amplitude and expanded in Hue layout
compared to the R−Y axis which is compressed in Hue layout and expanded in amplitude because B−Y axis has been
reduced to 56.1% of the the R−Y axis level creating a tall hexagon using the MRYGCB points that has been squashed on
each side. This means the Yellow−Blue axis is affected more by noise in regards to saturation level and less to Hue
changes but the opposite is true for the Cyan−Red axis and to a lesser extent the Green−Magenta axis since it is about
half the distance away from the R−Y axis as it is from the B−Y axis. For transmission and reception this does not have a
big detrimental effect and may be a benefit since the eye is less sensitive to amplitude and phase variations to the colors
centered around the Yellow−Blue axis very near to the B−Y axis compared to the colors centered around the Cyan−Red
& Green−Magenta axes which are closer to the R−Y axis.
NOTE: In the vector image above it also shows the ColorBurst for PAL, the German adaptation of NTSC where the R−Y
axis is phase inverted on every other line but the scaling for R−Y & B−Y is the same. In Europe channel spacing is 8
MHz, the field refresh rate is 50 Hz vs. 60 Hz for NTSC allowing higher resolution and the Chroma sub­carrier is at
4.43 MHz instead of 3.58 MHz for NTSC. There are several variations of PAL and PAL-M used in Brasil is basically
U.S. NTSC with 6 MHz channel spacing but R−Y is phase inverted every other line. They would have used NTSC since
they were already on the U.S. B & W standard but Philips and Telefunken persuaded them to Pay Another License.

Any vector noise added to the Chroma signal for colors centered around the B−Y axis will have more of an effect on
amplitude than phase since peak levels are lower and the Hue layout is expanded. For colors near the R−Y axis the
opposite is true but the signal is stronger for peak levels and noise does not change saturation much nor does it affect the
Hue much either.

However it is a different story for VCR recordings where phase jitter can be a problem with the Chroma signal regardless
of the amplitude. The colors where the Hue layout is compressed around the R−Y axis the detrimental effects are greater
and can be objectionable but the colors near the B−Y axis where the Hue layout is expanded phase jitter has less of an
effect and the eye is less sensitive to Hue and amplitude changes for these colors. VHS and probably other formats use the
color under method where the Chroma signal is hetrodyned down to around a 650 kHz carrier since phase jitter is less of
an issue in the lower frequencies of the recording medium. The unequal shape of the Chroma hexagon also does not
optimize peak tape saturation levels for colors near the B−Y axis.

[+100] ┌──┐◀━━ 1.2202381 [+130⅚] Values in [ ] are IRE Levels.

1.0000 ━▶ ┌──┤ 1 IRE = 1/140 = 5/700
White │ └──┐◀━━ 0.924679
Level │ └──┐◀━━ 0.802446 One Horizontal line.
│ └──┐◀━━ 0.727125
│ └──┐◀━━ 0.612161
│ 0.536839 ━━▶└──┐ 0.3393 [+7.5]
│ 0.414607 ━━▶└──┐ ┃
━▶ ─┘ ├───┐◀┛ ┌─── ◀┓ 0.0536
0.3393 [+7½] 0.11904762━▶└──┘ ┏▶└┐ ┌─ ─┘ ── ◀┛ SetUp [7½]
Black Level [─23⅓] 0.2857 │ │
[0] │ │ Color Burst [±20]
Voltage Levels Blank └─┘◀┓
│ Horizontal Sync 0 [─40]
Begin New Line│

Composite = (Luma + Chroma) × 0.66071429 + 0.33928571 (Sync + SetUp [47½])

[92.5]
For a 1Vp­p B & W video signal with sync 0.6607 composite scaling is used with Chroma levels of 0.5911Vp­p
minimum peak and 0.8356Vp­p maximum peak depending on the color. Blanking level is exactly 2/7 V [-40].
ColorBurst is ±1/7 V [±20], centered on blanking level, 1/7 V [-20] to 3/7 V [+20].

Anatomy of a Vertical Sync Pulse

has been moved to page 29 with timing annotations added.
 Specifications for a 6MHz Channel Space
To the right is the spectrum layout within the 6MHz channel
space. The bandwidth of a double sideband signal would waste
spectrum so a vestigial sideband signal is used. As long as the
lower frequencies of the signal are represented by both sidebands
the higher frequencies can be represented by only one sideband
without any detrimental effects. This is also used for the I channel
of the Chroma signal. Since quadrature modulation requires both
sidebands to carry 2 channels the Q channel bandwidth is limited
to ±500kHz modulation but the I channel has an extra +1MHz
added to the LSB to extend the 500kHz within the quadrature
sub­carrier for a 1½MHz bandwidth which is the channel that
handles flesh tones. The sound is on a 4½MHz sub­carrier that
can handle 3 separate channels of audio, L+R, L−R, and SAP.
For vestigial sideband modulation method see NTSC­DVD
General:
Aspect Ratio 4:3 = 1⅓ Usable Aspect Ratio 16:9
Total Picture Pixels (Digital) 640×480 ; 307200 Pixels NTSC 4.43
Analog Resolution (Kell Factor) 452×340 ; 153600 Pixels (Studio)
Broadcast (½ Contrast) 344×340 ; 116817 Avg. ; 413×340, 140181 Max.
Vertical: Pixel Aspect 1.315:1
Frames Per Second 29.97Hz (PsF) 3:2 DVD 1.191:1
Frame Period 33.3667ms 16:9 4.43 1.210:1
Total Lines Per Frame 525
Picture Lines Per Frame 480 Ideal Size
Field Sweep 59.94Hz
Field Period 16.68335ms 20”×15”⇒25”
Total Lines Per Field 262½ W H Diag
Picture Lines Per Field 240 794µm Line Pitch
Lines Per Blank 22½ 32 Lines/Inch
Blank 1.43ms
Sync 190.6µs ; 3 Lines 599 ; 1.009
Horizontal: 499 427½ 513 ; 0.992
Resolution ; Pixel Aspect Avg: 344⅙ ; 1.315 Max: 413 ; 1.096 (@-9dB)
Line Sweep 15.734264kHz (563) ; 52.6055µs (466)
Line (HP) ; Picture Period 63.55556µs (455) ; 52.6254µs (376¾)
Picture BW Pixels 368⅜≈1⅔×YBW×(HP−HB) ; (344⅙+24⅕)≈6⅔% OverScan, 3½µs
Blank (HB) ; Active Picture 10.930µs (78¼) 10.922 (96¾) (2×352=704) 49.168µs
Front Porch 1.502µs (10¾) 1.496 (13¼) (2×360=720) 50.287µs
Sync 4.714µs (33¾) 4.713 (41¾) (2×442=884) 49.896µs
Back Porch 4.714µs (33¾) 4.713 (41¾)
Luma & Chroma: ↑Chroma ½Cycles
Luma (Y) BandWidth 6 51 4⅕MHz Luma/Chroma BW Ratio 5⅗:2:⅔ Gamma 2.2
Chroma: 28.63636048 (8×) Y I Q
Sub-Carrier 3.57954506MHz 4.429195316 (4.43)
½H Odd Harmonic 455 (227½) 563 (281½)
I Bandwidth 1½ 1½MHz (120)
Q Bandwidth 1½ ½MHz ( 40) Chroma Cycles
Color Burst 2.709 2.794µs ; 10 Cycles ; 2×{1½+10+5⅜}=33¾ ←Fully optimized
Baseband Guard 2½ 2MHz 12 2×{2 +12+6⅞}=41¾ ← within specs.
MTS Sound: pg.18 Brzwy+Snc+Snc2Blnk
Carrier 4.4999995MHz FM ±25kHz, ±50kHz, ±73kHz
H Harmonic 286 (L+R) add(L−R) add(SAP)
L+R Equalization 75µs Pre-Emphasis
L−R Sub-Carrier 2×H 31.468528kHz AM DSB Suppressed Carrier
Encoding/Compression Zenith-dbx (THAT Corp.)
To extract test patterns from document click to highlight and copy, open image program and paste from clipboard.
Test pattern licenses on the next pages are ©2017 A.R.R. & Creative Commons (CC BY-ND 4.0)
↓640×480↓ ↓↓ Chroma LoR/Freq: 49½/503kHz, 123⅘/1.259MHz for [640|704|720]×480
704×480

361.89kHz
(35⅗)

519.23kHz
(51)

723.78MHz
(71⅙)

1.0227MHz
(100½)

1.4476MHz
(142⅓)

2.0455MHz
(201⅛)

2.8951MHz
(284⅔)

4.0909MHz
(402¼)
720×480

314.865kHz
(47¾)

597.902kHz
(60⅛)

849.650kHz
(85½)

1.19580MHz
(120¼)

1.68357MHz
(169⅓)

2.39161MHz
(240½)

3.36713MHz
(338⅝)

4.77273MHz
(480)
 NTSC Chrominance Locked Test Patterns
The test patterns on page 13 & 14 are 704 & 720 pixels wide even though the displayed aspect
ratio is nominally 4:3 (640×480). This also corresponds to the 4× sample rate of the Chroma
sub­carrier, 4×3.579545MHz=14.31818MHz. It is necessary to have the sample rate of the
Luma signal at a multiple of the Chroma sub­carrier frequency and at 4f it is Chrominance locked
so that the Chroma signal within the composite video signal is sampled every 90°. This 4f
frequency period is 69.84ns in length and represents the width of the 1 pixel sample. The
horizontal period is ~63.5µs=1/15.734kHz and the horizontal blank is ~10.9µs leaving 52.66µs for
the active picture area. Allowing for the standard 6.6% overscan, 3½µs (6.632%), this leaves
49.1683µs for the actual image, so 49.1683µs÷69.84ns=704 pixels, 176 Chroma cycles. For
standard TV broadcast using the standard 6.6% overscan and sampling the horizontal line at 704
pixels allows the Chroma signal, sampled at the 90° incremented ±I & ±Q marks, to be seamlessly
merged with the Luma to create the
composite video signal. If not using
the ±I,1½MHz / ±Q,½MHz dual
bandwidth setup then the 90° sample
points would be at ±U & ±V. Having
the master clock at 14.31818MHz
allows the development of the
quadrature oscillator for Chroma
signal generation. For digital
encoding in the studio all timing
components of the composite signal
are synchronized to the master clock
and for digital decoding in the
receiver the 14.31818MHz 4f
Chroma oscillator is PLL locked to
the incoming Colorburst signal on the
horizontal back porch. The 6.6%
overscan defined in the early days of
NTSC was needed to allow for the
limitation of older transmitter/receiver
technology and drifting so they would
function properly. As technology has
advanced a smaller amount of
overscan can be used and reducing it
to 4½%, 2⅜µs, allows for 720
chrominance locked Luma samples,
180 Chroma cycles @ 50.286µs, to
be used for a horizontal line to
support the 720×480 DVD format.

854×480 (~50µs) ⇒
 PAL
(Der SystemBruch)
PAL is a modified form of NTSC. It addresses the drifting Hue issues that are present in NTSC giving inaccurate
colors when the phase tracking of the Chroma decoder is in error. This is accomplished by inverting the phase
of the R−Y channel on every other horizontal line hence the name Phase Alternation Line. Any decoding phase
errors will cancel out visually on the screen in the PAL Simple decoding mode. PAL Simple mode also has the
effect of creating what is called Hanover Bars where under severe phase decoding errors two scan lines of a full
frame will have its Hue shifted in one direction and the next two will have its Hue shifted in an equal but
opposite direction. The visual addition of the two sets will produce the near perfect Hue as the eye's color
resolution is less than what it is for the B & W Luma portion of the image. Depending on how severe the phase
error is and the viewing distance from the screen they may or may not be noticeable. The greater the error the
greater the viewing distance is needed to have the eye blend them together and not be noticed. A more advanced
decoder uses a delay line of 1H or 1/FH to electronically blend two lines together before being put on screen
eliminating Hannover Bars. The enhanced version of this controls the delay time by a chrominance lock of so
many cycles for the perfect and most accurate delay. Both methods reduce color saturation levels when phase
decoding errors occur and the greater the error the greater the saturation reduction. This is more acceptable to
the viewer than the wrong Hue.
In NTSC each horizontal line ends with ½ Chroma cycle which causes the clusters of Chroma energy to fall in
between the clusters of Luma energy. In PAL the phase inversion of every other line of R−Y by H/2 effectively
modulates it with a square wave smearing its spectrum and creating sidebands of ±H/2 causing R−Y energy
clusters to fall directly on top of the Luma culsters causing interference. The solution to this is to adjust the
Chroma sub­carrier frequency so that each horizontal line ends with either ¼ or ¾ of a cycle of the Chroma.
Having the sub­carrier frequency end with ¼ or ¾ of a cycle and not ½ of a cycle does not cause interference
with the Luma even on the fine mesh level for B−Y. When R−Y is phase inverted at the H/2 rate its modulated
sidebands are fall on the ¾ cycle marks when the line ends with ¼ cycle of the Chroma sub­carrier and when
the sub­carrier ends with ¾ of a cycle the R−Y modulated sidebands fall on the ¼ cycle marks.
The downside to not ending the line with ½ cycle of the carrier breaks the dot pattern system of NTSC when the
Chroma signal is super­imposed onto the Luma which is designed to average out the Luma brightness for each
spot on screen both vertically and temporally. The vertical inversion breaks up the vertical stripes which is
realized by the lines ending in ½ cycle and the temporal inversion is created by having an odd number of lines
per frame. Ending the lines with ¼ or ¾ of a Chroma cycle creates a very noticeable dot pattern motion. In
NTSC this pattern repeats over two frames inverting on every frame and is unnoticeable to the eye when viewed
at a distance. In PAL this pattern repeats over 8 fields or 4 frames and produces a visible slanted vertical line
pattern that can move either to the left or right. To counteract this visible dot pattern motion the number of fame
cycles per second is added to the Chroma frequency to cause the Chroma phase to invert 180° at the beginning
of every field from its normal repeat pattern to simulate a similar effect as NTSC. This maintains a 4 frame
repeat pattern and a motion that is not visible. Adding the frame rate to the Chroma frequency creates phase
creep ; 4 × the Number of Frame Scan Lines is the number of unique Luma/Chroma scan line combinations, a
digital coding nightmare where NTSC has only, Two. For PAL[-N] this 180° inversion can also be realized by
using 621 scan lines without creating phase creep producing only Four line combinations. The R−Y switching
has a 4 line repeat pattern and adding the frame rate to the Chroma frequency aligns with the repeat rate of 8
fields or 4 frames, 41⅔% the speed of NTSC. Therefore NTSC with its higher frame and dot repeat rate handles
fast motion better and allows a picture 5 times as bright before flicker is noticed.

NTSC-M variant PAL-M used in Brasil is one example. In Brasil before Color TV was introduced they were on
the U.S. 525 lines 30 frames per second 'M' B & W system using a 6MHz channel spacing. Facing the same
compatibility issues as the U.S. and being a large country with many B & W sets in service it was only logical that
they use NTSC-M or a variant thereof and thus PAL-M was spawned. The two systems are so alike that it is very
easy to convert from one format to the other. PAL-M uses a slightly lower Chroma sub­carrier frequency of
3.5756115MHz so a horizontal line will end with ¼ of a cycle where NTSC-M uses 3.57954506MHz for the
chroma and 15.734264kHz horizontal scan which is 1/227½ of the Chroma. For PAL-M the frequencies are:
15.734264kHz × 227¼ = 3.5756115MHz. The horizontal, vertical, and sound frequencies are identical and
only the Chroma sub­carrier frequency was reduced enough to create the ¼ cycle per line offset but does not
need the frame rate cycle increase. Both systems are B & W and stereo sound compatible with each other.
The next image shows what is happening on screen where NTSC uses ½ cycle offset and PAL uses ¼ or ¾ cycle offset.

Why Brasil chose PAL over NTSC in the early 1970's when solid state Chroma decoders were coming on the
scene having much greater phase accuracy, greatly reduced the Hue issues that NTSC posed during the vacuum
tube days leaves one pondering. When IC decoders arrived in the mid 1970's in Japanese sets it was rare to
require any Hue adjustment once it was set. By the late 1970's and early 1980's all U.S. brands were using IC
decoders also. In the end NTSC with its simpler dot pattern and ½ cycle/line ending Chroma 3­Line and 3-D
Comb Filters provided an almost complete Chroma/Luma separation whereas PAL Chroma with its ¼ cycle and
frame cycle offset and the spectrum smearing H/2 switching make separation a much more complicated an
incomplete process. Here is a GIF animation of PAL On Screen Vector Rotation and V Switch (CC BY­SA 4.0).
 Repairing the Brokenness of PAL
The big mistake in PAL's design is adding the frame rate to the chroma frequency to re­arrange the on screen
chroma dots into a non­objectionable pattern. Unfortunately this creates an unlocked relationship between the
horizontal and Chroma frequencies, as they are locked in NTSC. This unlocked condition makes digital coding
impossible to do efficiently. This pattern varies in relation to the number of scan lines and the 4 phase states of
PAL color. It just so happens that 625 lines cause the 2 fields within a frame to pair lines with the same chroma
dot position on screen. The next pair of lines within a frame will be shifted to the left or right 90°. The results
produced are diagonal lines moving to the left or right in a 4 frame step repeat pattern. The pattern is not fast
enough to blur the motion so it must be altered. When adding the frame rate the 1st line in the next field will be
inverted 180° from its original phase. This same inverted order on a scan line in a field is 2 lines away (4 lines
in a frame). Adding or subtracting 4 frame lines from the total number of lines in a frame will also break up
this pattern. It then becomes unnecessary to add the frame rate to the chroma frequency thus keeping the
horizontal and Chroma frequencies locked. Conventional PAL[-N] sets should be able to handle the 4 line
adjustment as this is a <1% change in the lines per frame. Here are some new specs. for 625 line PAL formats.
PAL-EU
621 Lines/Frame
310½ Lines/Field
50Hz Vertical 49.97013256 ─0.05973%
1.33ms V. Blank
15.525kHz Horizontal 15.515726
6MHz Luma
4.43626875MHz Chroma 4.43361875
285¾ Factor
17.745075MHz 4 x Chroma 17.34475
1143 Factor
2½MHz LSB
1½MHz USB
6.504975MHz Sound (419) 6.5010892

PAL-N
621 Lines/Frame
310½ Lines/Field
50Hz Vertical 49.99528946 ─0.00942%
1.33ms V. Blank
15.525kHz Horizontal 15.523537
4⅕MHz Luma
3.57463125MHz Chroma 3.58205625
230¾ Factor
14.298525MHz 4 x Chroma 14.328225
923 Factor
1¾MHz LSB
⅝MHz USB
4.50225MHz Sound (290) 4.5018258

Determining the Need for Frame Rate Increase of Chroma Frequency

Depending on the number of scan lines in combination with the 4 phase states of PAL Color the Chroma dot
arrangement may create an on screen objectionable pattern.
There must be an odd number of scan lines in a frame for a 2:1 interlace. Divide the total number of scan lines
by 4. If the quotient is even with a ¼ remainder or if the quotient is odd with a ¾ remainder then add the
frame rate to the Chroma frequency. If the quotient is odd with a ¼ remainder or if the quotient is even with a
¾ remainder then it is not necessary to modify the Chroma frequency. This is the ideal situation as there will
be only 4 unique Luma/Chroma line combinations making it easy to digitally process.
PAL-M (The Frame Rate is NOT added to the Chroma Frequency! Easy to DSP.)

B&W 'M' Standard

525 Lines/Frame
262½ Lines/Field NTSC-M & PAL-M Vert.
60Hz Vertical 59.94Hz ─0.1%─────────┐
1.43ms V. Blank NTSC-M & PAL-M Horiz. │
15.75kHz Horizontal 15,734.264±0.044Hz────┼────NTSC Frequencies
4⅕MHz Luma │ │
3.5791875MHz Chroma 3.575611494MHz │ ├───3579545.06±10Hz
227¼ Factor PAL-M Adjusted Chroma │ ├─────227½
14.31675MHz 4 x Chroma 14.30244598MHz │ ├──14318180.24±40Hz
909 Factor │ └─────910
1¾MHz LSB (2⅓) │
⅝MHz USB ( ⅚) NTSC-M & PAL-M Sound │
4.5MHz Sound (286) 4.499999504MHz───────┘

An Alternate NTSC-M & PAL-M Frequency Arrangement

Looking at the commonality of the frequencies between NTSC-M & PAL-M an alternative to keeping the
Horizontal line Vertical refresh frequencies the same and changing the Chroma frequency is to use the same
Chroma frequency and slightly alter the Horizontal and Vertical frequencies. For PAL-M it turns out that this
keeps the Horizontal and Vertical frequencies the same as they are for a Type­ M B & W signal, making TV sets
that are dual NTSC-M & PAL-M simple to manufacture. A set would automatically switch between the two
systems depending on whether the colorburst has a static phase for NTSC-M or a swinging gate for PAL-M,
controlling the R−Y V Switch phase. This would apply to VCRs as well. The color under system would have to
properly preserve the Horizontal, Vertical and Chroma frequency relationships since 1 scan stripe of the tape is
one field and this would differ slightly between the two systems. The only side effect would be if a show was in
NTSC-M the frame rate would be 0.11% slower than a PAL-M show or faster vice­versa. The RC timing circuits
for the Horizontal and Vertical sections will operate well within these tolerances whereas the Chroma crystal
oscillator would stay exactly the same.

PAL-M NTSC
525 Lines/Frame
262½ Lines/Field
60Hz Vertical 59.934Hz ─0.11%
1.43ms V. Blank
15.75kHz Horizontal 15.732692kHz
4⅕MHz Luma
3.5791875MHz Chroma 3.5791875MHz
227¼ Factor 227½
14.31675MHz 4 x Chroma 14.31675MHz
909 Factor 910
1¾MHz LSB (2⅓)
⅝MHz USB ( ⅚)
4.5045MHz Sound (286) 4.49955MHz
Was OSKM really NTSC-D/K? 8/4.43/15625/625/50/25
Early in 1960, before adopting SÉCAM 7 years later, the USSR experimented with a 625 line 50Hz system on an 8MHz
channel space for about 3 years. The Chroma system was basically NTSC with a full DSB-SC ±1½MHz BW at 4.43MHz.
“Simultaneous System with Quadrature Modulation” (Одновременная Система с Квадратурной Модуляцией). Since
the quadrature modulation did not have a vestigial sideband it was not necessary to use the dual bandwidth I & Q system
but used the European B−Y & R−Y / U & V matrixing that PAL and SÉCAM adopted later. The use of an NTSC style
Chroma (an adaptation of NTSC for the current European scan and field/frame rate at the time) and given the rules in
selecting frequencies, syncs, and bandwidths for analog TV systems, this is maybe a fair description of the specification.
The potential resolution is not bad at all. If they would have stayed with this system with the hue drifting issues becoming
a minor issue with the advent of transistor/ IC Chroma decoders, the benefits of a simpler on screen Chroma dot pattern
with the use of 3 line and digital 3D comb filters would have provided superior Luma / Chroma separation and image
enhancement that NTSC greatly benefited from. SÉCAM (Système Extrêmement Contraire à la Américaine Méthode) or
PAL (Bild Immer Schön) never achieved this level of separation. It would have had the best picture out of all the systems
in use.
General:
Aspect Ratio 4:3 = 1⅓
Total Picture Pixels (Digital) 768×576 ; 442368 Pixels
Analog Resolution (Kell Factor) 543×407 ; 221184 Pixels (Studio)
Broadcast 498×407 ; 202742 Avg. 597×407 ; 243154 Max.
Vertical:
Frames Per Second 25 Hz
Frame Period 40 ms
Total Lines Per Frame 625
Picture Lines Per Frame 576
Field Sweep 50 Hz
Field Period 20 ms
Total Lines Per Field 312½
Picture Lines Per Field 288
Lines Per Blank 24½
Blank 1.568 ms
Sync 192 µs ; 3 Lines
Horizontal:
Resolution ; Pixel Aspect Avg: 497¾ ; 1.091 Max: 597 ; 0.91 (@─9dB)
Line Sweep 15.625 kHz
Line (HP) ; Picture Period 64 µs (567) ; 53.277 µs (472)
Picture BW Pixels 532¾≈1⅔×YBW×(HP−HB) ; (497¾+35)≈6⅗% OverScan
Blank (HB) 10.723 µs (95) 49.7µs (441)
Front Porch 1.467 µs (13)
Sync 4.515 µs (40)
Back Porch 4.741 µs (42)
Luma & Chroma:
Luma (Y) Bandwidth 6 MHz ; Vestigial 1¼ MHz, Corner ¾ MHz
Chroma:
Sub-Carrier 4.4296875 MHz
H/2 Odd Harmonic 567 (283½)
U Bandwidth 1½ MHz
V Bandwidth 1½ MHz
Color Burst 2.48 µs ; 11 Cycles ; 2×{5+11+5}=42
Baseband Guard 2⅞MHz
Sound:
Carrier 6.5 MHz
H Harmonic 416

Copyright ©2017 All Rights Reserved. ­ J. S. Gilstrap (KD5TVJ) & / ||

(CC BY­SA 4.0) Creative Commons, Attribution, Share Alike.
 Color Gamuts

RED GREEN BLUE

Rec.2020 (0.708, 0.292) (0.170, 0.797) (0.131, 0.046)
1953 NTSC (0.670, 0.330) (0.210, 0.710) (0.140, 0.080)
SMPTE NTSC (0.630, 0.340) (0.310, 0.595) (0.155, 0.070)
EBU (0.640, 0.330) (0.290, 0.600) (0.150, 0.060)

6504°K White Point (0.3127, 0.3290)

 Alternative Horizontal Blank Timings and Levels, a Blend of NTSC-M, NTSC-J & PAL-B/G Specifications,
Producing Increased Composite, Colorburst & Sync Levels with 0 SetUp for 720×480 Resolution
Timings Timings are within NTSC-M specifications. IRE Levels are an average mix of the three.
Active Picture 50.286µs (360) + ─┐ Gain: Sync & ColorBurst +5%, Luma/Chroma +5.95%
Overscan Lead In 1.362µs (9¾) + ─┼─ = (455) Levels [1 IRE] ¹/₁₄₀V
Overscan Lead Out 1.397µs (10) + ─┤ Setup: 0 V [0/140]
Blank (HB) 10.511µs (75¼) + ─┘ Video: 700mV [98/140] ⁷/₁₀V
Front Porch 1.292µs (9¼) (½Chroma Sync: 300mV [42/140] ³/₁₀V
Sync 4.470µs (32) Cycles) ┌──Burst: ±150mV [±21/140] ±³/₂₀V
Breezeway 384ns (2¾) ├──▶Peak: 450mV [63/140] ⁹/₂₀V
ColorBurst (11cyc) 3.073µs {1⅜+11+4⅝}×2=(34) ├──▶Bias: 300mV [42/140] ³/₁₀V
Back Porch 4.749µs (34) └▶Trough: 150mV [21/140] ³/₂₀V
 TRANSMISSION STANDARDS
Television Stereophonic Sound Standards
Television broadcast stations may transmit stereophonic sound by employing a subcarrier on the
aural carrier. The main channel modulating signal shall be the stereophonic sum modulating signal;
the subcarrier modulation shall be the stereophonic difference encoded signal.

The subcarrier shall be the second harmonic of a pilot signal which is transmitted at a frequency
equal to the horizontal line rate. Note: if the station is engaged in stereophonic sound transmission
accompanied by monochrome picture transmission the horizontal scanning frequency shall be 15,734
Hz ±2Hz.

The subcarrier shall be double sideband amplitude modulated with suppressed carrier and shall be
capable of accepting a stereophonic difference encoded signal over a range of 50 – 15,000 Hz.

The total modulation of the aural carrier, including that caused by all subcarriers, shall comply with the
requirements of § 73.1570 of the FCC Rules and Regulations.

Television Second Audio Program Standards

Television broadcast stations may transmit a subcarrier carrying a second audio program.

The subcarrier frequency shall nominally be equal to the fifth harmonic of the horizontal line rate.

The second program encoded signal shall frequency modulate the subcarrier to a peak deviation of
±10 kHz.

The second audio program subchannel shall be capable of accepting second program encoded
signals over a range of 50 – 10,000 Hz.

The modulation of the aural carrier by the second audio program subcarrier shall comply with
§ (D) (a) (1) (iv) of this bulletin (±15kHz deviation).

Television Sound Encoding Standards (dbx, now THAT Corp.)

The stereophonic difference audio signal and the second program audio signal shall be encoded prior
to modulating their respective subcarriers. A diagram of one method of obtaining this encoding is
shown as Fig. 8–1.

This encoding shall have the following characteristics where f is represented in kilohertz (kHz).

(i) Fixed pre-emphasis F(f) whose transfer function is as follows:

┌ ┐ ┌ ┐
│ 1+j2.451f │ │ 1+jf/2.19 │
F(f) = │───────────│ × │───────────│
│ 1+jf/5.23 │ │ 1+jf/62.5 │
└ ┘ └ ┘
(ii) Wideband amplitude compression wherein:

(a) The decibel gain (or loss) applied to the audio signal during encoding is equal to minus one
times the decibel ERMS value of the encoded signal (the result of the encoding process),
weighted by a transfer function P(f) as follows:

jf/0.0354
P(f) = ─────────────────────────
[1+jf/0.0354]×[1+jf/2.09]

(b) The exponential time weighting period T₁ of the ERMS detector referred to above in (a) is
34.7ms

(c) The zero decibel reference ERMS value for the encoded signal referred to above in (a) is
8.99% modulation of the subcarrier at 300 Hz.

(iii) Spectral compression wherein:

(a) The transfer function S(f,b) applied to the audio signal during encoding is:

1+j(f/F)×(b+51)÷(b+1)
S(f,b) = ──────────────────────
1+j(f/F)×(1+51b)÷(b+1)

F=20.1kHz; D=decibel rms value and b is the decibel ERMS value of the encoded signal (the result of
the encoding process) weighted according to a frequency transfer function Q(f) as follows:

j(f/5.86)³
Q(f) = ──────────────────────────────────────────────
[1+j(f/7.66)²+jf/7.31]×[1+jf/26.9]×[1+jf/3.92]

(b) The exponential time weighting period T₂ of the ERMS detector referred to above in (iii-a) is
11.4ms.

(c) The ERMS zero decibel reference for the encoded signal referred to above in (iii-a) in 5.16%
modulation of the subcarrier at 8kHz.

Note: This reference results in a +18.4dB gain throughout the encoding process at 32.0%
modulation using an 8kHz tone, when the output bandlimiting filter (see (iv) and (v) following)
gain is +18.4dB at 8kHz.

(iv) Overmodulation protection which functionally follows the functions i, ii, & iii above.

(v) Bandlimiting to appropriately restrict bandwidth which functionally functions i, ii, & iii above.

Television non-program related aural subcarrier standards.

Multiplexing of the aural carrier is subject to the requirements of § 73.682 (c) of the FCC Rules and
Regulations: provided, however, that when the stereophonic and/or second audio program
subchannels are transmitted, multiplexing, of the aural carrier by non-program related subchannels is
subject to the following changes:
(i) The maximum modulation of the aural carrier by the non-program related subcarrier shall comply
with the requirements of § (D) (a) (1) (iv) of this bulletin.

(ii) When the stereophonic and second program subcarriers are transmitted, the instantaneous
frequency of the non-program related subcarriers shall have the average value of six and one half
times the horizontal scanning frequency with a tolerance of ±500Hz.

(iii) When only the stereophonic subcarrier is transmitted, the instantaneous frequency of the
non-program related subcarrier shall lie between 47 and 120kHz with a tolerance of ±500Hz.
 TRANSMISSION SYSTEM REQUIREMENTS
/MULTICHANNEL SOUND REQUIREMENTS
Electrical Performance Standards for Stereophonic Operation
The aural transmitter must operate satisfactorily with a frequency deviation of ±73kHz.
It is recommended that the transmitter operate satisfactorily with a frequency of ±100kHz.

The pilot subcarrier shall be frequency locked to the horizontal scanning frequency of the transmitted
video signal.
The requirements of § 73.687 (b) (2) of the FCC Rules and Regulations shall be compiled for both
(L+R) main channel and (L–R) subchannel, except for pre-emphasis as specified in § (B) (c) of this
bulletin, with the additional requirement that the aural transmitter shall be capable of transmitting a
band of frequencies from 50 to 120,000 Hz.

Unless otherwise specified, the transmission system requirements are defined for 75µs pre-emphasis
(which is matched to that in the main channel in the case of stereophonic transmission) substituted
for encoding. Measurements are made over the band of 50 to 15,000 Hz and employ 75µs
de-emphasis in the measuring equipment.

The stereophonic subcarrier, being the second harmonic of the pilot signal, shall cross the time axis
with a positive slope simultaneously with each crossing of the time axis by the pilot subcarrier. The
pilot subcarrier shall cross the time axis at points locked within ±3° (approximately ±530ns) of the
zero crossings of the stereophonic subcarrier.

The unmodulated stereophonic subcarrier shall be suppressed to a level less than 250 Hz deviation
of the main carrier.

The combined audio frequency harmonics measured at the output of the transmitting system
(including the sound encoder), as defined in § 73.687 (b) (3) of the FCC Rules and Regulations, at
any audio frequency from 50 – 15,000 Hz and at modulating percentages of 25, 50 and 100%, 75µs
equivalent modulation shall not exceed the rms values in the following table:

50 to 100 Hz 3.5%
100 to 7,500 Hz 2.5%
7,500 to 15,000 Hz 3.0%

Harmonics shall be included to 30kHz.

The ratio of peak main channel deviation to the peak stereophonic subchannel deviation when only a
steady state Left (or Right) signal exists shall nominally be one half of all levels of this signal and for
frequencies 50 – 15,000 Hz.

The phase and amplitude characteristics of the stereophonic sum modulating signal and the
stereophonic difference encoded signal shall be such that the minimum equivalent input at 10%, 75µs
equivalent modulation is as follows:

(i) 30dB separation from 100 Hz to 8 kHz.

(ii) Smoothly decreasing separation below 100 Hz, from 30 dB to 26dB at 50 Hz.
(iii) Smoothly decreasing separation above 8 kHz, from 30 dB to 20 dB at 15kHz.

Note: it is recommended that the transmission system excluding encoding, shall meet a 40 dB sep-
aration requirement when 75µs pre-emphasis in substituted for sound encoding.

Crosstalk into the main channel caused by a signal in the stereophonic subchannel shall be at least
40 dB below 24 kHz main carrier deviation.

Crosstalk into the main channel caused by a non-stereophonic multiplex signal shall be at least 60 dB
below 25 kHz aural carrier deviation.
Crosstalk into the stereophonic subchannel caused by signal in the main channel shall be at least 40
dB below 50 kHz aural carrier deviation.

Crosstalk into the stereophonic subchannel caused by another multiplex signal shall be at least 60 dB
below 50 kHz aural carrier deviation.

The aural transmitting system output frequency modulation noise level in the band of 50 – 15,000 Hz
(with de-emphasis) must be at least 58 dB below the audio level representing a frequency deviation
of ±25 kHz. The frequency modulation noise level in the stereophonic subchannel, after
demodulation in the band of 50 – 15,000 Hz ( with de-emphasis) must be at least 55 dB below the
audio level representing a frequency deviation of ±50 kHz.

The pilot subcrrier to interference ratio, over a bandwidth of 1 kHz centered at the pilot subcarrier,
shall be at least 40 dB.

Electrical Performance Standards

for Second Program Operation
The aural transmitter frequency deviation capability must comply with the requirement of
§ (C) (a) (1) of this bulletin.

The aural transmitter modulation bandwidth capability shall comply with the requirement of
§ (C) (a) (1) of this bulletin.

The unmodulated subcarrier shall be frequency locked to the fifth harmonic of the horizontal line rate.
When modulated, the center frequency shall nominally be that of the fifth harmonic of the horizontal
line scanning frequency with a tolerance of ±500 Hz.

Frequency modulation of the subcarrier shall be used.

The subcarrier shall be shut off when the second audio program subchannel is not in use.

The combined audio frequency harmonics measured at the output of the transmitting system
(including the encoder) at any audio frequency from 50 – 15,000 Hz and at modulating percentages
of 25, 50 and 100% 75µs equivalent modulation shall not exceed the rms values in the following
table.
50 to 100 Hz 3.5%
100 to 7,500 Hz 4.0%
7,500 to 15,000 Hz 3.0%

Harmonics shall be included to 20 kHz.

Cross-talk into the SAP subchannel caused by a signal in the main channel and/or in the
stereophonic shall be at least 50 dB below level representing full modulation of the SAP subcarrier
(±10 kHz deviation).

The aural transmitting system output frequency modulation noise level after subcarrier demodulation
be at least 50 dB below the level representing full modulation of the SAP subcarrier (±10 kHz
deviation).

The aural transmitting system output frequency modulation noise level in the band of 63 – 94 kHz
shall be at least 50 db below the level representing 100% amplitude modulation.
 Electrical Performance Standards for Video Operation
The requirement of § 73.687 (a) (1) and (2) of the FCC Rules and Regulations shall be complied with
provided, that when the station is engaged in stereophonic sound transmission, or when the station
transmits stereophonic sound and/or second audio program, subparagraphs (1) and (2) apply, except
except as modified by the following: A sine of 4.5 kHz introduced at the terminals of the transmitter
which are normally fed the composite color picture signal shall produce a radiated signal having the
amplitude (as measured with a diode on the RF transmission line supplying power to the antenna
after the combination of visual and aural power) which is down at least 30 dB with respect to the
signal produced by a sine wave of 200 kHz.

In the situation where stereophonic sound and/or second audio program is transmitted, the following
requirements shall be met: the incidental phase modulation of the visual carrier by video signals in the
frequency band of 1 and 92 kHz shall be less than 3° for carrier amplitude below ¾ of the voltage at
synchronizing peaks and less than 5° for carrier amplitudes exceeding ¾ of the voltage at
synchronizing peaks.

Electrical Performance Standards for Sound Encoding

The equivalent input noise of the sound encoder, measured over a 15 kHz bandwidth, shall be more
than 70 dB below the 100 Hz, 100% 75µs equivalent modulation level.

The tracking characteristics of the sound encoder shall be such that the minimum equivalent input
separation a modulation percentages from 1 to 100% 75µs equivalent modulation is 26 dB from 100
Hz to 8 kHz.

Modulation Levels
When only a monophonic audio signal is transmitted, the modulation of the aural carrier shall not
exceed 25 kHz deviation on peaks of frequent recurrence, unless some other peak modulation level
is specified.

For stations transmitting more than one audio program channel the maximum modulation levels must
be meet the following limitations”

(1) TV stations stereophonic sound signals must limit the modulation of the aural carrier by the
stereophonic sum modulating signal to 25 kHz deviation on peaks of frequent recurrence.

(2) TV stations stereophonic sound signals must limit the modulation of the aural carrier by the
sum of stereophonic sum modulating signal and stereophonic difference encoded signal to
to 50 kHz deviation on peaks of frequent recurrence.

(3) The modulation of the aural carrier by the stereophonic pilot signal shall be 5 kHz deviation
wit a tolerance of ±500 Hz.

(4) TV stations transmitting a second audio program must limit the modulation of the aural carrier
by the SAP subcarrier to 15 kHz deviation.

(5) TV stations transmitting multiplex signals on the aural carrier for non-program related
purposes must limit the modulation of the aural carrier by the arithmetic sum of all
subcarriers, other than the stereophonic and second audio program to 3 kHz deviation.
 (6) The total modulation of the aural carrier by multichannel sound shall not exceed 73 kHz.

Definitions
Decibel ERMS Value: The exponentially time-weighted root mean square (ERMS) value converted to
dB as follows:
┌ ┐
│ ERMS Value │
decibel ERMS Value = 20 log │ ────────── │
│ Reference │
└ ┘
Where Reference is the 0dB ERMS value.

Exponentially Time-Weighted Root Mean Square (ERMS) Value: The ERMS value of a waveform
is obtained from the following formula:
Above is a block flow chart of NTSC advanced encoding. After matrixing into Y, I & Q they are then
low pass filtered at 4⅕, 1½ & ⅗ MHz respectively.
Adaptive Emphasis High Frequency Compensation¹ – This circuit boosts signal levels of higher
frequencies that lack the harmonics necessary to produce sharp edges. A square wave contains the
fundamental and odd harmonics to produce sharp image edges. A filtered square wave with all
harmonics removed contains a sine wave that is only 63⅔% of peak. This will boost the sine wave
peak to the same level of the square wave. It does not increase sharpness but it does restore peak
contrast and if circuits in the receiver square it up it will return the signal close to its original form.
Vestigial Sideband on I Channel – When eliminating one sideband there is a 6dB loss in envelope
modulation for frequencies above the cutoff frequency. To compensate those frequencies above the
cutoff will need a 6dB boost to restore a flat response.
Luma & Chroma Adaptive Pre-Combing¹ – In order to reduce cross color and hanging dots during
comb mesh failure or for receivers with poor Luma & Chroma separation pre-combing will reduce
those spectral components to a tolerable level that will make them minimally visual. The choice of
using this only for areas of motion is to optimize it for larger screen receivers that also use adaptive
motion (purple
dotted line).
Combing can
reduce resolution
and for still areas
this is noticeable on
larger screens.
Using adaptive
motion provides the
best performance
for larger screens
but for smaller
screens that may or
may not use a 3-line
comb filter the
artifacts can be
noticeable in still
areas. Full
non-adaptive
combing (orange
dotted line) will
reduce artifacts for
all screen sizes but
but dose not offer
the best
performance for
larger screens.
Since this advanced processing is mostly beneficial for larger screens and of limited
benefit to existing smaller screens implementing adaptive motion seems to be the
prudent choice. For still areas a field comb of 1 frame delay in the receiver will
provide complete artifact free Luma/Chroma separation. Not using pre-combing for
still areas offers the sharpest images for larger screens.
NTSC was designed to use I & Q chroma channels under the belief that a QAM
signal could only properly carry the higher frequencies of only one of the channels so
it was chosen to assign the wider bandwidth channel to flesh tones. However this
was a mistake that produces improper colors for signals from ½ to 1½MHz falling
45° between the I & Q channels. For signals that fall on either I or Q the hue will
be correct but as hues approach the 45° mark the hue error increases to its
maximum. The reason for this is that the I channel portion will contain modulation
that the Q channel does not. With a 50/50 duty cycle the filtered Q channel output
will be an average 50% of the peak modulation. The resulting modulated hue output
will bounce between two hues on either side of the original hue, hence the earned
moniker Never The Same Color. To the right are four sets of patterns that represent
the four vectors that are 45° to the I & Q axes in a before and after arrangement.
The input, above, is fully saturated and at full brightness that alternates between its
Luma equivalent with no color. The output is just below. From top to bottom the 45°
vector order is: I & Q, –I & –Q, I & –Q, –I & Q.
For a higher bandwidth Chroma using vestigial sideband QAM modulation for both
U & V channels is the better option. The two Chroma channels are usually thought
of as being separate but in reality they are a Cartesian representation of a polar
signal, R being saturation and θ being hue. With this in mind the QAM signal should be able to carry
the higher frequencies well of both channels, ²pg29. This has probably been employed on PAL-B/G
that uses a 7MHz channel space where the Luma has been reduced to 5MHz and thus the Chroma
USB has been reduced to ⅗MHz. Take for instance a Green–Magenta color bar pattern. The
vestigial sideband Chroma signal generated has 0° phase shift and resembles a suppressed carrier
signal from a single modulator similar to the Luma signal. It is off axis to the U & V channels which
represent its Cartesian co-ordinates. Upon de-matrixing into RGB sharper transitions are produced
compared to what is seen on the NTSC test pattern. It should be safe to assume that the non
vestigial sideband portion should do a good job on chroma modulation that contains hue changes.
This dual band filtering of I & Q which produces improper colors should be abandoned in favor of the
U & V scheme. A dual I / Q bandwidth receiver will still produce hue errors on a wideband U & V
signal but the outcome may be slightly different. On sets that use ⅗MHz Chroma this is a non-issue.

Above is a block flow diagram of advanced receiver decoding. Adaptive processing switches
between a field comb for still image areas to a 3-line comb for motion which is controlled by
comparing a two frame delay signal to the current to detect motion which then drives the fader
controls. The faders are necessary to transition the wipe over several pixels to avoid sharp
transitions that would be noticeable. The Chroma output is Super¹NTSC processed to square up the
signal by using the higher Luma frequencies above the Chroma cutoff frequency. This requires
proper amplitude and phase adjustments to the high frequencies before being added to the Chroma
signals.
Advanced reading:
1. NTSC and Beyond – Yves Faroudja – IEEE Transactions on Consumer Electronics, Vol.34#1 2/88
2. The Engineer’s Guide to Decoding & Encoding – John Watkinson – Snell & Wilcox Handbook Series
3. A Handbook for the Digital Engineer – Keith Jack – Newnes Elsevier
4. Improved Television Systems: NTSC & Beyond – William F. Schreiber
5. Design of FIR Filters – Elena Punskaya
 Horizontal & Vertical Blank & Sync Timings & Structure
Regarding the horizontal blank & sync components, front porch, sync, back porch and
colorburst the dot clock optimized timings are:

Horizontal Blank: 10.9µs Horizontal Blank Structure available in

Front Porch: 1½µs Composite Video Scope Image on page 4.
Sync: 4.7µs
Back Porch: 4.7µs
Colorburst: 2⅘µs, 10 cycles

The timings on page 6 reflect these within the tolerances that the dot clock, the chroma 8×
oscillator, can produce. When generating the signal these values should be adhearded to.
For better compliance with PAL-M, and instead of centering the colorburst on the back porch,
the minimum breezeway betweem sync and burst is 381ns, the average space after
centering is ~1µs, so this space can be reduced to the mimimum allowing for greater time
for the V switch to complete its operation. Using 419ns (1½ cycles) with a 10 cycle
colorburst leaves 1½µs of time for the V switch to complete its opertaion within the blank.

However specification tolerances are a bit looser and any decoding must accommidate these
ranges.

Horizontal Blank: 10.487µs (0.165H) min

Front Porch: 1.271µs (0.020H) min
Sync: 4.449µs (0.070H) min
5.084µs (0.080H) max
Breezeway Spacing: 381ns (0.006H) min
Sync Start to Burst End: 7.94 µs (0.125H) max
Sync Start to Blank End: 9.215µs (0.145H) min
Colorburst: 2.234µs (0.035H) min ( 8cycles)
3.073µs (0.048H) max (11cycles)
Back Porch: 4.131µs (0.065H) min

For information on vestigial sideband modulation as it pertains

to NTSC-M broadcast transmission see NTSC-DVD on pages 7 & 8.
 Enhanced NTSC Timings
In the goal of squeezing every last bit of performance out of the NTSC signal space it is probably
possible to modify the specs within limitations that most older sets shouldn't have an issue with. With
this in mind here are a set of parameters that may work. It offers a ~10 ¼% increase in detail
improvement.
Chroma Chroma
Aspect Ratio 4:3 = 1⅓ ½Cyc ½Cyc 3:2 = 1½
H.Freq 15,734.264Hz
H.Period 63.555µs 455 (227½) 423 (211½)
H.Blank 7.543µs 55 52½ 7.888µs
F.Porch 0.768µs 5½ 5 0.751µs
H.Sync 3.492µs 25 23½ 3.531µs
B.Porch 3.422µs 24½ 24 3.606µs
Space 0.419µs 3 3 0.451µs
Burst 2.235µs 16 16 2.404µs
Space 0.768µs 5½ 5 0.751µs
Pixels 55.873µs 400 370½ 55.667µs
Active Pixels 54.197µs 388 (776) 360 (720) 54.090µs
Overscan 1.676µs 12 3.1% 10½ 2⅚% 1.578µs
Chroma 3.57954506MHz 3.327796836MHz
Chroma BW 1½MHz
Luma BW 4⅕MHz
Ratio 2⅘:1:1
Luma Lines A:379/455 D:536/644 Avg/Max
Pxl.Asp. 1.193:1 1.345:1

IRE=1V/140
Luma (Y) Level: 98 700mV
Sync: ─42 300mV
ColorBurst: ±21 ±150mV
Max (Yl & Cy) 130⅔ 1.23V
Min (Rd & Bl) ─32⅔ 66⅔mV

1931 CIE
Rec.709 sRGB Gamut x y nm
Red 0.64 0.33 ~607
Green 0.30 0.60 ~556
Blue 0.15 0.06 ~467
White Point 0.3127 0.329 6504°K
Contrast 2¹²:1 Gamma 2.4

While the 720×480 DVD aspect is 3:2 at 4:3 the analog pixel aspect is already at 1.193:1 and stretching it
to 3:2 would increase this to an acceptable 1.342:1. With the goal of keeping the vertical edges sharp
using typical sharpness circuits 4:3 is probably best. Increasing it even more to 854×480 16:9 (1.592:1
pixel aspect) would make those edges fuzzy. Advanced circuitry can square up these edges and increase
contrast without overshoot but actual detail can never go above 455 lines because of luminance
bandwidth limitations, 227½ cycles of a 4⅕MHz sine wave within 54.197µs. This would set the pixel
aspect for 3:2 at 1.119:1 which is very good and 1.327:1 for 16:9 which is acceptable. This advanced
circuitry is great for 4:3 providing a pixel aspect 0.99:1, which is perfect and almost studio quality (500).
Using this circuitry this is good enough to make the default resolution 720×480 but in order to have 180
chroma cycles that would align to the 720 samples within ~54µs the chroma would need to be lowered to
211½×15,734.264kHz=3.327796836MHz and this would also increase the USB chroma to ⅞MHz.