EE 351 · Communication Systems

Week 7

Television Engineering — From Images to RF Signals

Raster scanning, composite video, video bandwidth, chrominance multiplexing, and vestigial-sideband AM.

By the end of this lesson you should be able to: explain how a picture becomes a time-varying voltage; describe scanning, blanking, synchronization and interlacing; calculate line time, active time, field and frame rates; relate picture detail to video bandwidth; explain how NTSC multiplexes colour into the same signal; justify vestigial-sideband AM; and read the spectrum of a historical 6-MHz analog TV channel.

Weeks 3–6 built AM, SSB, FM and stereo multiplexing. Television reuses every one of those ideas, in one system.

A — Why television is a communication problem

Radio transmits audio. How do we transmit a moving picture?

Audio

One number per instant: pressure \(p(t)\). A single wire, a single carrier, about 15 kHz.

A still picture

A number at every point: brightness \(I(x,y)\). Two spatial coordinates, no natural “time” axis.

A moving picture

\(I(x,y,t)\): three coordinates, yet one antenna and one radio channel must carry it.

That requirement forces four engineering tools: scanning (serialize), synchronization (re-assemble), modulation (reach the channel), and frequency multiplexing (add colour and sound).

A — Why television is a communication problem

The whole lesson in one chain

Image→Scan lines→Brightness waveform Y′→+ sync, blanking→+ chrominance (3.58 MHz)→Composite video, B ≈ 4.2 MHz→VSB-AM on a picture carrier→+ FM sound→6-MHz RF channel
PartQuestion answeredTool you already know
B–C Scanning & interlacingHow is a picture serialized?Sampling a signal in time
D Video bandwidthHow much spectrum does detail cost?Bandwidth of a square wave
E ColourHow can colour ride in the same signal?Quadrature DSB-SC (Week 4), FDM (Week 5)
F–G VSB and the channelHow is it sent in 6 MHz?DSB / SSB (Weeks 3–4), FM sound (Week 6)

A — Why television is a communication problem

Television is a review of the course so far

Earlier weekIdeaWhere television uses it
Week 3 — DSBProduct with a carrier creates two sidebands; \(B_{\text{RF}} = 2B\)The 8.4-MHz problem for 4.2-MHz video; chrominance is DSB-SC on a subcarrier
Week 4 — SSBKeep one sideband; quadrature branchesVSB is the compromise between DSB and SSB; I/Q quadrature carries colour
Week 5 — FM stereoFrequency-division multiplexing of audio and a subcarrierColour multiplexed on a subcarrier inside the video signal; sound on its own FM carrier
Week 6 — FMDeviation, modulation index, Carson bandwidthTV sound: FM with ±25 kHz peak deviation

Part B

Raster Scanning and Video Signal Generation

Image → scan line → brightness waveform → sequence of lines → reconstructed image.

B — Raster scanning

A picture is 2-D; a signal is 1-D

A camera sensor (a vidicon tube historically, a CCD today) examines the picture one narrow strip at a time. Along each strip it outputs a voltage proportional to brightness.

  • Horizontal position \(x\) becomes time within a line.
  • Vertical position \(y\) becomes which line — i.e. time within a field.
\[ v(t) \;\longleftarrow\; I(x,y) \quad\text{with}\quad x \propto t - t_{\text{line start}},\;\; y \propto \text{line number} \]

B — Raster scanning

Left to right, top to bottom

trace (beam on)left → rightretrace (beam off)fast flybacknext line starts alittle lowertop → bottom
The scanning beam traces each line at uniform speed (blue). Between lines it flies back to the left, much faster, with the beam turned off (red dashed). Because the vertical deflection keeps moving, each line is drawn slightly lower than the last. The standard picture shape is 4:3 (width : height).

B — Raster scanning

Worked example: a black letter F on white

1234567picture, lines numbered from the topline 1line 2line 3line 4line 5line 6line 7white (high)black (low)time along each line (left → right)
Each scan line produces its own waveform: high (white) where the beam sees the background, low (black) where it crosses the letter. Lines 1 uses no black at all; lines 2–3 cross the top bar; line 4 only the stem; lines 5–6 the middle bar; line 7 the stem again. (Idea after textbook Figs 23-3 and 23-4; redrawn.)

B — Raster scanning

Sent one after another: the video signal

line 1line 2line 3line 4line 5line 6line 7blanksyncwhiteone line = trace + blanking, repeated every 63.56 µs
The camera output for lines 1–7 joined in time. Between lines there is a short blanked interval containing a sync dip. This one-dimensional waveform is the brightness (luminance) signal, written \(Y\) in the textbook and \(Y'\) once gamma correction is acknowledged.

At the receiver the process is reversed: the same waveform drives the brightness of a beam that sweeps the screen in step with the camera. Image → line → waveform → lines → image.

B — Raster scanning

Demo 1 — Raster scan explorer

Original picture

Redrawn by the receiver

Video voltage of the selected line

This model draws 48 lines so you can see each one; NTSC transmits 525 (≈ 485 carry picture). Timing shown is the real NTSC line: 63.56 µs, of which about 10.9 µs is blanking (BP = back porch, FP = front porch).

Exercise 1 of 7 — Raster scanning

Draw the waveform

Exercise 1

The 8 × 8 pattern below is scanned top to bottom, each row left to right. Black cells are shaded. Sketch the video waveform for row 6 (white = high, black = low).

Which description matches row 6 (active video only)?

B — Raster scanning

Anatomy of one horizontal line

white 100blank 0sync tip −40black 7.5TH = 1 / 15 734 Hz = 63.56 µssyncactive video ≈ 52.7 µs (visible scan)back porchFPhorizontal blanking ≈ 10.9 µs = front porch 1.5 + sync 4.7 + back porch 4.7 µs (beam off)
One NTSC line is \(T_H = 1/f_H = 1/15\,734\text{ Hz} = 63.56\ \mu\text{s}\). About 10.9 µs is horizontal blanking (front porch 1.5 µs, sync 4.7 µs, back porch 4.7 µs); about 52.7 µs shows picture. The textbook rounds these to ≈ 10 µs and ≈ 53.5 µs. Levels are in IRE units; blanking is 0, black 7.5, white 100, sync tip −40.

B — Raster scanning

Four ideas that are easy to confuse

TermWhat happensWhy it is needed
Visible scanning (trace)Beam moves left → right at constant speed; its brightness follows the video voltageThis is the picture
Horizontal retraceBeam flies back to the left edge (physical motion)Start the next line
Horizontal blankingSignal sits below black so the beam is off (electrical interval, contains the retrace)Hide the retrace; leave time for the sweep circuit to reset
SynchronizationA pulse (sync tip, 4.7 µs) inside the blanking interval marks each line startKeeps the receiver sweep locked to the camera; no drift, no skew

B — Raster scanning

Why the receiver needs sync pulses

  • The receiver has its own sweep oscillators. Left alone, analog circuits drift, so lines would start at slightly different positions and the picture would skew sideways.
  • The transmitter therefore adds a sync pulse to every line (and a longer pattern at the end of every field). The receiver’s sync separator finds those pulses because they go below the black level, a range the picture never uses.
  • Sync is also what tells the receiver which line is the first line of a field.

Exercise 2 of 7 — Timing

Line period and active video

An NTSC colour system has a line frequency of 15 734 Hz and about 10.9 µs of horizontal blanking per line. What are the line period and the active (visible) time per line?

Part C

Interlaced Scanning

Fields, frames, half lines, and the NTSC timing numbers.

C — Interlaced scanning

Two ways to draw the same lines

ProgressiveInterlacedlines 1, 2, 3, 4 … in orderField 1: odd lines (solid) · Field 2: even lines (dashed)one pass = one full frametwo passes = one full frame
Progressive scanning draws lines 1, 2, 3 … in order: one pass is one complete frame. Interlaced scanning draws the odd lines first (field 1), then the even lines in the gaps (field 2): two passes make one frame. Broadcast analog television was interlaced.

C — Interlaced scanning

The flicker–bandwidth compromise

Flicker needs a fast refresh

Eyes see a screen flash as steady only if it is refreshed roughly 50–60 times per second.

Detail needs many lines

Sending 60 complete frames every second would double the line rate and so the video bandwidth.

The price: fast-moving edges are in different places in the two fields. They look jagged (“combing”) when the two fields are shown together, and thin horizontal detail can flicker (“line twitter”).

C — Interlaced scanning

The NTSC timing numbers

QuantityMonochrome (1941)Colour NTSC
Lines per frame525 (two fields of 262½)
Field rate60 Hz59.94 Hz
Frame rate30 Hz29.97 Hz
Line frequency \(f_H\)15 750 Hz15 734 Hz (15 734.27)
Line period \(T_H\)63.5 µs63.56 µs
\[ f_H=\tfrac{4.5\ \text{MHz}}{286}\approx 15\,734.27\ \text{Hz},\qquad f_{\text{field}}=\tfrac{2f_H}{525}\approx 59.94\ \text{Hz},\qquad f_{\text{frame}}=\tfrac{f_H}{525}\approx 29.97\ \text{Hz} \]

C — Interlaced scanning

Why 262½ lines per field?

start of field 1ends mid-linevertical retrace ↗Field 1: 5½ lines in this sketchstarts mid-line (hidden)ends at the end of a lineField 2: also 5½ linesThe half line gives both fields equal length: one master oscillator and dividers can drive H and V timing.
An odd line count (525) means each field contains a half line. Field 1 ends mid-way across the bottom; field 2 begins mid-way across the top. Both fields then have the same duration and the same retrace distance, so one master oscillator and simple dividers produce all the timing. Half lines fall in the part of the picture hidden by overscan.

C — Interlaced scanning

Vertical blanking and vertical sync

vertical blanking≈ 20 lines≈ 242 lines of picture information per fieldFirst 9 lines of the blanking: 3 pre-equalising + 3 broad (vertical sync) + 3 post-equalising pulse lines.One field = 262½ line periods = 16.68 ms (59.94 Hz). A frame (two fields) shows at most ≈ 485 lines; about 480 are used.
After each field the beam must fly back from bottom to top. About 20 line periods are blanked for this; that is why only ≈ 485 of the 525 lines can carry picture, and why textbook estimates quote 480–495. Inside that interval the signal carries a recognisable pulse pattern so the receiver can tell a field boundary from an ordinary line sync.

The pulses use a double-rate pattern (equalising pulses) so that field 1 and field 2, which begin half a line apart, look the same to the receiver’s sync circuits.

C — Interlaced scanning

Demo 2 — Interlacing explorer

Progressive (same line rate)

Interlaced: two fields per frame

Line-rate and field sync (stylised)

Local brightness: amber progressive, blue interlaced

Reference: NTSC has 525 lines (262½ per field), 15.734 kHz, 59.94 fields/s, 29.97 frames/s. Here 21 lines and a slowed clock keep the structure visible. Brightness fades after the beam passes, like a phosphor.

C — Interlaced scanning

625-line / 50-field systems

525 / 60 (NTSC System M)625 / 50 (e.g. PAL B/G)
Lines per frame525625
Fields / frames per second59.94 / 29.9750 / 25
Line frequency15 734 Hz15 625 Hz
Line period63.56 µs64 µs
Typical lines carrying picture≈ 480–485≈ 575–580
Colour subcarrier3.579545 MHz4.43 MHz (PAL)
Video bandwidth (nominal)4.2 MHz5.0 MHz (B/G)
RF channel width6 MHz7 MHz VHF / 8 MHz UHF (B/G)

Exercise 3 of 7 — Interlacing

Fields, frames and lines

A 625-line, 25-frame/s interlaced system sends two fields per frame. What are its field rate, lines per field and line frequency?

Part D

Video Signal Bandwidth

From picture detail to hertz.

D — Video bandwidth

Fine detail is a fast signal

picturewhiteblackone cycle = one black + one white element = 1 / fN = 8 elements → 4 cycles during Tactivetime across the active line →
Scan across alternating black and white vertical stripes: the video voltage is a square wave. One black stripe plus one white stripe is one cycle. Thinner stripes mean more cycles in the same active time, so a higher frequency.

D — Video bandwidth

From stripes to a frequency

  1. Let the active scan time be \(T_{\text{active}}\), and let \(N\) be the number of alternating black and white elements along one line.
  2. One cycle = one black + one white element, so the line holds \(N/2\) cycles.
  3. The period of one cycle is \(T_0 = T_{\text{active}} / (N/2) = 2T_{\text{active}}/N\).
  4. The fundamental frequency of the square wave is its reciprocal:
\[ f_{\text{video}}=\frac{1}{T_0}=\frac{N}{2\,T_{\text{active}}} \]

Doubling the detail (\(N\)) doubles the frequency. Halving the active time (scanning twice as fast) also doubles it. Higher picture detail or faster scanning demands more bandwidth.

D — Video bandwidth

Check the units; state the assumptions

\[ [f]=\frac{\text{elements}}{2\times\text{seconds}} = \frac{1}{\text{s}}=\text{Hz}\qquad\Bigl(\text{with }T_{\text{active}}\text{ in }\mu\text{s: } \tfrac{\text{elements}}{\mu\text{s}}=\text{MHz}\Bigr) \]
What the formula assumes
  • The worst-case pattern: alternating black/white elements.
  • Ideal square-wave picture; only the fundamental counted as the needed bandwidth.
  • \(T_{\text{active}}\) excludes blanking.
What it ignores
  • Sharper edges need harmonics (higher bandwidth).
  • Real filters roll off gradually.
  • Camera and display limits, and the eye’s own limits.

So the result is an estimate of the highest useful video frequency, good to a few per cent, not a precise standard.

D — Video bandwidth

Demo 3 — Video bandwidth explorer

Original stripes

After bandwidth limiting

Frequencies on a line (MHz)

D — Video bandwidth

The NTSC 4.2-MHz case, worked backwards

  1. Cycle period at the limit: \(1/4.2\text{ MHz} = 0.238\ \mu\text{s}\). One element (half a cycle): \(0.119\ \mu\text{s}\).
  2. Textbook active time: about \(53.5\ \mu\text{s}\) (a 63.6-µs line less roughly 10 µs of horizontal blanking).
  3. Elements along the line: \(N = 53.5/0.119 \approx 449.5\), so \(R_H \approx 450\).
\[ N = 2\,T_{\text{active}}\,f_{\text{video}} = 2(53.5\ \mu\text{s})(4.2\ \text{MHz}) \approx 449 \]

D — Video bandwidth

“450 lines” across the width is not “450 lines” per height

Horizontal, across the full width

\(R_H \approx 450\) alternating elements fit across the whole 4:3-wide picture.

Per picture height (TVL)

Because the picture is only 3/4 as high as it is wide, the same detail corresponds to \(450 \times \tfrac34 \approx 337\) TV lines per picture height.

\[ R_V \approx 0.7\,N_L \quad\Rightarrow\quad 0.7\times 485 \approx 340\ \text{lines (vertical resolution)} \]

Only some of the visible lines are independently resolvable, since lines and picture details are not perfectly aligned. Notice that the ≈ 337 TVL horizontal figure and the ≈ 340-line vertical figure are close: NTSC was designed so that horizontal and vertical detail are about balanced.

Always say which one you mean: across the full width, or per picture height.

D — Video bandwidth

A PAL-based example (textbook Example 23-2)

Given: 625 lines, 25 frames/s, \(f_H = 15\,625\) Hz, 80 % of each line active, \(R_H \approx 512\), about 580 displayed lines. Find the bandwidth and vertical resolution.

  1. \(T_H = 1/15\,625 = 64\ \mu\text{s}\); \(T_{\text{active}} = 0.8 \times 64 = 51.2\ \mu\text{s}\).
  2. One element: \(51.2/512 = 0.1\ \mu\text{s}\); one cycle: \(0.2\ \mu\text{s}\).
  3. \(B \approx 1/(0.2\ \mu\text{s}) = 5\ \text{MHz}\) — the same as \(f=N/(2T)=512/(2\times51.2\ \mu\text{s})\).
  4. \(R_V = 0.8 N_L = 0.8 \times 580 \approx 464\) lines.

Exercise 4 of 7 — Video bandwidth

Frequency of a stripe pattern

A scan line crosses 240 alternating black and white stripes during an active time of 52.7 µs. What fundamental video frequency results?

Exercise 5 of 7 — Resolution

Bandwidth for a target resolution

A 625-line system must resolve 600 elements across the full picture width, with 51.2 µs of active line time. What video bandwidth is needed (idealised)?

Part E

Adding Colour Without Tripling the Bandwidth

RGB, luminance, chrominance, and quadrature multiplexing.

E — Adding colour

The obvious way: three pictures

Any colour can be made by mixing red, green and blue light, so a colour camera is really three cameras behind colour filters, producing \(R'\), \(G'\), \(B'\) signals (the prime marks that each has been gamma-corrected for the display).

R′+G′+B′=3 × 4.2 MHz ≈ 12.6 MHz?
  • Sending three full-bandwidth signals would need three times the spectrum.
  • It would also be incompatible: millions of black-and-white receivers already existed in the 1950s.
  • The eye resolves fine brightness detail far better than fine colour detail, so colour can be sent with much less bandwidth than brightness.

E — Adding colour

Luminance: the brightness signal

\[ Y' \approx 0.299\,R' + 0.587\,G' + 0.114\,B' \]
  • The weights reflect how bright the eye finds each primary: green most, blue least. The textbook writes them as 30 %, 59 % and 11 %.
  • Notation: the textbook calls this \(Y\). Strictly it is \(Y'\) (“luma”), computed from gamma-corrected \(R'G'B'\), not true photometric luminance. This lesson writes \(Y'\) for the signal and \(Y\) only where quoting the textbook.
  • White (\(R'=G'=B'=1\)) gives \(Y'=1\); any grey gives \(Y'\) equal to that grey level; pure blue gives only \(0.114\).

E — Adding colour

Why a black-and-white set still works

Monochrome receiver

Uses only \(Y'\) and the sync. The chrominance is a small, fine-grained pattern near 3.58 MHz that it largely ignores or that appears as a faint dot texture.

Colour receiver

Recovers \(Y'\), then separately recovers the colour information from the 3.58-MHz subcarrier, and recombines them into \(R'G'B'\).

\[ \text{composite video} \;=\; Y'(t) + C(t),\qquad C(t)\ \text{centred on}\ 3.579545\ \text{MHz} \]

Compatibility works in both directions: a colour broadcast shows a normal grey picture on an old set, and a monochrome broadcast has no colour, so the colour set shows a normal grey picture.

E — Adding colour

Colour-difference signals: I and Q

Brightness is already sent as \(Y'\), so only two more numbers per point are needed. Take colour differences, which are zero for any grey:

\[ R'-Y',\qquad B'-Y' \]

NTSC rotates this pair by 33° into I (in-phase) and Q (quadrature), chosen to match how sharply the eye sees orange–cyan versus green–magenta detail:

\[ I = 0.596R'-0.274G'-0.322B',\qquad Q = 0.211R'-0.523G'+0.312B' \]

The textbook rounds these to I = 60 % R − 28 % G − 32 % B and Q = 21 % R − 52 % G + 31 % B. Because I carries the detail the eye resolves best it is allowed about 1.5 MHz; Q is limited to about 0.5 MHz. (Many later receivers demodulate with equal narrow bandwidths; the I/Q split is the historical design.)

E — Adding colour

Two colour signals on one subcarrier

R′G′B′from cameraY′ matrix0.299 0.587 0.114I matrix+0.60 −0.28 −0.32Q matrix+0.21 −0.52 +0.31LPF≈ 1.5 MHzLPF≈ 0.5 MHz××3.579545 MHzsubcarrier oscillator90° shift+Y′ + Ccomposite videoIQEach × is a balanced modulator: DSB suppressed-carrier AM.
I and Q each pass through a low-pass filter, then a balanced modulator. The two modulators use the same 3.579545-MHz subcarrier but 90° apart. Adding them gives the chrominance \(C(t)=I\cos(\omega_{sc}t+33^\circ)+Q\sin(\omega_{sc}t+33^\circ)\), which is then added to \(Y'\) (after textbook Fig. 23-9a, redrawn).

E — Adding colour

Why two signals can share one frequency

This is the same idea as Week 4’s quadrature branches. A cosine and a sine at the same frequency are orthogonal, so two independent signals can be sent and separated:

\[ C(t)=U(t)\sin\omega_{sc}t+V(t)\cos\omega_{sc}t,\qquad U\propto B'-Y',\ \ V\propto R'-Y' \]

A synchronous detector multiplies by \(2\sin(\omega_{sc}t)\) or \(2\cos(\omega_{sc}t)\) and low-pass filters; the cross terms average to zero:

\[ \operatorname{LPF}\{2C\sin\omega_{sc}t\}=U,\qquad \operatorname{LPF}\{2C\cos\omega_{sc}t\}=V \]

Each modulator is a DSB suppressed-carrier AM stage: no subcarrier is transmitted, only sidebands. That is why the receiver must regenerate the subcarrier, with the right phase.

E — Adding colour

Hue is phase; saturation is amplitude

+(B−Y)+(R−Y)I axisQ axisburst, 180°redyellowgreencyanbluemagentaPhasor length = saturationPhasor angle = hueZero length: grey (no chroma).The burst is the phase reference:every hue angle is measured against it.Drawn with +(B−Y) at 0° and +(R−Y) at 90°.I and Q are 90° apart; I sits 57° from the burst axis.Colours shown at full-saturation bar positions;lengths are approximate.
Write the chrominance as a single sinusoid: \(C(t)=M\cos(\omega_{sc}t+\varphi)\). Its amplitude \(M=\sqrt{U^2+V^2}\) is the colour’s saturation (how far from grey); its phase \(\varphi\), measured against the burst, is the hue (which colour). A grey has zero chroma amplitude. (Textbook Fig. 23-9b shows the same phasors.)

E — Adding colour

Frequency multiplexing: luminance and chrominance interleaved

01234MHzluminance Y′ (spectral lines at multiples of fH)3.579545 MHz (suppressed)I, Qzoom: spectral lines near 3.58 MHz — the two sets interleaveY′ lines: multiples of fH = 15.734 kHzchroma lines: odd multiples of fH / 2
A scanned picture’s spectrum is not smooth: it clusters at multiples of the line frequency \(f_H\). The subcarrier is placed at an odd multiple of \(f_H/2\) (\(227.5\,f_H = 3.579545\) MHz), so the chrominance lines fall between the luminance lines. This interleaving lets both share the upper part of the 4.2-MHz video band with limited mutual interference.

E — Adding colour

The colour burst: a phase reference

sync 4.7 µsburst ≈ 2.5 µsback porchvideo9 cycles of 3.579545 MHz
Because the subcarrier is suppressed, the receiver’s regenerated oscillator has no phase information. A burst of about 8–12 cycles (nominally 9) of the unmodulated subcarrier is placed on the back porch of every line. A phase-locked oscillator compares to it and locks, so demodulation uses the same phase reference the transmitter did.

Phase error rotates every hue by the same angle: a 20° error shifts all colours around the colour wheel, which is what the old “tint” control adjusted.

E — Adding colour

Demo 4 — Chrominance multiplexing explorer

1 · Original R′G′B′

2 · Y′ only: what a B&W set shows

3 · Chroma only (on mid-grey)

4 · Reconstructed colour

Composite waveform: burst, then chrominance on Y′

Phasor: I/Q axes, burst reference

Baseband multiplexing: the 3.58-MHz subcarrier lives inside the video; the whole composite then amplitude-modulates the RF picture carrier (Part G). Gains not to scale. Move the phase slider: every hue rotates, so the receiver must lock to the burst.

Exercise 6 of 7 — Chrominance

Luminance from RGB; why quadrature?

A pixel has \((R',G',B') = (0.8,\,0.4,\,0.2)\). What is \(Y'\), and why are the colour signals sent in quadrature on one subcarrier?

Part F

Vestigial-Sideband Modulation

Getting 4.2 MHz of video onto an RF carrier.

F — VSB

Why not ordinary AM?

Week 3: amplitude-modulating a carrier with a message of maximum frequency \(B\) creates an upper and a lower sideband, each \(B\) wide:

\[ B_{\text{RF,DSB}} = 2B = 2(4.2\ \text{MHz}) = 8.4\ \text{MHz} \]
  • That is for the picture sidebands alone. Adding the FM sound carrier 4.5 MHz above the picture carrier makes the total even wider.
  • The whole TV channel is 6 MHz. Ordinary DSB-AM does not fit, even before any guard band.
  • The lower sideband duplicates the upper one. It is the redundant half.

F — VSB

Three ways to send the same video

carrierfcDSB-AM≈ 2B = 8.4 MHzcarrierfcSSB-AM≈ B = 4.2 MHz (sharp cut at fc)carrierfcVSB-AM≈ B + vestige ≈ 5 MHz
Idealised spectra for \(B=4.2\) MHz: both sidebands (DSB); upper only (SSB); full upper plus a controlled remnant, the vestige, of the lower (VSB; about 0.75 MHz drawn).
DSB-AMSSB-AMVSB-AM
RF bandwidth\(2B\) = 8.4 MHz\(B\) = 4.2 MHz\(\approx B+\)vestige ≈ 5 MHz
Filter at \(f_c\)none needednearly ideal “brick wall”gentle, controlled slope
Low-frequency contentpreservedhard to preservepreserved

F — VSB

Why SSB is a poor fit for television

  • Video has important energy at and near 0 Hz: the average brightness of a whole picture is a DC component, and slowly changing areas of the picture are very low frequencies. In RF terms that is right next to the carrier.
  • SSB would need a filter that passes everything above \(f_c\) and kills everything below it, with a vanishingly narrow transition. That cannot be built with practical filters at VHF/UHF.
  • Any gradual cut near the carrier damages the very frequencies carrying picture brightness and large-area tone, producing smeared, uneven pictures.

F — VSB

What makes a vestigial filter “faithful”?

Let the transmit signal be the DSB spectrum shaped by a filter \(H(f)\). Coherent detection (multiply by the carrier, low-pass filter) returns

\[ M_{\text{out}}(f) \;\propto\; M(f)\,\bigl[\,H(f_c+f)+H(f_c-f)\,\bigr] \]

The picture is recovered without distortion if the bracket is constant for all \(|f|\le B\):

\[ H(f_c+f)+H(f_c-f)=\text{const.}\qquad (|f|\le B) \]
  • Above the vestige width \(f_v\): \(H(f_c+f)=1\) and \(H(f_c-f)=0\), sum = 1.
  • Inside \(\pm f_v\): the slope must have odd symmetry about the carrier, passing through 50 %: what the lower side lacks the upper side contributes, and vice versa.
  • Vestige abruptly kept (no slope): the sum is 2 near DC, so low frequencies are doubled.

F — VSB

Demo 5a — Sideband spectrum explorer

Transmitted RF spectrum around the carrier

Recovered baseband gain after detection

F — VSB

Where the filtering really happens: the receiver

  • The transmitter removes most of the lower sideband; the receiver’s IF filter (a surface-acoustic-wave, SAW, filter in sets of the textbook’s era) supplies the final precise Nyquist slope.
  • The slope is arranged so the picture carrier sits at the 50 % (−6 dB) point of the response, giving \(H(f_c+f)+H(f_c-f)=\text{const.}\) automatically.
  • The same filter strongly attenuates the sound IF, so the sound does not leak into the video detector.
  • US standard IFs: picture 45.75 MHz, sound 41.25 MHz. Example: channel 4, picture 67.25 MHz, sound 71.75 MHz, local oscillator 113 MHz, so \(113-67.25 = 45.75\) MHz and \(113-71.75 = 41.25\) MHz. Because the oscillator is above the signal, the spectrum is mirrored at the IF.

An envelope detector works because the picture carrier is large; later receivers use synchronous detection, which reduces the distortion that VSB envelope detection can introduce.

Part G

The Complete Analog Television RF Channel

NTSC System M: picture, colour, sound and a 6-MHz allocation.

G — The RF channel

One channel, three signals

L+0L+1L+2L+3L+4L+5L+6L = lower channel edgepicture carrierL + 1.25 MHz (AM, VSB)sound carrierL + 5.75 MHz (FM)colour subcarriervestigefull upper sideband, video to P + 4.2 MHz4.5 MHz picture-to-sound spacing6 MHz channel
Picture carrier 1.25 MHz above the lower channel edge (AM, vestigial sideband); FM sound carrier 4.5 MHz above the picture carrier (5.75 MHz above the lower edge); colour subcarrier 3.579545 MHz above the picture carrier, inside the picture sidebands. Redrawn after textbook Figs 23-1 and 23-10.

G — The RF channel

Demo 5b — Historical 6-MHz channel explorer

Idealised sketch of the historical System M channel. The real transmitted spectrum is bounded by a regulatory emission mask, and the exact vestige roll-off is set by the transmitter and, above all, by the receiver’s IF filter; neither is drawn to scale here. Channel numbers 2–69 follow the textbook’s US chart.

G — The RF channel

Two different “bandwidths”

4.2 MHz6 MHz
What is it?Highest frequency in the baseband video signalWidth of the allocated RF channel
DomainAround 0 Hz (before modulation)Around the picture carrier (VHF/UHF)
Fixed byScanning (detail, line time) and the filter on the videoRegulation (spectrum allocation)
ContentsY′ and the chrominanceVSB picture + FM sound + guard regions

Budget from the lower channel edge \(L\): 1.25 MHz up to the picture carrier (this region holds the vestige) + 4.2 MHz of upper video sideband + 0.3 MHz to the sound carrier + 0.25 MHz above the sound carrier = 6 MHz.

Exercise 7 of 7 — RF channel

Carriers for channel 13, and VSB versus DSB

US channel 13 occupies 210–216 MHz. Find the picture and sound carriers.

G — The RF channel

The complete chain, simplified

TRANSMITTERCamerascan → R′G′B′MatrixY′, I, QQuadrature mod.I, Q on 3.58 MHzAdd sync + burstcomposite videoVSB-AMpicture carrierDiplexeradds FM soundFM soundcarrier +4.5 MHzRECEIVERTuner→ IFIF / SAW filterNyquist slopeVideo detectorcomposite videoY′ / chromaseparateDemodulateI, Q (burst lock)Matrix → displayR′G′B′Sync separatorH, V timingSweepscanning4.5 MHz soundFM detector
Transmitter: scan → matrix → quadrature colour modulators → add sync and burst → VSB-AM on the picture carrier → combine with FM sound. Receiver: tuner → IF with Nyquist slope → video detector → separate Y′ and chroma → demodulate with burst-locked subcarrier → matrix → display, while sync is separated to drive the sweep and the 4.5-MHz sound difference is FM-detected. (Simplified from textbook Figs 23-12 and 23-13.)

G — The RF channel

History, not present-day broadcasting

  • Everything above describes analog NTSC System M. Full-power analog broadcasting in the United States ended in 2009, replaced by the digital ATSC system, which also fits in 6-MHz channels.
  • Digital broadcasts compress the picture and send bits, not a scanned voltage. ATSC 1.0 uses an 8-level vestigial-sideband modulation (8-VSB), so VSB survived even though analog video did not.
  • Interlacing, however, did not disappear: formats such as 1080i still scan two fields per frame.
  • The textbook chapter you used is historical for its analog TV sections: use it for the engineering principles, not as a description of what a modern receiver does.

Optional engineering activity

Try it in MATLAB

The five demos need only a browser. This optional activity repeats the chain numerically:

  1. Turn rows of a test image into a sampled brightness waveform (with blanking and sync).
  2. Plot that waveform and its FFT, and find the stripe frequency \(N/(2T)\).
  3. Amplitude-modulate it onto a carrier and compare DSB with VSB filter responses.

Plain MATLAB; no toolbox or Simulink model is required, and none is provided this week. See the study article for the exact verification statement.

EE 351 · Week 7

From a picture to a 6-MHz RF channel

  • Scanning turns \(I(x,y,t)\) into one voltage; sync keeps the receiver in step.
  • NTSC: 525 lines, 29.97 frames/s (two fields, 59.94 Hz); \(T_H=63.56\) µs.
  • Detail costs bandwidth: \(f=N/(2T_{\text{active}})\); about 4.2 MHz for NTSC.
  • Colour: \(Y'\) plus quadrature I/Q on 3.579545 MHz, locked by the burst.
  • VSB: one sideband plus a vestige, with \(H(f_c+f)+H(f_c-f)\) constant.
  • Channel: picture \(L+1.25\) MHz, FM sound 4.5 MHz above it, all in 6 MHz.