Chapter 1
Basic video and magnetic theory
Introduction
Timecode was originally developed in order to clearly identify positional information on videotape in a manner similar to traditional motion-picture film. On film, this information is printed in human-readable form along the film edge, and the individual frames are clearly seen by inspecting the film. Videotape frames are recorded as magnetic imprints that cannot be seen by inspection, so when the timecode was introduced, it needed to contain details of the frames. The introduction of colour to the original monochrome signal increased the complexity of the video signal, so care had to be taken in joining non-contiguous sections during editing if disturbances in the picture were to be avoided. Developments in postproduction meant that the sound could be dealt with separately, as long as a guide video was provided, together with timecode. There have recently been great strides forward in the marriage between film and videotape for the purposes of post-production. This chapter explains the video and magnetic theory relevant to the recording and replay of timecode on video- and audiotape in order that the timecode processes described later may be easily understood.
The video signal
A scene viewed by the television camera is converted into an electronic signal by scanning the image formed by the lens in a series of horizontal lines, much as one would read text in a book. These lines are grouped together in frames (pages in the book in our analogy). The rate at which individual pictures are reproduced has to be high enough to minimize flicker. The technology available when television was developed did not permit a frame rate sufficiently high to achieve this, so each frame was split into two fields, each field containing alternative lines (Figure 1.1).
Originally the frame rate was linked to the nominal frequency of the mains. This meant that in Europe a rate of 25 frames (50 fields) per second (fps) was chosen, and in the USA the rate was 30 frames (60 fields) per second. When colour was added to the original monochrome signal it was coded into a very stable high-frequency sine wave, the frequency of which had to be chosen very carefully to keep interference to a minimum for those people who still viewed in monochrome.
Figure 1.1 A television frame comprises two fields (a), which are interlaced (b) to produce a frame.
In the PAL system developed in the UK, each frame contains 625 lines of video information, divided into two fields of 312½ lines. Each line of picture information contains a signal allowing the colour information to be decoded. In some countries on the European mainland the French SECAM system is used. This has the same frame rate and number of lines per frame as PAL, and a stable high-frequency sine wave is also used to carry the colour information, but the decoding information is placed in the start of each field. In the USA, Japan and a number of other Asian countries the NTSC system is employed. In this system the colour information is also coded into a high-frequency sine wave but in a simpler manner than either the PAL or SECAM systems. There are 525 lines to each frame, divided into two fields of 262½ lines.
Video source synchronization
In all sequential scanning systems it is important that all items of equipment concerned with viewing or processing the scene maintain synchronization, so that each is dealing with the same frame, field and line of the scanned picture at the same instant as the other items of equipment. Each line starts with a clearly defined and identifiable pulse (the line synchronizing pulse), followed by a short interval before the picture information starts. The period of time containing both this pulse and the interval is called Tine blanking' (Figure 1.2).
Figure 1.2 Video information extends from 0.3V to 1.0V. Sync information extends from 0.3 V to 0 V.
Each field of information is preceded by a complex series of narrow and broad pulses which define both the start of the field and the particular field within a frame (Field 1 or Field 2). The start of the frame is indicated by the field having a half-line space between the narrow pulses and the first synchronizing pulse. A series of lines containing no video information follows this series of pulses. The period of time containing this series of pulses and the blank lines is called 'field blanking' (Figure 1.3). In the PAL system, picture black (blanking level) is represented by a signal level of 0.3 V, peak signal level (peak white) is 1.0 V, and the synchronizing pulses go below blanking level down to 0 V. In the NTSC system picture content from blanking level to peak white is represented by 100 IRE units, and sync pulses by 40 IRE units. The 1 V peak-to-peak value remains the same, but sync tip level is 0.29 V below blanking. The combination of synchronizing pulses and detail of the scene brightness is called the luminance signal.
Video bandwidth requirements
As Figure 1.4 illustrates, there is a clear relationship between frequency and the ability of a system to resolve fine detail. A detailed examination of the arguments that relate the frequency response of a system with resolution obtained is out of place here, but a television system needs to be capable of handling a frequency range (bandwidth) of 25 Hz to 5.5 MHz in PAL systems, 30 Hz to 4.5 MHz for NTSC.
Figure 1.3 There are some 25 lines in each field clean of any picture information. Seven of these are used for field synchronization. This leaves 18 available for other purposes.
Figure 1.4 The ability of a video system to convey fine detail is dependent on its frequency response.
Adding colour to the signal
The perception of the human eye to colour information detail is much lower than to luminance because the eye contains fewer receptors for colour than for luminance. This is fortunate, as it permits colour information to be processed as a comparatively low-resolution signal. This 'chrominance' signal thus requires a narrower bandwidth than the luminance signal, typically about 1 MHz, and can be incorporated into an existing monochrome signal, as Figure 1.5 illustrates. Although the human eye is capable of identifying a large number of different colour shades (hues and saturations), it is possible to re-create most of them by combining just three colours (red, green and blue) in an additive mixing process, so that, for example, red arid green combined in various proportions can produce an extensive range of oranges, yellows and browns. The colour information presented by the camera is mixed (matrixed) into two signals, called 'colour difference signals', which vary (modulate) the amplitude and phase of a high-frequency signal, called the colour subcarrier, which is superimposed on the luminance signal. If this sub-carrier were to be an exact multiple of the line frequency, the patterning caused would be very obtrusive. To minimize this, the phase of the sub-carrier with respect to the start of each line is shifted by 90° on successive lines. In the NTSC and SECAM. systems this phase shift results in a sequence that repeats every four fields. In the PAL system a further phase inversion on alternate lines reduces the effects of phase distortion in the transmission/reception process. This results in the phase relationship between the sub-carrier and the start of each field repeating only once in every eight fields (the eight-field sequence). These phase relationships have to be taken into account in the video post-production process if there is to be minimum disturbance to the colour (chroma) information during editing.
Figure 1.5 The chrominance information is contained within the luminance frequency bandwidth. Care has to be taken to minimize mutual interference.
With the need for an exact relationship between the colour sub-carrier phase, line and frame frequencies, television systems must run at very precise frequencies. In the PAL and SECAM systems this caused no particular problems; sync pulse generators were used that had highly stable internal clocks, instead of using the mains as a reference. However, engineers developing the NTSC system ran into problems with transmitting colour information at a precise 30 Hz frame rate, and had to reduce it to approximately 29.97 fps. When shooting or editing in NTSC, account has to be taken of this somewhat odd framing rate.
Colour difference signals
The red, green and blue colour components are not used in their raw form. Instead, the red (R) and blue (B) signals are each combined with the luminance signal by subtracting the luminance signal (Y) from the red and blue separately to give R—Y and B—Y. In this form they are known as 'colour difference signals'. It is these signals that modulate two separate feeds of colour sub-carrier. In the PAL and NTSC systems these two colour sub-carrier sine waves are identical in frequency, but are held 90° out of phase with each other. In the PAL system the phase swings on alternate lines between advance and retard (hence PAL or 'phase alternating line'). This reversal of phase on alternate lines was considered necessary in the development of the PAL system in order to minimize colour degradation should there be any decoder misalignment. The two modulated sub-carriers are added together to give a resultant sine wave whose amplitude and phase depends on the proportions of red and blue present in the original signal. It is this signal that is superimposed on the luminance signal as the chrominance signal. The SECAM system, on the other hand, employs sub-carriers of two different frequencies, carrying R and B information on alternate lines.
In all systems it is the amplitude of the modulated colour sub-carrier that represents the saturation of the colour. In PAL and NTSC systems the phase of the sub-carrier is compared with a reference signal (the 'colour burst') to determine the hue (Figure 1.6). The colour burst is inserted at the start of each picture information (active) line in the line blanking period, just after the line synchronizing pulse. In the SECAM system the decoding information is carried in several lines within the vertical interval,
Figure 1.6 The instantaneous value (amplitude and relative phase) of the colour sub-carrier determines both intensity and hue of the chroma.
Any modulation process generates additional frequency bands ('sidebands') which extend both above and below the carrier frequency. In all systems these sidebands sit within the luminance bandwidth, making it impossible to remove the sub-carrier completely from the luminance signal.
Component analogue video systems
It is fair to say that if colour television were to be invented with the technology that exists today, a sub-carrier system would not be employed. Available technology has been exploited to develop colour television systems, known as component analogue systems, which avoid modulating the chrominance information onto the luminance signal. In the two component analogue VCR systems in use today, Betacam (and its development) Beta SP from the Sony Corporation, and Mil, developed by National Panasonic, the colour difference signals are processed without the need for a colour sub-carrier. For distribution purposes, three signals have to be sent around a building instead of just one, but the additional complexity is more than compensated for by the improvement in picture quality and the reduction of signal degradation during post-production.
The two chrominance information signals are recorded onto tape in time-division multiplexed form, line by line. Both luminance and chrominance signals are recorded with timing signals, instead of the traditional synchronizing pulses, or colour burst, to allow a more favourable packing density on tape. The two chrominance signals, known as Pr and Pb, are recorded on a part of the tape completely separate from the luminance signal. They are time-expanded and demultiplexed on replay.
When a component VCR is recording signals decoded from a composite source, some form of sub-carrier phase identification is needed. This takes the form of a pulse placed in the chrominance channel and a line of subcarrier ('vertical interval sub-carrier' or VISC) inserted into a line within the field blanking interval. The manner in which the chrominance is recorded, together with its limited bandwidth requirement, makes it possible to record two high-quality audio signals on frequency-modulated carriers along the chrominance tracks.
Magnetic recording
The basic principles of magnetic recording are reasonably well known; a plastic tape, coated with finely-divided magnetic powder, is passed at constant speed in intimate contact with a pair of magnetic poles called a 'recording head'. Currents flowing in the head coils cause corresponding variations in magnetic flux. The particles of magnetic powder are magnetized to varying degrees, depending on the strength of the current flowing in the recording head coils.
Magnetic replay
During replay the magnetized particles are caused to pass at constant speed in front of a similar head, where their external flux links with the head coil, generating a voltage. The strength of this voltage depends approximately on the rate of change of the magnetic flux rather than on its absolute level, so that the voltage that appears is the first derivative of the magnetic flux. This differentiation of the flux strength is modified by a number of factors, including the inductance and resistance of the head coil and the intimacy of contact between the tape and the face of the replay head (Figure 1.7). The consistency of the tape coating will also have an effect, variations in coating consistency resulting in corresponding variations in sensitivity. Should the particles of magnetic powder clump together, or be absent (perhaps as a result of poor adhesion), there will be a momentary loss of signal, or 'dropout'.
Figure 1.7 At high frequencies tape heads suffer from a variety of effects causing loss of available output voltage.
Implications for video recording
The variations in linear tape speed that are acceptable for audio signals are unacceptable for video processing where timing and synchronization are critical. For example, the accuracy of colour sub-carrier timing has to be within ±0.01 μs for video editing. In videotape recordings, variations in coating thickness and tape-to-head contact would cause unacceptable variations in brightness of the replayed picture. The video bandwidth extends over 18 octaves. Over this range, the differentiation effects described above would produce variations in replay output levels of over 108 dB. Differentiation also results in sine waves being effectively shifted in phase by 90°, and rectangular waves being reproduced as positive- and negative-going spikes (Figure 1.8). Obviously this is unacceptable, so some way has to be found of overcoming these problems.
Figure 1.8 A sine wave (a) when differentiated becomes a cosine wave (b). A waveform comprising a sine wave and its 3rd harmonic (c) will produce the waveform (d) when differentiated. A square wave comprises a sine wave and an infinite number of odd harmonics; (e) comprises a fundamental sine wave and odd order harmonics up to the 15th. When differentiated the waveform (f) results.
Use of frequency modulation for video recording
The video signal is not recorded directly onto tape, but is used first to modulate the frequency of a constant-amplitude carrier. The resulting signal is recorded as a magnetic imprint on tape. In this way the effects that result from variations in replay levels that occur as a result of variations in tape/head contact and magnetic coating inconsistencies are reduced. Frequency modulation also allows the d.c. component of the video signal, representing brightness, to be recorded (Figure 1.9).
Although current magnetic tape and replay head technologies mean that wavelengths less than 1 μm can be recorded and replayed, there is still a requirement for tape-to-head speeds to be much higher than is realistically possible with longitudinal tracks traditionally used for analogue audio recordings. This is achieved by the use of helical scan techniques.
Figure 1.9 Unmodulated carrier has constant amplitude and frequency (a). This carrier is modified by a signal (b) to change its frequency but not its amplitude (c). Specific carrier frequencies (d) can represent specific voltage levels (e). Sync tip, blanking and peak white may thus be represented by an a.c. signal and so recorded onto tape. The difference in frequencies between sync tip and peak white is known as the deviation (f).
Use of helical scan to improve write/read speeds
The high tape-to-head speed required for video recording is obtained by having the heads mounted on a spinning drum, the complete assembly being called a 'scanner'. The tape is wrapped around the drum in an open spiral. This results in a series of recorded tracks being laid in shallow diagonal lines across the tape (Figure 1.10). In analogue video recording each of these tracks corresponds to an individual field of information. In digital video systems the information related to an individual field may be recorded over a number of tracks. In this manner, although the linear speed of the tape may be quite low, the writing speed will be very high One typical system employs a linear (longitudinal) tape speed of 0.066 metres per second (m/s), but has a writing speed of 6.9 m/s, a writing-to-linear speed ratio of the order of 100:1.
Some video recording formats leave guard-bands between the recorded video tracks, others (notably component analogue and digital) may make use of azimuth recording techniques, where the write/read heads for luminance and chrominance tracks have azimuths offset in opposite directions. In azimuth recording, the heads are slightly wider than the recorded tracks, so will partially over-write (and over-read) adjacent tracks. The offset in the azimuth angles minimizes the resulting crosstalk to acceptable levels. Some digital audio systems also employ helical scanning techniques, both to minimize the linear tape speed (smaller cassettes = longer recording times) and to accommodate the very high rate of data that digital systems incur.
Figure 1.10 The videotape is wrapped as a part helix around a spinning drum (a). Heads set in the drum write and read diagonal tracks on tape as it moves relatively slowly through the transport system. Note in practice these diagonal tracks are at a very shallow angle.
Control track
When material is recorded on tape as a series of stripes, some way has to be found of ensuring that the scanning head (on both replay and over-recording during editing) follows the originally-recorded tracks. If that recorded material is video information, some way has to be found of ensuring that the individually-recorded fields can be correctly identified, particularly as regards the correct relationship of colour sub-carrier to field. This relationship is still important even with a component analogue recorder, as such a machine may well have to interface with equipment of composite format during production, post-production and playout.
Figure 1.11 In the D-2 recording system the control track carries servo reference signals together with video and colour framing pulses. Figures given are for the PAL system, those for the NTSC system are in parentheses.
One way of achieving this is by recording, on a longitudinal track, a series of pulses which enables the scanning heads to follow the prerecorded tracks correctly on replay. This track is called the 'control track'. It will often contain other pulses that identify the 8- or 4-field colour-framing sequence, and in digital video systems may carry information concerning the packaging, in segments, of the digital data on tape (Figure 1.11).
As the control track is frame-related, it can be used to control the editing process. However, it cannot be read while the tape is stationary or is moving at very slow speed (as when starting or stopping), because the replayed level is either non-existent or very low. A unique time identification for each frame, with this time related to the colour frame sequence in a known and unambiguous way, is the only satisfactory option.
Timebase correction
The timing accuracy necessary for the replay of video information is not possible using purely mechanical or electromechanical systems because of the inertia present with any mechanical device. The timing has to be corrected electronically. One method of doing this within a broadcast-quality system is to convert the off-tape demodulated signal into digital form. A digital signal can easily be stored until the correct time for transfer out of the machine, when it will be converted back into analogue form. This process is performed by a timebase corrector (TBC), a device which often incorporates a facility to adjust video and sync levels, and overall system timing and phase. Very often the vertical interval will be regenerated within the TBC, especially if it is external to the machine. Note that if the machine has to interface synchronously with the outside world, some form of external reference will be required by both the videotape machine and the TBC. The TBC may also be capable of outputting a correctly-timed video signal even while the machine is playing at nonstandard speed, perhaps for special effects.
Dropout compensation
The high packing densities employed in video recorders, mean that momentary loss of output due to dropout is going to be much more noticeable than with audio. A dropout lasting just 1/15 000 s would result in the loss of a complete line of video. To minimize the effect, analogue video recorders employ devices called dropout compensators. Basically, these devices are short-term (1-line) stores constantly replenished by the FM video signal coming off tape. When a dropout occurs, signalled by a drop in the off-tape FM signal, the output to the demodulator is switched from the direct to the delayed signal in the store, and an uncorrupted line replaces the one with dropout.