Mouseover the Fig. will bring up the relevant picture
Clicking on the picture will close the window
One aspect of the televising of the 1966 World Cup football matches that attracted a good deal of interest was the ease and speed with which slow-motion recordings of the highlights were presented, sometimes only seconds after the event. This article describes how the complex equipment that achieved these results was developed and built in a relatively short period.
A SLOW - MOTION VIDEO TAPE
RECORDER
by P. RAINGER, B.Sc., C.Eng., A.M.I.E.E.
IT IS well known that television standards in the United Kingdom provide 50 fields /s; in contrast with film, which has only 24 fields (or rather frames), there is sufficient information to permit the slow-motion reproduction of a recording obtained from a standard television camera. There have been a number of attempts to make slow-motion replay equipment. A Japanese system was developed and used for the 1964 Olympic Games, but, as this was designed for operation on 60 field standards, it is not applicable to the United Kingdom 50 field system. The machine was also large and complex. Other systems have been tried abroad with varying success. It was therefore suggested that some form of slow-motion video tape recorder (v.t.r.) should be constructed in Britain, using available standard v.t.r. equipment and employing techniques which required the minimum of new development. In this case, maintenance would be assisted by the spare parts and the extensive experience which existed for basic machines of this type.
Standard record and replay
After some consideration, it was decided that the ideal machine should be able to replay in slow motion standard transverse-track 2in tapes. The machine should have the following facilities :
  1. It should have a standard record process and be capable of making tapes identical with those recorded elsewhere.
  2. The machine should be capable of a standard replay process and produce satisfactory pictures at normal speeds.
  3. It should be able to replay these standard tapes in slow motion.
  4. The machine should be able to produce a still picture ('freeze') from either the incoming pictures, the normal-motion replay pictures or the slow-motion replay signal. The transition to this mode of operation should be instantaneous.
Although it was expected that this machine would be of use in a large number of different sports programmes, the first area of application was to be World Cup football (the Jules Rimet Trophy series, 1966). For this purpose, it was decided that the machine would have to operate on 625 lines, using an f.m. standard known as the EBU interim standard (black level 5Mc/s; white level 6.8Mc/s). Operation on 405 line standards would also be possible on changing one or two units.
Simulated slow-motion reproduction was obtained by step-printing motion-picture film. The results were studied with a view to determining what degree of slow motion could be tolerated without causing the signal to appear to be unduly jerky. Our conclusions were that a slow-motion reproduction of one quarter to one fifth of the normal speed would be valuable as a programme facility; this only exhibited a jerky motion on a few very difficult pictures. Although the Japanese used a 5:1 slow-motion reproduction, they were able to work on a 60 field/s television system; therefore it seemed reasonable to settle on a 4:1 slow-motion reproduction in the United Kingdom.
Preliminary studies also demonstrated clearly that the repetition of a complete picture of 625 lines gave an unacceptable result, and the slow-motion equipment would have to operate on single fields, i.e. 312½ lines. A study of the technical difficulties which may arise suggested that both 4:1 and 5:1 reproduction could be obtained equally easily, but that it would be relatively difficult to change the speed of reproduction once the design had been completed. It was therefore decided that the machine should be designed to produce a fixed 4:1 reduction in the speed of movement.
In order to obtain a 4:1 slow-motion reduction, it was proposed that we should replay a standard tape in an intermittent manner. The signal leaving the video tape recorder would therefore follow the cycle shown in Table 1.
Table 1. Record-replay cycle for v.t.r. and disc store
The gaps in the Table are filled with a signal from the magnetic-disc field store; the cycle involves the use of two heads and two alternate tracks on the disc
.
TIME V.T.R. OUTPUT DISC-STORE OUTPUT
HEAD 1 HEAD 2
Output field 1 Replay field A Record A Replay *
Output field 2

Replay *
Output field 3
Replay A Erase *
Output field 4
Replay A Erase *
Output field 5 Replay field B Replay A Record B
Output field 6
Replay A
Output field 7
Erase A Replay B
Output field 8
Erase A Replay B
Output field 9 Replay field C Record C Replay B
Output field 10

Replay B
Output field 11
Replay C Erase B
Output field 12
Replay C Erase B
Output field 13 Replay field D Replay C Record D
Output field 14
Replay C
Output field 15
Erase C Replay D

Magnetic-disc store
It was further proposed that the gaps in the sequence shown in Table 1 should be filled with the signal from a magnetic-disc field store. The field leaving the video tape recorder was therefore recorded on the disc store and repeated a total of four times before the next field arrived. In order to achieve the record, erase and replay cycle for the disc store as detailed in Table 1, it is necessary to use two heads and to record two tracks alternately on the magnetic disc. In order to minimise the cyclic variations which were likely to occur, all four fields would in fact be obtained by replaying from the disc store; the original signal from the v.t.r. is not seen at any part of the cycle.
The only mechanical work carried out was the modification of the v.t.r. to provide intermittent motion of tape. It was decided to use a series of tape rollers (Fig.1) to guide the tape into an omega-shaped path. In order to reduce the inertia of the system to a minimum, the tape guides are pumped-air bearings and do not rotate in any way. The two guides B and B' are linked together and attached to the arm A, the guides E and F being stationary. The tape is pulled by the conventional capstan mechanism at one quarter of its normal speed, and the arm A oscillates backwards and forwards, driven by an eccentric coupled to a synchronous motor. This synchronous motor is driven by a 12½ c/s signal, so that, when the tape velocities due to the combined action of the capstan and the oscillating arm are in the same sense, they add together, to cause the tape to travel past the video heads at the required normal speed of 15 in/s (Fig.2).
In order to replay satisfactorily at least one complete field each cycle, it is necessary to ensure that the tape velocity remains correct over this period of 20ms and is suitably phased, i.e.'tracking' the recorded signal. The eccentric attached to the driving arm A causes a sinusoidal displacement of the tape, and the quasilinear-motion (constant-velocity) part in the centre of the sinewave is used to produce the required signal. In practice, the linear portion of the sinewave does not extend far enough, and a simple 'following-link' accelerator is used to couple the motor to the eccentric. This expands the linear portion of the tape motion to include a number of lines over and above the required field. This is necessary, because it is of the utmost importance that the field-synchronising waveform leaving the video tape recorder should be complete, as this is used to control many other functions in the record-replay process of the slow-motion machine.

Mechanical alterations

It was also necessary to keep the inertia of all oscillating parts to a minimum, and the rollers E, F, B and B' were constructed as fixed guides, lubricated by a supply of air fed to each bearing surface. The original design was based on the use of a 20lbf/in² air supply to feed the air bearings. In practice, the tape lifted off the rollers with only 11bf/in², and the normal operating pressure was adjusted to be 5lbf/in². The only other mechanical alterations to the v.t.r. machlne involved mounting the control panels and fault indicators for the complete equipment alongside the normal v.t.r. controls and moving the two tape spools further apart, to make way for the slow-motion mechanism.
Detailed consideration of the tape motion due to the combined effect of the oscillating arm and the capstan reveals that it momentarily reverses at one point in the cycle. A second capstan and pinch roller were therefore added on the other side of the replay-head assembly, to keep the tape under tension at all times. A general view of the tape-transport mechanism is shown in Fig.3.
As stated earlier, it is necessary to carry out the complex record, erase and playback cycle shown in Table 1. It was decided that the signal would be modulated on its f.m. standards at the beginning of the record cycle, and that, when replaying a 2in tape, the f.m. output signal would remain in this form throughout the equipment. In particular, the field-store and delay-line-carrier systems all operate on f.m. It was therefore necessary to construct fast-acting electronic switches which could operate at this rate; it had to be ensured that switching the f.m. signal did not cause undue difficulty due to switching transients appearing at the output. As already explained, two record-replay heads are used on the disc store, and it was also necessary to ensure that a very high degree of isolation existed between these two heads. At one part of the cycle, when, for example, head 1 is recording, there is a signal of approximately 100V across this head; at the same time, head 2 is providing a replay signal of approximately 5mV. The isolation problem was such that both electronic switches and reed relays were used in cascade; even then, the results were only marginally satisfactory.

Line delays

As a result of this switching operation, the output from the field store consists of four even and four odd fields alternately. The field timing at this stage should be a regular 50c/s, but the line rate would have a phase discontinuity or jump of 180° at the beginning of some fields. In order to produce a standard waveform, it is necessary first to ensure that the video signal is based on a line-synchronising signal of a continuous nature. It is therefore necessary to insert a ½ television-line delay in the rather curious cycle shown in Table 2.

Table 2. Timing cycle of switches S2 and S3

The half-line delay is simulated by the alternate insertion of a 1-line delay and a 1½-line delay.

OUTPUT FIELD 1 2 3 4 5 6 7 8
Field from disc store EVEN EVEN EVEN EVEN ODD ODD ODD ODD
½-line video delay (S2) OUT IN OUT IN IN OUT IN OUT
½-field trigger delay (S3) IN OUT IN OUT OUT IN OUT IN

In practice, a satisfactory half-line delay was not available, and this effect had to be simulated by alternately inserting a 1-line delay (lH) and a 1½-line delay (1½H). Large bandwidth quartz delay lines were employed, operating at 30 Mc/s, and it was necessary to construct frequency changers to move the f.m. signal up to this band and back again.
A continuous line frequency having been achieved by inserting appropriate delay, the regular sequence of the field-synchronising waveform has now been disturbed. It is necessary to add one half-line delay of the field-trigger signal in an inverse cycle (Table 2) to obtain a correct waveform.
A simplified block diagram of the complete machine is shown in Fig.4.
On operating the switch S1 the input to the field store may be changed from the input signal to the v.t.r. output operating in slow motion or at normal speed. Selecting the correct switching sequence to S2 and S3 provides the required normal-speed or slow-motion output. If the switch S1 is moved to the lower position, no further signal is given to the field store, and there remains a still picture extracted from whichever source had been supplying the field store. The transition to 'freeze'is achieved in a near-synchronous manner, i.e. without undue disturbance of synchronism. During all modes of operation, the machine is approximately field-synchronous with the incoming signal or station pulses.
The normal method of operation is to replay the required sequence in slow motion and then terminate this slow-motion excerpt by freezing the picture at the end of the recorded sequence. The programme can then cut away from the frozen picture at a convenient time. It is, of course, possible to use the facilities quoted here in many different ways, and experience will show which are the most valuable. An overall view of the v.t.r. and the associated slow-motion controls is shown in Fig.5. On the left of the v.t.r. is a small control panel to start and stop the disc store and to select, as required, one of ten alternative tracks on the disc store; a total of ten still pictures can be recorded if required.
The right-hand side of the machine supports a similar control panel, which carries the controls determining the mode of operation. This panel also contains fault indicators, to assist maintenance and adjustment of the machine.

Successful operation
The mechanism can give a most interesting picture, providing a new insight into many sporting events. Although there is some loss in resolution and some degradation of the signal/noise ratio, these would not appear to be major problems. Successful operation of this machine depends on obtaining in the first instance a picture which contains sufficient information. As photographers well know, exposures of 20ms are not short enough to stop very rapid motion. The television camera cannot have exposures shorter than this duration without sacrificing sensitivity, and all cameras in use today have an exposure time of 40ms in theory, although in practice the effective exposure time lies between 20ms and a few hundred milliseconds. Slow-motion reproduction is used to study fast action. However, fast action will involve movement blur, which, although acceptable when seen for a fleeting moment, can be quite disturbing when studied at leisure. Defects of this sort are fundamentally unavoidable if we use standard television cameras. An improvement in this respect would require the use of a special slow-motion film camera or the equivalent nonstandard television camera.
Similar, but not so important, difficulties occur when using the 'freeze' facility, because, when all movement ceases, the frozen picture is sometimes quite difficult to interpret. The successful operation of this equipment also depends on the synchronous operation of the video tape recorder, oscillating arm and field store. In particular, synchronism must be maintained between adjacent fields to an accuracy within a few tens of nanoseconds. It was hoped that the inertia of the video tape recorder and the field store would be sufficient to maintain reasonable timing at least over the short period of one field, i.e. 20ms.
The experimental results obtained so far suggest that, while it might be possible to reduce the timing errors on the v.t.r. to about l s per field, the field store has much larger variations in its speed. These variations are such as to cause a discontinuity of line synchronism at the beginning of the field, which is not within the range of correction by the variable-delay devices often associated with video tape recorders. In practice, the signals would be acceptable for most driven timebases, but some 'flywheel' time-bases with long time constants would be disturbed in an objectionable manner.
It was therefore necessary to resort to standards-conversion techniques, to ensure that the output synchronising waveform was beyond criticism. In practice, it was most convenient to use an optical standards convertor (i.e. one using a display and camera combination) for this standards-conversion process; the movement blur normally associated with convertors of this type is of negligible importance, because slow-motion techniques have reduced the speed of movement to the point where the convertor can contribute no further loss. There are, nevertheless, the normal losses of resolution etc. associated with this method of standards conversion.
Storage capabilities
No mention has been made of the deficiencies and errors in the vertical resolution due to creating a sequence of four odd and even fields from one single field, i.e. 312½ lines. Techniques used in line-store standards convertors and similar electronic devices have shown that suitable interpolation between lines can give a much improved result. Construction of a fully interlaced picture using these techniques has not been employed in this device. The result is an apparent vertical hop of fine-detail picture information, which, although somewhat disturbing on test waveforms, is acceptable on all practical camera signals. The storage capabilities etc. of the optical convertors employed with this equipment do much to remove this difficulty, even on the most difficult pictures.

Much credit is due to the engineers who assisted in making this relatively complex equipment and completing the project in about nine months. In particular, I would like to acknowledge the efforts of D. P. Robinson and P. White of the BBC Designs Department and their colleagues throughout the BBC.
I would also like to acknowledge that, without the Siemens & Halske disc store and the Ampex recorder, this project would not have been possible. Its success is also due in part to Emitape Ltd., who developed and manufactured at short notice the discs used in the field store. My thanks are also due to the Director of Engineering of the BBC for permission to publish this article.

Electronics & Power November 1966 pp 386-389