Fullscreen HTML5Med Res (??MB)HTML4

The History of Videotape – part 2 Helical Scan By Ian Batty, Andre Switzer & Rod Humphris Last month, we described the major innovation that was the Ampex quadruplex videotape recording and playback system. Of course, technology did not stand still, and it was only a few years before more breakthroughs were made, enabling not only better video quality but also some significant new features... Thanks to the Toshiba Science Museum for use of this image: toshiba-mirai-kagakukan.jp/en/learn/history/ichigoki/1959vtr/index.htm A mpex’s quadruplex video recording was a revolutionary technology. Casting off the existing linear tape paradigm, Alex Poniatoff’s company invented a system where four tape heads, mounted on a spinning disc, scanned the tape transversely. Coupled with the adoption of frequency modulation, ‘quad’ established videotape recording (VTR) machines as television broadcasting’s workhorse for replay, editing, distribution and archival work. Yes, the first VTRs were horrendously expensive, and the size of a few refrigerators. And yes, the tape is not entirely robust – it can break and 64 Silicon Chip distort. But its added flexibility was well worth it for news and broadcast companies. For the rest, videotape recording was out of reach. But the principles established by quad were sound: rotating head scanners and frequency modulation were clearly the way ahead. If only someone could devise a simpler, cheaper system. And it would be helpful for it to produce a picture in pause, or at slow or fast picture search; things impossible with quad. Enter Toshiba Dr Norikazu Sawazaki at Toshiba’s Matsuda Research Laboratory develAustralia’s electronics magazine oped a prototype helical scan recorder in 1953. The first experimental VTR1 was completed in 1958 and demonstrated to the public in September 1959. Commercial production of the new videotape recorder followed. At around the same time, Eduard Schuller of Telefunken had also devoted himself to the recording of television signals. Having already invented the “ring-shaped” audiotape head still in use today, he was awarded a 1953 patent for magnetic recording and playback of television pictures using helical scanning. The tape runs around the head drum, giving much longer video tracks siliconchip.com.au Fig.9: the basic concept of helical scan recording. The tape is wrapped around a drum head at an angle so that as the head spins, it scans diagonal strips. This means that the diagonal tracks overlap continuously along the length of the tape, avoiding the segmentation necessary with the quad system. Fig.10: this gives you an idea of how the tracks are laid down on the tap in a helical scan system. While they are diagonal when the tape is laid flat, when the tape is wrapped around the drum, the tracks actually form a helix shape. than was possible with quadruplex. Figs.9 & 10 show a simplified single-head system. The tape engages the head drum (the scanner) high and exits low, so the system records a number of slanted tracks at a shallow angle of perhaps 5°. Viewing the tape on the drum, the video tracks appear as a series of spirals, a bit like a coil spring, hence the term “helical scanning”. Early helical-scan VTRs used the available 2-inch tape. Despite not needing vacuum air to form the tape path, they were hardly more compact than their quad predecessors. A slower tape speed of 3.7ips allowed five hours recording or playback on 12.5inch tape reels. Video recording and playback demand continuous head-to-tape contact. Quad solved this by always having one of four heads engaged with the tape, and switching to the active head, but this resulted in the possibility of mismatches causing head banding. Helical scanning aimed to record an entire field of 312.5 lines over 20ms in a single scan over the tape. This demanded a much longer track length than quad’s 46mm, with its 16 lines per scan. Quad systems were able to record signals in the megahertz range by virtue of the high headwheel speed, and helical scan would also need high head-to-tape speeds. siliconchip.com.au Helical scan needed to use realistic tape speeds, say 7.5ips, but the headto-tape speed needed to be in the order of 20m/s. The solution was to use a head drum with a large enough diameter to give the required head-totape speed for FM recording. The Ampex 5800~7900 series VTRs (Fig.11) used a head drum diameter of 135mm, creating a track length of some 425mm. This gave a writing speed of just over 11m/s, adequate for FM recording. They matured with the 7950, a timebase-corrected VTR capable of broadcast performance. Using a single head with one field for each scan of the tape, this system’s head drum rotated at 50 revolutions per second (3000 RPM) for our CCIR/PAL standard. But with such a long track, tape tension has much more effect on the horizontal rate. Television broadcasters had been the market for the first generation of VTRs, and broadcast demands very stable images. With a track length of only 46mm laid across the tape (and thus much less affected by tape stretch), quad’s greater immunity to tape variations meant that it remained the preferred format. Helical systems would have to play catch-up for some time. Broadcast vs non-broadcast video tape recorders As described in the last article, broadcast VTRs must be locked to station sync, both in frequency (to prevent vertical rolling or horizontal drifting) and in-phase (to register VTR pictures over the station program). But if a VTR program is to be replayed on a local monitor, or sent Fig.11: an Ampex helical-scan VTR which used 2-inch tape. One big advantage over the quad system was lower tape speeds, which meant longer recording and playback times (more photos at www.ebay.com/ itm/182696338060). Source: www. labguysworld.com/Ampex_VR-660.htm Australia’s electronics magazine April 2021  65 done quickly and accurately. The long tracks of helical-scan formats made cutting-and-splicing impractical, so helical systems need to use electronic editing (re-recording) in some form. The end of segmentation Fig.12: this type of ‘flag waving’ image distortion was a result of timing errors due to the tape stretching slightly, or the tape or head speed varying slightly between recording and playback. to a non-station destination, the rigid demands of broadcast don’t apply. Non-broadcast equipment can have relaxed timebase stability, as the VTR will supply vertical and horizontal references for any destination equipment: monitors, other VTRs, etc. Non-broadcast programs may be in colour, and of high visual quality; non-broadcast does not imply poor quality. The best off-tape video may be as good as – or better than – off-air programs. Non-broadcast just means that the destination equipment is more tolerant of variations in the exact line and field rates and phasing of video signals. Domestic TV receivers were designed with high-performance timebases capable of locking to very weak signals. Such designs respond well to weak but constant signals. They do not easily tolerate signals with timing errors. It was common when early VTRs were fed to high-performing television sets for the TVs to lose sync with picture rolling or horizontal tearing (or ‘flag-waving’; see Fig.12). The solution was to speed up the monitor’s timebase response, allowing better tracking of the VTR video with its higher degree of timing errors. Tape editing Since quad recorded transversely, it was practical to cut-and-splice tape for editing. This skill, adopted from movie film editing, could be Segmentation – the splitting of a field into discrete scans – was a systemic problem with quad. The smallest mismatches in playback level, timing or equalisation caused problems. Helical scanning would solve this by recording an entire field in one scan of the tape. The simplest way of doing this was with just one rotating head. The head would need to be continuously in contact with the tape, so this dictated a full 360° wrap, as shown in Fig.9. VTR development was driven by the opportunity of bringing the technology to education, commerce and industry. A teacher could show a science video at any time, not just when it came to air. A sports coach could play back a tennis player’s serve and analyse just how to get that drop shot. A plant supervisor could not only explain, but actually show the company’s board just what the problem was. The rapid onslaught of solid-state technology and its radical miniaturisation of electronic circuitry helped, of course. No longer would VTRs be the size of several equipment racks. Before long, the physical mechanism would be the main determining factor on the size of VTRs. Just about every major electronics manufacturer would have a go. Get ready for the first VTR format war. Format wars: the first battle Helical scan systems use a rotating head, or heads, to provide the very Fig.13: the layout of the magnetic recordings on Ampex 1-inch helical-scan videotape. 66 Silicon Chip Australia’s electronics magazine siliconchip.com.au Fig.14: an Ampex VR-6000 helical scan VTR also used 1-inch tape. It went on sale in 1966 (more photos at www.ebay.com/itm/183994149450). Source: www.labguysworld.com/Ampex_VR-6000.htm high head-to-tape speeds used in all videotape systems. The head disk rotates within a drum, with the drum accurately guiding the tape to give just the right amount of head-to-tape contact and the correct head-to-tape path. The tape must be wrapped around the head drum, but how? Do we use a full 360° wrap, with just one head, or do we use a 180° wrap and two heads? (See Fig.15) A single head removes the problem of matching head amplitude/frequency differences. But since a 360° wrap implies that the entire tape width must be reserved for the video tracks, where will we put the control and audio tracks? The solution was for the video head to scan less than the full tape’s width. While this could be made to work, it left a short period of each video field unrecorded; there was an inbuilt dropout period in the video playback. But a two-head system could be designed so that the video heads scanned less than the full tape width, allowing for control and audio tracks. Since there were two heads, the design allowed each head to record a full field, with electronic switching guaranteeing an uninterrupted playback signal. Ampex & IVC 1-inch systems These two pioneers adopted the single-head, 360° wrap format using 1-inch (25mm) tape. They released incompatible 1-inch systems: Ampex (see Figs.13 & 14) used the “alpha” wrap while IVC used the “omega” wrap; both names are derived from the Greek letters. With Ampex’s alpha wrap, the tape is led around a near-90° entry guide before contacting the drum. The tape runs anti-clockwise. On exit, the tape is led around another near-90° exit guide. Tape loading is done with the two guides retracted. When ready, the operator closes the guides to give the correct tape path over the head drum. Fig.15: the three most commonly used helical scan tape paths. The alpha and omega systems have the advantage of only needing a single head. In contrast, with the omega system, there is no discontinuity in tape scanning, so any signals in the blanking periods are recorded. This was critical for broadcast use, and almost all videotape systems standardised on the two-head approach. siliconchip.com.au Australia’s electronics magazine The tape enters the scanner station from the reel table and ascends as it traverses around the drum, to exit one inch above the entry point, thus giving almost the complete 360° (see Figs.16 & 17). There is a small gap where the head loses contact with the tape, and thus creates a loss of signal. This is timed to occur during the vertical blanking interval. Although this prevents the disturbance from being seen, the loss of sync pulses during this dropout period renders the format fundamentally incompatible with broadcast standards. Each video track stores one field of signal. Audio is recorded on a conventional linear track using a bias signal. A control track is laid down during record to allow accurate scanning in playback. International Video Corporation (IVC) led the tape directly on to the head drum, also running it anti-clockwise. This meant that tape guiding was simpler than Ampex’s, but there was still a short gap in the signal. Like Ampex’s 1-inch system, the IVC format could not reproduce the entire vertical blanking period’s synch pulse block. Audio is recorded on a conventional linear track using a bias signal, and a second audio track, used for cueing, is provided. A control track is laid down during recording to allow accurate scanning in playback. German engineers working for Bosch-Fernseh broke out with BCN, a segmented helical scan system using a single 1-inch tape (Fig.18). With a high slant angle and a small two-head drum rotating at 9600 RPM, this system recorded only 52 lines per track. Like quad, it could not display a still picture, nor a picture during search. Released in 1976, BCN was widely used in Europe. A, B and C formats Ampex’s single-head, 1-inch system was developed to the point where its resolution was equal to quad’s. Capable of recovering and playing back the full video bandwidth, timebase correction (TBC) gave this system full-colour capability, but still with the loss of signal in the vertical synch block. This could be corrected by a digital TBC that re-inserted sync pulses (which they commonly do), but the format was not intrinsically broadcast-standard. It was, however, registered by the Society of Motion Picture and Television Engineers (SMPTE) as Type A in 1965. April 2021  67 Fig.16: this shows how the alphawrap system used in Ampex 1-inch helical scan VTRs was implemented. Bosch-Fernseh’s BCN was registered as Type B. Signal loss in the vertical synch block was more than a nuisance. It potentially destroyed vital engineering information: the vertical interval test signal (VITS). Not visible to the viewer, VITS was valuable to engineers and technicians. There was also the SMPTE’s vertical interval time code (VITC) that uniquely identified each frame on the tape, critical to editing and verification of events recorded on tape. So, if no 360° wrap system could record a full field, why bother trying? Why not allow a laneway in the slanted video tracks? By adjusting the phasing of the video head against that of the active vid- eo signal, it would be possible to start the video track someway in from the tape edge, but end it before the bottom edge of the tape. This means there is no loss of contact (dropout) period in the active video. However, the loss of the vertical synch block would have to be addressed. The solution was to add a second head to the drum, around 30° behind the video head. The second head simply recorded the vertical synch block, also without any loss of contact during its active period. So the system records (and plays back) the active video and the vertical synch block, both without any interruption or dropout disturbances. This was a system Sony pioneered. Fig.17: this IVC 1-inch omega-wrap VTR is mechanically a bit simpler than the Ampex VR-6000, and like the Ampex system it uses a single record-playback head. Source: https://youtu.be/EIhI85cHIfg 68 Silicon Chip Australia’s electronics magazine The whole field (active video and vertical synch bloc) could be recovered by switching and combining the outputs of the two heads. Known as the “one-and-a-half head” system, this reproduced an entire field with no gaps or losses. Ampex and Sony co-operated to mature and formalise their designs, registered by the SMPTE as “Type C” in 1976 (see Fig.19). It would supplant quad and become the open-reel standard, surviving into the 1990s. Easeof-handling, enhanced still, slow forward and reverse play and fast forward and reverse play made Type C the system of preference, especially for editing. Prior to Type C’s release in 1976, Fig.18: a Bosch-Fernseh Type B helical tape scanner head. Source: https://w.wiki/gyb siliconchip.com.au Fig.19: the layout of the Ampex/Sony “Type C” tape format of 1976. It supplanted quad to become the open-reel standard, surviving into the 1990s. single-head systems could not record an entire field without some period of signal loss. A two-head system can use each head to lay down an entire field, and reconstruct the whole frame from the combined, sequential output of the two playback heads. This eliminated the problem of signal dropout during the vertical interval. Two-head omega wrap systems Single-head systems require a complete 360° scan in 20ms (PAL/CCIR), giving a speed of 50 revolutions per second or 3000 RPM. A two-head system sees each head scanning only 180°, halving the drum speed to 25 RPS/1500 RPM (Fig.20). In practice, the wrap was slightly more than 180°, ensuring uninterrupted recovery of the entire video frame. Sony released an omega wrap twohead system, and 180° omega wrap became the preferred format for the successful and well-known ¾-inch U-matic, ½-inch Electronic Industry Association of Japan (EIAJ), Betamax, VHS, Philips VCR, Akai ¼-inch and Sony 8mm Video 8 systems. Two-head omega wrap was also used in digital audio tape (DAT), in computer implementations of DAT for data storage, and Digital Video (DV) handycams. Armistice: the EIAJ format The format wars came to an end when the Electronic Industries Association of Japan released the EIAJ-1 standard for half-inch open-reel videotape recorders (see Fig.21). Initially monochrome only, it was re-engineered for colour operation and appeared in at least two cartridge/ cassette systems. It was intended for non-professional use by businesses, schools, government agencies and hospitals but was also adopted by some consumers. Timebase errors remained For all of helical scan’s advantages, it was even less suited to broadcast than quad. With their long video tracks, helical format machines had worse timing stability than quad. Around the time that helical scan was being taken up, advances in semiconductor technology were delivering digital integrated circuits of some complexity. Digital signal processing, also in development, made it possible to digitise analog video signals. Fig.20: the mechanical layout of a basic two-head omega-wrap VTR system. siliconchip.com.au Australia’s electronics magazine April 2021  69 Frame store also freed cameras from the need for station lock. With a frame store, a remote non-synchronous camera feed could be accepted, then mixed in directly. Previously, such “outside broadcast” (OB) programs would be recorded, then played back from a station-synchronised VTR. On rare occasions, a producer would punch to the OB camera, and run the entire station in sync with the OB. Not desirable, but sometimes, “you gotta do what you gotta do!” DTBC technology advanced to the point that it could be offered in the pro versions of domestic video cassette recorders, such as Panasonic’s ProLine AG-1980. Colour made it harder Fig.21: a Sony EIAJ ½-inch VTR. Comparing this to its predecessors demonstrates the degree of miniaturisation which made Sony famous. More photos at https://historictech.com/product/sony-cv-2000-videocorder-c1965/ With digitisation came the possibility of highly-responsive timebase correction. The principle is simple: digitise the off-tape video at its own varying rate and store it in digital memory. Then read the data out of memory at the station sync rate, convert the digital data back to analog and deliver fully-corrected, station-synchronous video (see Fig.22). Early digital timebase correctors (DTBCs) had only enough memory to store a few lines, and could not correct a video signal unless it was vertically-locked to station sync. Further developments offered larger memories, and it eventually became possible to store an entire video frame. A frame store system can correct timing errors, but also to accept a video signal that is not locked to station sync. This allows any video signal with the correct format (PAL, NTSC etc) to be combined with station video sources. A version of frame store was used in the Bosch- Fernseh’s BCN system to display still frames, otherwise impossible with its 52-lines-pertrack format. Before the introduction of frame store, satellite feeds were commonly recorded and then played back on a VTR locked to station sync. Frame store allowed satellite feeds to be corrected to station sync, then mixed directly into station programs. Both PAL and NTSC encode colour (chroma) as a quadrature amplitude modulated (QAM) signal. This appears as a phase-modulated signal, and it must fit in the same bandwidth as the monochrome (luminance) signal. To reduce interference, the chroma signal has its carrier removed, leaving only the signal’s upper and lower sidebands. The problems of phase modulation are explained below. The receiver’s demodulators must have a suitable carrier to work, so a short reference “burst” is added at the start of each line of video, For PAL, it’s about 4.5µs of a 4.43361875MHz sinewave. This is vital to a receiver’s colour processing. This makes the stability problems even worse. NTSC’s chroma frequency is exactly 3.579545MHz, and PAL’s is 4.43361875MHz(!) Any colour system must deliver the colour (chroma) signal at very close to those precise frequencies. Also, both the American NTSC and European PAL systems encode colour signals using phase modulation. Even Fig.22: once digital technology had matured sufficiently, it became possible to implement timebase correction (TBC) mostly in the digital domain. This shows the basic layout of such systems. Once mature, they finally provided a simple means to interface a colour VTR to just about any broadcast system, providing stable phase, line and frame sync. 70 Silicon Chip Australia’s electronics magazine siliconchip.com.au Fig.23: hetereodyne VTRs accept the full colour signal, then use a low-pass filter to remove the chroma component. The remaining luminance is fed to the frequency modulator to create the FM signal for recording. On playback, the FM signal is demodulated to recover the luminace component of the original video signal. if the chroma signal frequency can be made accurate, any phase errors will cause colours to “slew” in one direction or another up and down the spectrum. Just a few degrees of phase error will be obvious, especially in the range of human skin tones. Given that the 4.433MHz PAL subcarrier has a period of only 225ns, an error of just 10ns translates to a phase error of 16°. That’s enough to make a healthy skin tone look either badly sunburned or dangerously jaundiced! Recalling the size and expense of quad machines, it was feasible to add colour correction and still sell the hardware. Correction used a recorded pilot tone signal. In replay, you would expect minor tape speed variation, and variations in tape tension, to affect all signal frequency/phases. Luminance phase and timing errors were corrected by the timebase corrector. Any errors in the pilot tone’s phase could be applied as a correction to cancel out errors in the chroma signal. See, for example, E. M. Leyton’s 1957 US Patent 2,979,558 (https://patents. google.com/patent/US2979558A/en). Helical scan systems had two particular barriers to proper colour operation. While quad could accommodate NTSC’s 4MHz bandwidth and PAL’s 5MHz bandwidth, only the highest-performing helical systems could meet this demand. 1-inch systems, siliconchip.com.au Fig.24: the tape bandwidth occupied by a monochrome video signal. As you can see, there is plenty of spare bandwidth to fit colour information. Fig.25: the bandwidth occupied by a composite PAL video signal. As can be seen, the chroma (colour signal) occupies a relatively narrow bandwidth centered on the chroma carrier frequency of ~4.43MHz. This allows the luminance and chroma signals to fit in the 5MHz original monochrome bandwidth, but with minimal interference with each other. Australia’s electronics magazine April 2021  71 Fig.26: this is the scheme eventually arrived upon to shift the colour (chroma) information to lower frequencies so that it can occupy tape spectrum not used by the FM luminance signal. such as Ampex’s VR-6000 (released in 1966, well after their first 1-inch outing) had a video bandwidth of only 3.5MHz, not enough even for NTSC (see Figs.24 & 25). Also, timebase errors in helical systems are far more severe than for quad. Even if a full-bandwidth colour signal could be squeezed onto a helical machine, colour correction would be vital even for CCTV use, let alone broadcast. To overcome both problems, they separated the chroma signal from the luminance signal and handled them separately. Colour television’s chroma (colour) bandwidth is quite small, despite its 3.58/4.43MHz carrier frequency; it’s -1.5/+0.5MHz for NTSC (a wider lower sideband) and -1.0/+0.6MHz for PAL. Now, there’s a lot of tape bandwidth not being used; even low-definition helical systems used signal frequencies above 2MHz for their low end. Fig.25 shows the 3.8~4.8MHz FM bandwidth of VHS. Sony’s U-matic and Betamax and JVC’s Video Home System (VHS) used the similar solution. The chroma content was filtered out, heterodyned (“down-converted”) to 626.953kHz (~627kHz), then recorded in the unused spectrum below the luminance signal. Fig.26 shows a simplified block diagram of this scheme, while Fig.27 shows the resulting on-tape spectrum for VHS. Yes, down-converting to around 627kHz reduced the colour bandwidth, and thus its fine detail, but this is domestic-grade equipment that’s not expected to give broadcast resolution. 72 Silicon Chip Just as recorded analog audio needs a bias signal to overcome tape non-linearity, so does this analog chroma recording. Happily, there is already a high-amplitude signal at maybe five to ten times the chroma frequency being recorded, ie, the luminance signal. So the luminance signal acts as a bias signal for the chroma, without creating any interference. On replay, the chroma signal is heterodyned (up-converted) back to 3.579545MHz or 4.43361875MHz, mixed with the off-tape luminance signal, and hey presto! Colour recording and playback. But let’s recall the problems of timebase errors, and the need to keep the chroma signal’s phase errors as low as a few nanoseconds. Now, converting the highly-precise 3.579545/4.43361875MHz signal down to 627kHz for recording, then (in replay) attempting to reconvert up to exactly 3.579545/4.43361875MHz with no frequency errors or phase jitter is a big ask. To keep the discussion simple, let’s consider a PAL colour signal, calling it 4.433MHz, and the down-converted signal 627kHz. Any heterodyne/colour-under system must be able to correct the chroma signal phase errors. Several different methods were developed, relying on newly-available digital circuitry to manage the down- and up-conversions with sufficient accuracy. The actual signal processing would continue to use plain old analog techniques. The mature solution arrived at by both Beta and VHS used a phaselocked loop (PLL) to generate the down-converter’s local oscillator, shown in Fig.26. This description uses VHS frequencies; Beta is similar. The PLL was locked to the incoming video’s line rate (15.625kHz), and it produced an output of 40.125 times Fig.27: the bandwidth occupied by the video signal after the processing shown in Fig.24 & 28. This assumes that the FM carrier for the luminance information is still over 4MHz; however, that can easily be changed to suit different tape speeds. Australia’s electronics magazine siliconchip.com.au Fig.28: how down-converted colour video signals are played back; it is basically the reverse of Fig.26. The colour signals must be recovered with very accurate phases and frequencies or the hues will be different from the originals. the line frequency (~627kHz). This was added to the 4.43MHz colour burst from the incoming signal to create the 5.06MHz local oscillator. The incoming video signal’s chroma component was filtered off through a 4.43MHz bandpass filter, then applied to the mixer, along with the 5.06MHz local oscillator, to produce the 627kHz chroma signal. This was combined with the frequency-modulated luminance signal and recorded onto the tape. Fig.27 shows the record signal’s spectrum Colour playback The hard part was up-converting the 627kHz chroma signal back to 4.433MHz in a stable manner. Remember that the recording LO was generated partly from the incoming signal’s line rate of 15.625kHz, and partly from the incoming signal’s chroma frequency of 4.433MHz. This means there was a fixed frequency ratio between the original and highly accurate 4.433MHz input chroma and the 627kHz down-converted signal. We can expect some phase errors and jitter in the off-tape 627kHz chroma signal. But this 627kHz signal was derived using a local oscillator phaselocked to the 15.625kHz line rate. So we can use the line rate itself as a stable reference for the replay up-converter’s local oscillator. And that’s what is done, as shown in Fig.28. A PLL recovers the 15.625kHz line frequency from the luminance playback circuitry, and creates a 627kHz reference. Another PLL recovers the 4.433MHz chroma frequency from the upconverter’s output. The local oscillator takes the 627kHz reference and the 4.433MHz chroma signal to create a local oscillator signal of 5.06MHz. The local oscillator is now applied to the up-converter’s mixer and heterodyned with the 627kHz off-tape chroma to produce 4.433MHz replay chroma. Using the replay signal’s line rate reference gives sufficiently good phase correction for a domestic colour television. The final stage in playback processing mixes the replay colour signal with the replay luminance signal, to Fig.29: once the luminance and chrominance signals have been extracted from the videotape, it is a relatively simple matter to mix them to produce a standard video signal, which a colour TV will happily accept. siliconchip.com.au Australia’s electronics magazine re-create the composite video output, as shown in Fig.29. Heterodyne colour systems are complicated, but were implemented for two reasons. First, it made colour recording possible on video tape systems that could not provide the full broadcast bandwidth of 4.2MHz (NTSC) or 5MHz (PAL). Second, heterodyne colour applies correction during replay, making the colour signal stable enough for display on monitors and television sets, and for editing and copying. The alternative to heterodyne colour’s replay processing would be a TBC in every VCR, making VCRs too expensive to market. That’s it for this article; next month, we will discuss the cassette systems that were used as a convenient means of storing and protecting videotape. Thanks to Randall Hodges, Richard Berryman and Rod Humphris for their help in preparing this article. References • A write-up on the history of video recorders etc: www.labguysworld. com/VTR_TimeLine.htm • Dana Lee’s website on TV and more: www.danalee.ca/ttt/ • An introduction to VCRs: https:// youtu.be/KfuARMCyTvg Many other videos on the above YouTube channel are also worth taking a look at. • Video Tape Recorders, 2nd Ed. Kybett, Harry, Howard W. Sams, Indianapolis, 1978 • Video Recording Record and Replay Systems. White, Gordon, Newnes-Butterworths, London, 1972 April 2021  73 Transports, Mechanisms and Servos As stated in last month’s article, this is a full description of the operation of servo motors as used in helical scans and the like. A tape transport draws tape from the supply reel, passes it over the heads and collects it on the takeup reel. The tape needs to move at a constant speed, and the usual mechanism is a spinning shaft (the capstan). The rubber-covered pinch roller presses the tape against the capstan to ensure a steady speed. Audio recorders, with their heads in fixed positions, can use mains-powered capstan motors, or speed-controlled DC motors. However it is achieved, the motor just needs to run at a constant speed. For different tape speeds, it’s common to see a stepped drive shaft, like on a multi-speed record player. Video recorders use a combination of fixed (audio, control track) and moving (video) heads. It’s vital for the video drum to spin at precisely the correct speed for the heads to scan the video tracks on the tape accurately. There is a reference for tracking: the control track, with its 25 ‘pips’ per second, indicating where the video tracks are located. So the head drum’s speed and position (phase) must be accurately forced (by a control system) to follow the control track signal. This control system is a servo. Phase servo The simplest VTR transports relied on a mains-powered motor running at a predictable speed to drive the tape capstan, and thus to transport the tape. Since the control track was part of the original recording, it would indicate the head drum’s desired position for correct playback. The main motor also drove the head drum mechanism, so it was naturally ‘in step’. The drum servo’s simple task was to adjust the position of the head drum relative to the tape, so that the heads scanned the slanted video tracks precisely. It isn’t enough to just have the correct speed; the position relative to the tracks needs to be correct, too. Fig.30 shows a simple phase servo. A pickup on the head drum feeds a trapezoidal waveform former, and the control track pulse is amplified to form a narrow sampling pulse. The sampling pulse operates an electronic sample-and-hold switch that delivers the trapezoid’s instantaneous amplitude at the time of sampling. A capacitor stores the instantaneous value as a DC voltage. The voltage across the capacitor will be low for early sampling or high for late sampling. This voltage is fed to the inverting input of a differential amplifier, with its non-inverting (reference) input voltage being adjustable via the ‘tracking’ control pot so that tapes from other VTRs can be played back successfully. Fig.30: an example of how a simple phase servo operates. 74 Silicon Chip Australia’s electronics magazine The output of this amplifier is proportional to the difference between the actual and desired phase, and this is then amplified to control the tape speed and thus bring the system into phase lock. Fig.31 shows a simplified mains-powered head drum mechanism. An eddy current brake, incorporating an aluminium disc mounted on the head drum’s driveshaft, applies a small amount of ‘drag’ against the drive belt’s force as the DC control current passes through the brake’s coil. This force is enough to create a minute amount of slippage between the belt and its drive wheel, and give an adjustable head drum position relative to the moving tape. The head drum speed was set just a little too fast, so that the drum servo would be able to adjust the drum phase to advance (less braking) or retard (more braking). Speed servos Mains-powered VTRs relied on the stable mains frequency to transport the tape at the correct speed, and the drum’s phase servo to deliver accurate tracking. Battery-powered VTRs also needed to transport the tape at the correct speed, and two methods were adopted. Akai’s VT-100 applied their clever DC brushless servomotor design first used in their X-IV and X-V portable audio recorders. It’s a three-phase motor driven by a high-power phase-shift oscillator. This design delivered excellent speed accuracy, but the drum servo could not use eddy-current braking for head positioning. Instead, the differential amplifier sent a control signal to the motor drive amplifier (MDA), and the MDA’s DC output powered the drum motor directly. So Akai’s circuit replaced the eddy current brake of Fig.30 with a DC drum motor. Sometimes the tape transport would also use a conventional DC motor. In this case, the transport motor would need a speed servo. A simple speed servo generates a voltage proportional to the difference between the motor’s actual speed and desired speed. If the actual speed is too low, this signals the Motor Drive Amplifier (MDA) to increase power to the motor. When the motor speed reaches siliconchip.com.au Build the world’s most popular D-I-Y computer! ALL-NEW COLOUR 2 Plastic Case Optional See SILICON CHIP July & August 2020 Fig.31: a simplified mains-powered head drum mechanism as used in a videotape recorder. its desired (setpoint) rate, this voltage moderates the MDA’s output to hold the motor at the setpoint speed. If the motor runs too fast, the voltage will swing in the opposite direction and signal the MDA to reduce power to the motor. As before, once the motor’s speed reaches to set point, the differential amplifier will moderate the MDA output to hold it at the set point. The basic speed servo (Fig.32) uses a simple speed pickup that delivers one pulse for each motor revolution. It could be a simple magnetic pickup, or it could use an LED with its light is transmitted to a phototransistor through a slit in a disk on the motor shaft. The tacho(meter) amplifier takes the incoming tacho pulses and converts them to a DC voltage proportional to the pulse frequency. The differential amplifier produces a voltage proportional to the difference between its two inputs. When they match, its output is such that the MDA maintains a constant speed. But if there’s a difference between the + and – inputs, the voltage will swing to signal to the MDA that it should change the motor speed and consequently, to bring the inputs back to balance. The actual setpoint speed is easily changed by adjusting the speed reference potentiometer. Combined speed/phase servos Phase servos are accurate, slow-responding systems. Speed servos respond quickly, but lack phase accuracy. High-performance designs combine a speed loop (for rapid startup) and a phase loop (for accurate positioning). Ultimately, mains-powered VCRs would take up these techniques, and would incorporate sophisticated direct-drive motors for capstan and head drum mechanisms. While more complicated, these advanced designs did not need speed-reducing belts or gears, were lighter and could be controlled more accurately, and could easily be slowed or reversed for slow-motion, reverse play and other useful ‘trick’ modes. SC 480MHz, 32-bit processor; 9MB of RAM; 2MB flash memory; 800 x 600 pixel colour display Don’t miss your opportunity to experience Australia’s own worldclass, world-famous single board computer that you build and program yourself, using the world’s easiest programming language – MMBASIC. Learn as you build! And it’s so easy to build because all the hard work is done for you: the heart of the Colour Maximite II, the Waveshare CPU Module (arrowed) is completely pre-assembled and soldered. YOU SIMPLY CAN’T GO WRONG! Short Form Kit includes: Waveshare CPU module pre-loaded with MMBasic the PCB – with solder mask and screen overlay front & rear panels to suit plastic case shown above and all other components required to build the Does not include plastic instrument case, Colour Maximite 2 CR12xx cell or USB power supply/cable 14000 $ All this for only Plus $10.00 p&p in Aust Fig.32: this basic speed servo uses a simple speed pickup to deliver one pulse for each motor revolution. siliconchip.com.au SILICON CHIP SUBSCRIBERS: $AVE 10%! Subscriber’s price just $126 plus p&p Order now (or more information) at www.siliconchip.com.au/shop/20/5508 Australia’s electronics magazine April 2021  75