The mechanical requirements of a videotape machine are very severe, and it is a triumph of modern manufacturing that they could be mass-produced at all. We will now look at some of the mechanical engineering problems that had to be overcome to build a videotape machine.

The heart of a videotape machine is the video head assembly. It has undergone a lot of changes over the years. In the early days, the quadruplex headwheel was turned at 14,400 RPM by a polyphase AC motor. It used ball bearings, and slip rings to get the signals to and from the rotating heads. Precision machined parts were used to guide the tape and form it into the correct 'U' shape so it could be scanned by the heads. Vaccuum was applied to the inside of the guide to pull the tape into alignment. The guide could be moved in and out, as well as up and down, to correct timing errors. On many machines, the in-out motion was under servo control. The video heads themselves were originally made of Alfesil (Also known as sendust.). Newer heads, as well as all modern heads, are made of ferrite.

The first problem encountered with the quadruplex video head was a particular form of video distortion known as the waterfall effect. It looked like a waterfall in the background of the picture. The problem was traced to tiny abnormalities in the ball bearings used in the head. No matter how carefully the bearings were manufactured, they would eventually wear in such a way that they would develop a vibration. This vibration was coupled to the video headwheel, and caused tiny timing variations. These vibrations would occur at the rate the ball cage turned in the bearing, so it caused the noise pattern to run down the picture. The solution was to eliminate the ball bearings, and replace them with air suspension bearings. Now, quad videotape machines required a fairly hefty compressed air source to operate the air bearings. The presence of compressed air in the machine eventually led to a wide variety of pneumatic systems in other parts of the machine.

The second major problem encountered with the quadruplex head was brush noise in the picture from the slip rings. The solution to this problem was most ingenious: the development of rotary transformers. Concentric coils of wire were mounted in grooves on a ferrite disc. One of these would be fixed, and the other would be on the rotating headwheel. They were spaced as close as possible so the magnetic fields would couple. Rotary transformers can be used to bring any kind of high frequency AC signal to the heads, and have left little to improve upon since their inception.

Quadrature error on a quad head was a timing error produced by the fact that the heads were not exactly 90 degrees apart. It produced a stubborn timing error that could only be corrected by physically adjusting the position of the delicate heads. Today, the same error exists in helical scan heads, and is called Dihedral. Modern professional machines can have five or more pairs of heads on the headwheel. Manufacturing techniques have improved to the point where dihedral correction isn't even available as a field adjustment.

Running very weak signals from a video head to a preamp that may be a couple feet away caused a lot of performance loss in early VTR's. With miniature tubes, and later transistors, manufaturers were able to build the preamp inside the head assembly, just after the rotary transformers. Ampex marketed one version of their quad head assembly, the Mark 10, with three different optional preamp modules: None at all, Tube, and Solid state. Any head could be converted to another type just by changing preamp modules.

Thus, an older machine could be retrofitted with a better head by simply replacing the assembly with a new one containing the correct module. The last improvement, only seen in helical scan machines, was to move the preamp right on to the headwheel itself. This provided the best signal-to-noise ratio. However, slip rings are again required to bring power and control signals onto the rotating headwheel. The actual video signals go through rotary transformers. Because of the expense of slip rings and miniature high-G relays (The centrifugal force on the inside of a 1" type C video head is 1,000 g's!) for record/play switching, this is rarely seen in consumer machines.

In the early sixties, helical scan machines began to become practical as the engineering problems they presented were solved. They did not require air bearings, as their rotational speed was much lower (1800 to 3600 rpm vs 14,400 rpm.). But, they had problems of their own.

In the quadruplex format, the length of tape that needed to be precision guided was about 3/4 of an inch. Furthermore, vacuum served to hold the thick tape in proper alignment. In helical machines, it was required to hold a piece of tape as long as 16 inches in an angle wrap around a cylinder. This tape had to be in precisely the correct position for it's entire length, and lie flat on the surface of the cylinder. The tendency towards thinner tape made this even more difficult. The solution was very complex. Many precision guides were used to put the tape at the right angle to follow the head drum.

It is not unusual to encounter guides that were mounted at strange angles, and have tapers or steps on them as well. A precision guide, called a Rabbet Guide was built into the lower part of the head drum cylinder. The tape lightly rests on this guide as it wraps around the head drum. Finally, the head drum assembly itself is usually mounted at an odd angle to facilitate tape wrapping around it. Open reel helical machines often have the reels at different heights to eliminate a lot of the oddly angled guides. Obviously, you can't easily do this with a cassette!

Another development that came with newer helical scan machines is azimuth recording. As the need for longer playing times pushed tape packing densities higher and higher, the narrow guard band between video tracks shrank until the tracks practically overlap. All of the newer formats use two-head, 180 degree wrap helical scan. This means that every other track is recorded by every other head. Since the adjacent track was recorded by another head, introducing a deliberate azimuth error between the two heads means that the track recorded by head 'A' cannot be read by head 'B'. The azimuth errors designed into modern heads varies between 7 and 14 degrees, meaning a 14 to 28 degree azimuth error between tracks. As a result, there is very little crosstalk between tracks, even though they practically overlap. This system is called, oddly enough, azimuth recording.

All videotape stretches a little while in use. The specifications for each videotape format specify how much tension needs to be on the tape for proper take-up without damage. They also specify how much holdback tension needs to be applied from the supply side to maintain proper tension around the head drum. (Tape tension wasn't nearly as critical for quadruplex.) All but the top-of-the-line professional studio decks tend to use a somewhat primitive holdback tension brake. A felt band is wrapped loosely around the supply reel table, and is connected to an arm which rests against the tape. If the tape tension drops, the arm moves inward against the tape. This inward motion tightens the felt band through a linkage, and restores proper tension. Conversely, if the tape tension gets too high, the tape pushes the arm out, and the band is loosened through the linkage. As crude as it sounds, this system is quite reliable. Top-of-the-line machine detect tape tension through sensors, and the reel servo makes the correction.

Cassette-based machines need some way of pulling the tape out of the cassette, and wrapping the tape around the head drum without operator intervention. Special quadruplex machines, called spot players, were built to specifically play back commercials during commercial breaks. These machines used small cassettes that could contain up to five minutes of tape. Puffs of air were frequently used to pull the tape into it's proper position. The straight path of the tape, along with a retractable vacuum guide on the video head, made threading straightforward in these machines.

Helical scan cassette machines posed a major problem. The tape had to be wrapped around a cylinder, and past a number of fixed guides and heads. Sony provided the first solution, by using a ring that pulled a loop of tape around the video heads and the rest of the guides. The system was simple and worked well. Sony also came up with another method: Two tape guides were mounted on moving tracks, and pulled the tape out of the cassette, and around the video head. They couldn't get this to work well, and abandoned the idea in 1966. JVC quietly bought the patent to that threading system and used it for the first time in a machine called video home system, or VHS! The ring threading system was simple, and Sony owned the patent on it. The M-load system, as the moving guides system came to be known, was hard to manufacture, and required excessive tape tension. But, they didn't have to buy rights to use it from Sony; JVC already owned the rights. Interestingly enough, many professional machines built by Sony today use the M-load system, while the corresponding machines built by JVC and Panasonic use the ring load system!

There are many other mechanical systems in a typical videotape machine, but they don't differ from an audio machine or they didn't pose a severe design challenge. They will not be covered here.