11 Recording

Part 1 – Recording devices and systems

There are two broad categories of audio recorder:

1.   Analogue, in which the degree of magnetization of a suitable tape varies with the amplitude of the audio signal.

2.   Digital, where the original signal is converted into a particular code and it is the latter which is recorded, possibly magnetically.

The professional world is increasingly using digital recorders, but it is likely that for some time to come analogue machines will be useful tools for many non-professionals, so we will deal with the main aspects of these devices.

Analogue recording

Analogue machines themselves come into two categories. There is the familiar cassette type and there is what is termed the open reel or reel-to-reel type. Because of the width of the tape in the latter, it is often referred to as quarter-inch tape – even in a metric world! In this type of machine the tape itself is wound on spools which can vary in diameter from about 27 cm down to half that. The spools, one with the tape on it, the feed spool, and the other, the take-up spool, are locked on to suitable hubs and the tape is manually threaded through the necessary mechanism. The tape speed on such machines is usually switchable, the standards being 76 cm/s (30 i.p.s. = 30 inches per second), 38 cm/s (popularly referred to as 15 i.p.s.), 19 cm/s (7.5 i.p.s.) and 9.5 cm/s (3.75 i.p.s.). Although reel-to-reel tape machines are capable of high-quality recording, and editing of the tape is easy, they are being displaced by digital machines.

Because of their still popular usage, we shall concentrate here on cassette recorders, as the likelihood is that they will be with us for a considerable time yet. First, though, we will look briefly at the fundamental processes in magnetic recording.

The fundamentals of magnetic recording

It is well known that some substances can be magnetized fairly easily. One such substance is a particular type of iron oxide – in fact a form of rust! This is very finely ground and then made to adhere to a thin plastic strip, thus forming the tape. In any tape recorder the tape is made to move past a series of heads, these being basically electromagnets with coils of fine wire. One such head, the record head, is fed with the electrical audio signal to be recorded. The tape coating – the oxide – then becomes magnetized in such a way that the strength of the magnetization is proportional to the voltage of the original audio signal.

A second head is placed close to the record head and is known as the replay head. The tape with its now variably magnetized surface passes this head and small electrical voltages are induced in its coil. When these are suitably amplified we have, all being well, a signal which is a replica of the original one fed into the record head. In Figure 11.1, the letters N and S represent the north and south poles of the tiny magnets formed on the tape during recording. The left diagram illustrates a low frequency signal, where the rate of alternation is low; a higher frequency signal is shown on the right.

So that tape can be reused it is necessary to have some means of ‘wiping’ it – demagnetizing it – when necessary. This is achieved by a third head called the erase head. This is inoperative during replay, but when the machine is switched to record, the erase head, which the tape meets first of all, is fed with an alternating current of large amplitude and high frequency – of the order of 100 kHz – and this has the effect of demagnetizing the tape before it reaches the record head.

From what we have said, there seems to be the need for three heads: erase, record and replay. However, many cassette machines manage with only two – the record and replay heads being combined. They are, after all, very similar in construction. In theory, it is possible to get better results if the two heads are slightly different in their design, and this is the case in professional ‘quarter-inch’ machines. There is, though, another reason, apart from quality, for having three heads and that is that the replay head can be used to monitor what has just been recorded. The more expensive cassette machines often have three heads. In such machines there will be a switch or button which in a typical case selects ‘source’ or ‘tape’. In replay this control may be inoperative, but when set to record and switched to ‘source’ the signal which is being fed to the record head is also sent to meters, loudspeaker, etc.

Figure 11.1 Tape passing a replay head

In the ‘tape’ position it is the replay head which is connected to these outputs. By switching between the two it is possible to detect a delay, generally quite a small fraction of a second, this being due to the time it takes the tape to travel from record to replay head. Consequently, when recording, the condition of the recorded signal can be continuously checked for faults in the tape (very rare with good quality tapes these days), for accidental overloads causing distortion (then probably too late to do much about, but further overloads can be avoided by reducing the recording level) and for satisfactory recording in general. For example, dirty heads can result in loss of the high frequencies, a point we shall return to.

It may be worth noting that the magnetic material used on tapes has to have rather special qualities. To begin with, this material must be capable of being magnetized to a high degree. It must also be able to hold its magnetization permanently yet, at the same time, deliberate demagnetization (erasing) must not be difficult.

Cassette quality

When all things are considered the quality of recording achievable with cassettes is remarkably good. To begin with, the slow speed (4.75 cm/s – just under two inches a second) means that at high frequencies the individual ‘magnets’ are very short and extremely close together, but nevertheless frequencies approaching 20 kHz are recordable.

The main defect with cassettes is poor signal-to-noise ratio. The individual tracks are extremely narrow, there being four (one each for stereo left and stereo right, times two because the cassette can be turned over) in a width of just over 3 mm. And since there has to be a space between tracks then each one is less than three-quarters of a millimetre wide! Now the magnetic particles, although ground very finely, are not infinitely small, so that each one makes its own tiny and individual contribution to the recording. If the tape were only one particle wide, the result to the ear would be a series of clicks whose overall effect might perhaps have some semblance to the desired recording. The wider the track, the less noticeable these clicks become until they merge to create a faint hiss, known as ‘tape hiss’. In the early days of cassettes, tape hiss was a major obstacle to their serious use, but improved materials and the adoption of good noise reduction systems (see later) means that nowadays modern cassettes on good equipment can provide very respectable recordings.

Figure 11.2 (a) Quarter-inch tape (b) Cassette tracks

Note that if the diagrams were to scale the cassette tape would be seen to be half the width of the quarter-inch tape.

The hiss, of course, sets the lower limit of the signal-to-noise ratio. The upper limit is determined by saturation of the magnetic particles. As the magnetizing force, a consequence of the voltage in the record head, increases there comes a stage where the magnetic particles are fully magnetized. Any further increases in the signal voltage are not matched by further magnetization. This condition is termed ‘saturation’ and on playback it appears as distortion, maybe very unpleasant distortion at that.

Noise reduction

All but the cheapest cassette machines these days, and most car radio/cassette units, have the ability to make use of ‘Dolby’® circuitry. It is important to say that ‘Dolby’ is a registered trade name, but like many others such as ‘Biro’, ‘Thermos’ and ‘Hoover’ has almost become a generic term. There are three commonly used versions, ‘B’, ‘C’ and ‘S’. There are others – the A system is confined to professional analogue equipment, for example. The primary aim of all the Dolby versions is to reduce the effects of tape hiss and, in the case of the cassette systems, to put it simply, they do this by increasing the high frequency content of a recording. Such recordings are often described as being ‘Dolbyed’ or ‘Dolby-encoded’, and pre-recorded ones have the Dolby ‘Double D’ trade mark – this can be found easily on virtually all pre-recorded cassettes.

In the replay machine a corresponding amount of high frequency reduction is used. Since tape hiss is predominantly high frequency, the result is that the second process reduces the effect of the hiss while bringing the recorded signal back to normal. In the B system there is about 10 dB of hiss reduction; with C, a somewhat more complicated version, around 20 dB reduction is achievable – even more in the S type.

To put this into a practical context, recordings taken ‘off-air’, i.e. recording a reasonably strong f.m. radio signal, with good quality tape and Dolby C, are barely detectably lower in quality than the incoming radio signal.

Head and tape cleanliness

A glance at the diagrams in Figure 11.1 will help to explain that for good recording and reproduction of the higher frequencies it is essential that the tape is as close to the heads as possible. It should be clear that the closeness of the poles in the high frequency case means that their effects will, as it were, cancel each other out a short distance away.

Normally, the tape is held tightly against the heads by a tension provided by the tape drive system. However, with time and use, it is easy for a deposit of the oxide material from the tape to build up on the heads, and this has the effect of keeping the tape at a distance from the heads. A loss of the high frequencies may then be apparent. The answer is to keep the heads clean. Many proprietary devices exist for cleaning the heads. A simple and very effective way, though, is to rub the heads gently with a cotton bud which has been dipped in isopropyl alcohol – generally available at pharmacists. (Some seem reluctant to sell this, but it can be bought as a proprietary head-cleaning material from hi-fi shops.) It is instructive to look at the cotton bud afterwards. The word ‘gently’, above, has been put into italics to emphasize the nature of the treatment. Rough handling of the heads can cause misalignment, which may be as bad as, or worse than, dirt!

This is assuming that it is easy to get at the heads. The better machines are generally designed so that there is access to the heads, but if this is not the case then it may be necessary to use a head-cleaning device which typically has the dimensions of a cassette and is slotted into the machine which is then set to play. After half a minute or so, the cleaning process will be finished. (Such devices are essential for car cassette players.)

It is not only the heads which must be clean, but also all the parts of the drive system and the actual tape. Normally, tapes do not get dirty themselves but careless handling can leave a thin film of grease from the skin on the exposed surface.

While dealing with the cleanliness of the heads it may be as well to include a few words about demagnetizing. If the metal of the heads acquires some permanent magnetism there is likely to be a change in their electric characteristics, and this can show as an increase in the hiss level. Worse still, it is possible for a tape to be permanently affected after being used on a machine with magnetized heads. Demagnetizers are readily available in good hi-fi shops and their use is easy. The instruction of both the makers of the cassette machine and the demagnetizer should be followed.

Digital recording

Having said above that digital audio cannot be handled by conventional recording machines, we will look briefly at the problem of how the enormous bit rate (number of bits per second) can be put on to tape. Let us, for the sake of simplicity, take the bit rate to be the same as the maximum frequency in Hz. As we have said, this is not quite correct, but the inaccuracy is not too serious. Let us say, then, that a conventional tape machine can handle frequencies up to 20 kHz and with digital audio we need to record frequencies up to 1.4 MHz – at least. We could do this if the tape were speeded up in the same proportion – 20 000:1 400 000, or around 70 times.

Quite apart from anything else, a reel of tape which would run for 30 minutes with conventional recording would now run for less than 30 seconds. That is faster than the spooling motors would go!

There are two tape-based approaches:

1.   The important thing, when one thinks about it, is not the actual tape speed but the relative speed between tape and head. Instead of moving the tape rapidly, why not move the head rapidly? This principle is used in many digital tape machines and it is also, incidentally, used in video recorders, both of the professional and domestic type, where there is a similar problem of trying to record frequencies of the order of megahertz. In such machines the head(s) are mounted in a drum which rotates rapidly. The axis of the drum is tilted slightly so that when the tape passes in front of it the recorded tracks are at a slant along the tape (see Figure 11.3).

A compact and reasonably priced digital audio recorder with this system is known as either R-DAT (rotary head digital audio tape), or more usually simply as DAT. The original intention of the companies that developed DAT seems to have been that it would replace CDs in the domestic market, but this has not happened. DAT is, however, used extensively in the professional world.

Figure 11.3 The principle of ‘slant-head’ recording (in practice the tape is partially wrapped round the head drum)

2.   Stationary head recorders. These look very much like conventional ‘quarter-inch’ machines. There are several models available to the professional, but they are all very expensive. Here the tape moves relatively slowly but it is of a special type which fits snugly round the heads and there may be several tracks carrying different parts of the data. Paradoxically, perhaps, the most expensive digital recorders allow the cheapest form of editing, in that the tape can be cut and spliced just like analogue tape.

DAT

1.   The cassettes used are actually smaller than the familiar compact cassette, being about the same thickness but about 3 cm less in width and 1 cm less in depth.

2.   The tape speed is about one sixth of that in a conventional compact cassette (8 mm/s against 48 mm/s). However, the playing time is 2 hours, with the possible option of 3 hours at slightly reduced quality.

3.   The head drum inside the machine is 30 mm in diameter (just over an inch) and rotates at 2000 r.p.m.

4.   A typical portable DAT machine is two or three times the thickness of this book, but its width and depth are comparable. It will run off batteries.

MiniDisc®

As we have said, it is possible, using data compression methods, to record only a small proportion – about 20% – of the sound signals that come from a microphone, and MiniDiscs do this with ATRAC, mentioned earlier. The system uses discs which at first glance look rather like computer ‘floppy’ discs. The rotatable disc is inside a shell. There are actually two types of MiniDisc. There are the pre-recorded ones, and these are really small CDs and are read in an exactly similar way. They cannot be recorded on!

Recordable discs are magnetic and the audio data are put on in a very clever way. The magnetic material needs to be heated to a fairly high temperature – about 180°C – to have its magnetic state changed and this is done with the laser working at high power but only on a very small part of the track. The heated part is exposed to the magnetic effects from a record head on the other side of the disc. Replay is done with the laser working at a much lower power – otherwise it might erase a wanted recording!

Now comes another very clever trick. In replay the laser beam shines on the recorded track and some of the beam is reflected, as with a CD. However, the polarization (the angle of vibration of the light waves) is changed by the amount of magnetization. This is detected using something in principle not unlike the lens of polaroid sunglasses and a suitable sensor converts this into an electrical signal. By using data reduction, which we've already mentioned, a MiniDisc can hold about 75 minutes of stereo (nominally 74 minutes) or twice that amount in mono.

A most important question about MiniDiscs is to what extent the quality is degraded, because such a large part of the incoming audio signal is discarded. The only answer I can give is that I once took a good quality CD, with a wide variety of music on its tracks. I copied a minute or two of each track on to a MiniDisc and then played each track on the CD in as near as possible sync with the corresponding MiniDisc track. (A–B comparisons, you'll note!) I switched from one player to the other, listening on good loudspeakers, and neither I nor other listeners could tell any difference! I knew which I was listening to, because I was doing the switching, but the other listeners didn't.

It is, of course, just possible that some people listening on extremely high-quality loudspeakers might just detect a difference, but I'm prepared to say that the reduction in quality on MiniDiscs is so small as to be negligible.

Solid state recording

A final point in this chapter. Since so much has been made in the sections above about storing digital audio in memory chips, it may be wondered why these are not used for all recording. The answer is quite simple. There is a vast amount of ‘data’ in even a minute's worth of sound! Let us look at it like this: we have already seen that 1 second of stereo contains about 1.4 megabits, and this is without taking into account bits for timing, error detection and correction, and so on. Two megabits per second would be a reasonable bare minimum. For a minute's duration of high-quality sound, this means about 120 megabits. Now the unit of memory storage in computers is the byte, usually equivalent to 8 bits, so a minute of good stereo needs around 14 or 15 megabytes.

Only a few years ago this would have been a total impossibility, and with conventional chips it's still not all that easy. However, two developments have taken place. First various ‘memory cards’ with many megabytes of storage have appeared. (Digital cameras use ‘flash cards’ with many megabytes of storage.) Secondly, there have been advances in ‘data compression’ (see below). This has led to the appearance of MP3 recorders, which can store long periods of quite acceptable music with no moving parts.

In the professional world there have been some ‘solid state’ recording machines but they, and the ‘blanks’, are currently very expensive. It has to be said, though, that high-quality recording and reproduction with no moving parts is a very attractive idea. A large part of the cost of a high-quality tape recorder is in the rigid but preferably not too heavy deck to avoid mechanical distortion of the tape path. There is also the necessity of having very accurate speed control of the tape without any risk of stretching it when spooling and braking. If all that can be avoided…!

There is, however, what may be the start of affordable solid state recording and that is MP3.

MP3

This is a now well-known recording device. By using heavy data compression, the machines can record music for considerable durations on to a solid state storage card. The quality? Well, it's difficult to say because the system is designed for headphone listening and this makes it difficult to make A–B comparisons with anything else.

Questions

1.   In a three-head tape recorder, what is the order in which the tape meets the heads?

2.   At what speed does the tape travel in a cassette recorder?

3.   What is the likely effect on the reproduced sound signal of dirt on the heads?