Confused about DAT? Vic Lennard guides you gently through the ins and outs, the pros and cons and the whys and wherefores of Digital Audio Tape.
The word "DAT" has established itself as part of the musicians' vocabulary. But what will Digital Audio Tape do for you and what are its limitations - should you believe the hype?
THESE DAYS, YOU can hardly hope to read an interview with a musician without them expounding the virtues of either recording or mastering "digitally". Unless the interviewee is Big League, chances are that mastering on R-DAT will be mentioned along with a comment to the effect "Now I've used DAT, I'll never use a two-track tape recorder again". From such talk are legends built. And while R-DAT may achieve this in due course, it will only do so if potential purchasers - you - are aware of what it actually is.
TO UNDERSTAND DIGITAL recording, it is important to appreciate how magnetic tape works in general. Look at it as being comprised of millions of tiny magnets. When a signal is applied to the tape via the record head, the magnets are aligned in a direction dependent on the content of the signal. The higher the signal, the higher the number of magnets that are aligned, until the point where all of them are in the same direction. This is called the saturation level, as no more signal can be recorded onto tape and any situation below saturation will have some magnets directionalised and some not. If there is no incoming signal, the magnets will be randomly positioned - the same state achieved by the erasing process.
This description is true for both analogue and digital recording; the difference between the two lies in what the magnets are actually representing. In analogue recording, their direction precisely corresponds to the analogue signal, which made up of the waveforms appearing at the recording head. This means that any extraneous noise component from the tape medium will be recorded as part of the program and will be present on playback. In digital recording, incoming signals are "digitised", meaning that they are translated into binary as a series of 0's and 1's, and so there will only be two important positions for the magnets which can be defined. Decoding the signal from tape requires the two directions to be recognised. Hence the signal is only indirectly represented by the tape particles, and tape noise is of practically no importance.
In general terms, the range of frequencies (or bandwidth) which can be recorded onto magnetic tape is proportional to the head/tape speed. In analogue recording this is why an open reel machine running at 38cm/sec (I5ips) will give better fidelity than a cassette deck running at 4.75cm/sec (1 7/8ips). There are other criteria, such as track width and tape formulation, but the most important is noise reduction. Without this, the amount of noise on cassette would be deafening - literally. This leads to another point, that of Dynamic Range which is the ratio of the highest recordable signal to the lowest (usually the noise basement level) and is measured in decibels (dB). The higher this figure, the greater the fidelity of reproducing dynamics within a program. This is quite often quoted in terms of a Signal to Noise ratio, which is a similar measurement but can be misleading, as the human ear is more sensitive to high-frequency noise like hiss, than to lower-frequency components like rumble or mains hum. The use of Dolby B or C noise reduction will only hide the top-end noise, making the signal to noise ratio dependent on the frequency at which it was measured. A quality cassette deck can return figures of 75dB measured with Dolby C at 1kHz and above but you can probably knock 20dB or so off at lower frequencies which is why the bottom end usually lacks clear definition.
Digital recording, as they say, is something completely different. The incoming audio signal is "sampled" a number of times each second. This is called the sampling frequency; a sampling frequency of 48kHz meaning the signal is sampled 48,000 times per second. It is then converted into a number depending on the number of "bits" of resolution, each of which will be either a 0 or a 1. Sixteen-bit resolution will allow the height (or amplitude) of the waveform to take one of 65,536 values. This process is called "digitisation" and because of the order of the numbers involved, means that a frequency response of about 4.7MHz is needed. The only way to achieve this is to have a head/tape speed of 313cm/sec which is, at present, impossible to achieve with the stationary head systems on reel-to-reel and cassette machines. Instead R-DAT (or Rotary Digital Audio Tape) machines employ a system similar to that of a video recorder - there is a rotating drum on which there are two heads each writing or reading tracks consecutively as the tape travels past. The head revolves at 2000rpm while the actual tape speed is only 0.8cm/sec. This gives a very low dynamic range in analogue terms but as only a change in state (from 0 to 1 or vice versa) has to be recognised, a dynamic range of 10dB is adequate. The actual dynamic range of the program will be constant across the frequency spectrum at between 90 and 96dB for 16-bit digitisation (depending on the quality of the circuitry).
As a final comparison, analogue recordings are laid as parallel tracks along the length of the tape. Digital recordings take the form of oblique stripes across the tape about a quarter the width of a human hair.
So what we appear to have in R-DAT is a medium capable of reproducing wide dynamics with no tape noise. Inherent in the coding is a degree of error correction which is recorded as extra data and, at playback, should be able to detect whether what is on tape is the same as what was recorded and make any necessary corrections.
LIKE VIDEO RECORDERS, R-DAT recorders record in "frames", each of which will have the relevant encoded signal, but also have what are termed "User bits" for inserting extra information which can be recovered on playback. R-DAT recorders can generally mark the start of a song or section by writing a Start ID to the tape. This can be done after the main recording has taken place. This mark can be erased without affecting anything else on tape, and IDs can be renumbered whenever necessary. As an example of its use, it is possible that too long a gap will be left between songs on a tape and that it would be useful to automatically fast forward the tape to the next Start ID after the end of a song. This is accomplished by writing a Skip ID. On reaching this, the tape is fast forwarded and stopped at the next Start ID and then starts to play. Again, Skip IDs can be erased, and most machines have a switch to select whether skip is in operation or not.
Other information can also be stored in this area. Casio's DA1 and DA2 machines use it for writing up to 28 characters for song name, performer, date or anything else pertaining to the recording. This acts in a similar manner to the Start ID in that it can be used to uniquely define a point on tape and can then be used as a reference to find that place, by entering the first five characters and then searching the tape for this sub-code.
One of the major disadvantages which has been put forward against the R-DAT format is that it cannot record a time code on a centre track, in the way that professional open reel machines can. This would severely limit the ability of R-DAT to be taken seriously in terms of postproduction and audio-visual work. This is another aspect for which the User area on tape can be used - as has been shown by Fostex. Their D20 reads SMPTE timecode, encodes it to tape at the correct position with the audio and then decodes it on playback. The advantages of this are that varispeeding will not affect the reading of the code from tape as it is only indirectly dealing with the actual signal.
THERE ARE TWO different types of signal input and output from an R-DAT machine; analogue and digital. Analogue inputs and outputs can be connected straight from/to a mixing desk or cassette recorder. As the signal has to be passed through an analogue-to-digital converter (ADC) to encode the audio signal and then through a digital-to-analogue converter (DAC) to decode, there will be a certain degree of loss of fidelity but very little in comparison to applying the same process to two cassette decks. Digital connections allow an R-DAT tape to be precisely copied from one R-DAT machine to another, as the transference is purely in terms of numbers. There are primarily two different standards found on R-DAT recorders; Sony Phillips Digital Interface (SPDif) and AES/EBU or SDIF2. The former is intended for domestic machines and usually consists of two phono sockets for input/output with no separate facility for synchronisation, while the latter two are the professional standards, utilising three BNC connectors or one XLR connector and can send an independent clock signal.
The sampling frequency at which R-DAT records has caused an enormous amount of controversy. Because of the ability of R-DAT to faithfully reproduce audio programs, and as the 16-bit format is the same as CD, it is possible to use a CD player with digital output to send the information on a CD to an R-DAT recorder through its digital input, and to then endlessly pirate perfect copies of the CD. To prevent this, a domestic sampling rate of 48k Hz was agreed upon as opposed to the standard CD rate of 44.1kHz, as conversion from one rate to the other is not that simple an affair. However, most early R-DAT machines were categorised as being aimed at the professional market and so had both frequencies available, and those which were semi-professional (such as the Sony DTC1000ES) could be modified to run at 44.1kHz. A variety of anti-copy measures were investigated, but none have inflicted themselves on the current generation of R-DAT machines. Later machines have the domestic rate burned into their hardware, so making any modification impossible. The only other sampling rate is 32kHz which is used solely for the playback of pre-recorded tapes and is seldom seen.
Emphasis is a gain curve giving 0dB at 3.1kHz rising to 10dB at 10kHz and above and acts in a similar manner to a presence boost. R-DAT machines can sense when a signal has been recorded with emphasis on - another use of codes - and will adjust their playback accordingly.
It is essential that no component of the sampling clock at 48 or 44.1kHz is present in the input signal. To prevent this, a lowpass filter (called an anti-aliasing filter) is used to roll off all frequencies above 20kHz at a steep rate. A relatively new idea is to move these frequencies up a few octaves making filtering easier - this often uses a method called Oversampling, which you may see mentioned in R-DAT literature. A similar technique is used in the output stage where steep filtering could upset the accurate phasing of program material.
Finally, some machines use two DAC's per channel in a process called Interleaving, where one is used for each half of a waveform and so operate 180 degrees out of phase from one another. This should give an extra one bit of resolution. Alternatively, some cheaper models share one DAC between both channels, which may lead to a slight loss in fidelity.
WHEN YOU CONSIDER how long it takes to to get good results recording with analogue, it should come as no surprise that digital recording also needs treating with respect.
Due to the nature of analogue, certain things are taken for granted. The obvious one is what happens when meters are pushed "into the red". While no two analogue recorders will respond in exactly the same way, it's rare for distortion to immediately become audibly apparent. A degree of tape compression will set in first, followed by the top end becoming harsh at the edges. The harshness will continue to get worse and the point at which it becomes categorised as distortion will vary from person to person.
In fact, the initial effect of signal overload can be used creatively. "Soft" compression often manifests itself as "warmth" - although you should never lose sight of the fact that it's still distortion. Where digital recording is concerned, distortion isn't as kind. There's only one way to describe the sound of digital distortion - painful. There's no forgiving "warmth", just chronic distortion. Another point to note when using DAT is that there's no facility for monitoring off tape, which means that you appreciate your mistakes only too late.
So much for the changes in attitude DAT requires. The first time you master digitally will herald the demise of half of the equipment in your studio. That synth with the noisy outputs (Yamaha TX7, Roland MKS70?) which always tended to get hidden in the general confusion of the mix will now stand out like a sore thumb. The hiss on the release of the bass drum sample which you always intended to clean up but never got around to will come back to haunt you every first and third beat throughout a song. That crackle from the noisy slider on the mixing desk will now sound like a crisp packet at a silent movie. And that untidy drop in? Forget it.
What is hopefully coming across here is that DAT tells the truth. Dropping in and out of record on a bass drum or snare drum hit will hide any clunks due to the multitrack transport, and using a noise gate, especially one with a downward expander, will also keep recorded noise to a minimum. Single-ended noise reduction units such as the Symetrix 511 or the dbx SNR1 (reviewed MT, June '89) are invaluable both for recording onto a multitrack and across the mixer outputs on mixdown, where their program-conscious nature will ensure that background noise is minimal.
Even after taking every precaution possible, there's one area which is still likely to be a problem. Because R-DAT recorders are at the kind of price which interests studios using predominantly semi-professional equipment, noise will make its presence felt. For example, if a noise gate is used for auto-fading, the noise attributable to its circuitry may be still be in the region of 30dB worse than that which can be recorded by DAT, and will set itself the reference point for the noise basement. A system will only be as good as the last link in the chain if that unit is being used to reduce the overall noise of the system.
SO HOW USABLE is the R-DAT format? It's certainly an unfriendly medium when it comes to editing. There is no equivalent to the cutting and splicing of analogue tape, and although Hybrid Arts have just released ADAP 2 which should be capable of visually editing DAT via digital connectors and an Atari computer, editors are likely to be thin on the ground for a while. Bearing in mind that DAT was never intended for the professional market, the longevity of the tape medium has also been brought into question.
However, the quality available from even the cheapest DAT recorder is beyond that of any semi-pro analogue recorder be it open reel or cassette. Dynamic range, distortion, you name it - DAT will outperform analogue. Do I own one? No. I was keeping an open mind on the subject but having experimented with three different models over the past month in the course of this review, I will certainly be buying one.
Thanks to HHB Communications Ltd and Darrell Lockhart of Pepper Records for loan of equipment.
Feature by Vic Lennard
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!