Digital Recording (Part 1)
A New Landmark!
Article from Electronics & Music Maker, October 1982
The first of a special two-part feature highlighting the dawn of a new era in audio recording.
In E&MM this month we have the first of two articles dealing with the current developments in Digital Audio recording.
We take a look at the reasons for going digital, Sony's 24 track digital recorder for studios, and equally exciting, Sony's PCM F1 processor which mates with domestic video recorders (VHS or Betamax) to produce what is predictably a landmark in quality home or small studio recording.
To enable readers to appreciate the information in the subsequent pages here are some digital Facts.
Analogue Audio — Conventional electrical signals which are continuous and take shape in direct analogy to the sound waves they represent.
PCM — Pulse Code Modulation, the technique used to encode digital information from an analogue signal.
Bit — A computer term meaning Binary Digit or single element which makes up a binary code.
The concept of PCM, now simply termed Digital, was proposed in the 1930s by a British Engineer, Alec Reeves. PCM has been used extensively in the telephone business (as much by British Telecom as anyone) and by the BBC for feeding FM radio and TV sound from studios to transmitters. However, the system had not been commercially viable until the development of large scale integrated circuits brought the cost down to a practical level. Once in Digital form there are enormous advantages in the way quality can be maintained in the transmission or recording of such signals.
No matter how complex a waveform is in terms of its frequency content, all we need to know at any one moment, is its Amplitude. See Fig. 2. If we repeatedly measure the Amplitude at very, very short time intervals we have all the information needed to describe the waveform and thus transmit or store it.
Using computer techniques, the easiest way of describing the amplitude is to use a Binary Code. So a series of pulse/no pulse is created to describe the amplitude at each sample taken. The number of pulse (or Bits), used for each sample will determine how many different amplitudes can be accurately described. Eight bits are used in telephony (28 = 256 levels). Thirteen bits are used currently by the BBC for VHF transmitter feeds (213 = 8129 levels) whereas sixteen bits is the accepted norm now for the highest quality audio recording (216 = 65,536 levels). So in the latter case, each waveform sample can take over 65 thousand different values which obviously allows very accurate coding, and therefore high reproduction quality.
On playback, the codes are converted to the instantaneous amplitudes they represent and a 'ragged' reconstruction takes place. This produces irregularities which contain high frequency harmonics that have to be removed by sharp cut-off low pass filters to result in a faithful reconstruction of the signal.
The breaking down of a waveform into a stream of pulses in a binary code has one particular attribute — amazing resilience to any alteration that a transmission or storage medium could produce. It's the logic level that matters and not the shape of the signal or how much noise has been added. As long as the pulses and their 'order' can be recognised then near perfect reconstruction is possible. This is the thing to be grasped about Digital — a present day digital recording as near to perfection as possible. In fact improvements in microphones, amplifiers and loudspeakers will be the way forward.
1. The fewer bits used in the code for each sample, the greater the noise generated — called quantization noise. With 16 bits, signal to noise ratios of over 90dB can be achieved!
2. Sampling of each channel must be carried out at least, at twice the highest audio frequency needed. Telephone systems are sampled at 8 thousand times a second. The current BBC links are at 38 thousand times a second. In tape recording there are a number of systems using 44 thousand times a second. There seems to be a move to standardise on 50 thousand times a second though.
3. Being in 'computer language' form there are emerging some amazing error correction processes which are dramatically accounting for the latest digital equipments overcoming the alleged failings of earlier types.
Just before the annual APRS Exhibition in London, earlier this year, E&MM attended the Press launch of Sony's Studio Digital Multitrack Recorder — the PCM 3324. This was at the Advision Studios in London — a conventional Studer multitrack studio, where for instance, Kate Bush records.
Sony have joined MCI and Studer in adopting what is hoped will become world studio digital format standards. To this end the PCM 3324 has 28 tracks on the tape, with 24 for the 'music', two for conventional analogue recording of editing memos or talkback etc, and two for time code and digital control. Using time code synchronization for example would allow additional 3324's to be synchronized giving 48 channels or more — like conventional analogue machines linked, but without losing two tracks in the process.
However, as was explained by Keith Smith, Sony's Professional Audio products manager, the advantages of digital in multitracking perhaps makes doubling up unnecessary. This, with the absence of tape hiss, very low cross talk between tracks and the perfect multi-generation copying will allow re-thinking of studio practices.
With conventional analogue recorders, the engineer has to be careful where he lays down certain instruments as crosstalk breakthrough is worst between adjacent tracks. If a 'pre-mix' bouncedown is done to clear some tracks there is the little 'nagging' loss of quality to consider. With digital multitrack there is no inhibition on the choice of track as any cross talk present is not due to the tape but will be in the analogue part of the machine. Multi-bouncedowns produce no change in quality or in added noise.
Top quality analogue 24 track recorders use 2" tape — the new Sony uses ½" tape. Capital and storage space costs are therefore much reduced. Drop-ins (or punch-ins) with conventional multitrack machines need care, as a 'clipped' signal already on tape will show up on playback. The new system does a digital cross-fade at the punch in point avoiding any sound discontinuity. The punch-in can also be rehearsed. This cross-fade feature also allows conventional spliced editing, however, it is envisaged that editing will be done using equipment such as the DAE1100 Electronic Digital Audio Editor.
With the advent of the Compact Disc there will undoubtedly be a need to originate multitrack material in digital form so that all the recording stages to the listener are digital.
Good as multitrack analogue is with Dolby A and careful use, we were treated at the press reception to a fascinating comparison of bouncing a signal 24 times down on a conventional analogue system at 30 ips, and on the PCM 3324 (also incidentally at the same tape speed to get the frequency response necessary for the 16 bit/50.4 kHz sampling used). The analogue showed problems with hiss by the eighth generation and later on down the sequence the bass 'woodle' problem of 30 ips operation. By the 24th generation the analogue was decidedly sick! In contrast, comparison of 1st generation with the 24th on the digital, showed no audible difference, in the listening conditions that prevailed at the press release!
Is it possible to predict what is likely to be a landmark in the recording world? In mid-1982 it is looking as though one particular piece of recording equipment is very likely to be at the centre of just such a prediction. And this writer is fully prepared to stick his neck out!
It's digital, it records on Video cassettes, it's around high quality reel to reel prices, although much cheaper to run than high quality reel to reel at 15 ips, and has created behind the scenes, reactions among hardbitten reviewers and digital sceptics that it is the first digital system without any curiosities!
The system is Sony's PCM F1, with separate mains power unit which feeds any domestic video recorder — VHS or Sony's own Betamax.
There is a matching, similarly sized Betamax recorder (see photo), without TV tuner, which also has a separate mains supply unit. If both the mains units are discarded and Nicad battery packs installed in the F1 and the Recorder, a system is created which must be the highest quality 'portable' yet. For video use there is a third matching programmable tuner/timer unit.
This article will confine itself to the Digital Audio use of the system, but overall the scheme produces very much the complete home video/audio installation.
We had the Sony PCM F1 and the supplied AC700 mains unit, shoulder strap, special connection cord for the matching video unit and phono to BNC cords for other recorders or digital dubbing on trial. The matching recorder SL-F1E/UB (PAL/SECAM) and its supplied mains units AC-F1E/UB completed the set up.
The audio inputs, levels and impedances for line and mic, and the line outputs are all similar to present day cassette decks. The video output, input and 'copy out' on the PCM F1 processor are also phono sockets (gold plated!) and are at the standard 1V P to P at 75 ohms.
There is absolutely no problem connecting the arrangement into the standard HiFi system. When recording, the input is fed to the output (after A to D and D to A) which can be handy in a monitoring situation. The mic inputs suit low impedance mics (200 to 600 ohms) and of course with the quality potential available one must have at least a decent pair of capacitor microphones.
Stereo inputs, line or mic are connected to the PCM F1 A to D unit. Conversion of both channels at 16 bits (optional 14 bit use provided), sampled at 44.1 kHz is used. A Time Division Multiplex arrangement is used to handle the two channels i.e. the left is sampled and quantised and its 16 bits fed to the video recorder, followed by the right channel in a similar way and so on.
On playing back, via the PCM F1, the appropriate D to A conversion provides one of the most superb stereo line signals.
Three small problems were encountered during the use of the system. The first concerns the noise of the video deck — not obtrusive when the music is present and certainly nowhere near that from the U-matic once used to master a disc release, but it can be heard in a quiet domestic environment and at on-site monitoring rooms.
The second relates to the tape position counter on the SL-F1E/UB which is far too coarse in its operation — it counts hours, minutes and tens of seconds. This is probably fine for video programme search (who cares if ten seconds of Coronation Street is missed) but not enough resolution for audio work. Out live with mics the technique has evolved of letting the recorder run on for 5 secs after a take and another 5 secs before a new take. Then the very useful 'return to zero' rewind function is used to get a safe accurate start for playback or a new take. After a take is considered 'in the bag' the counter is set to zero to facilitate this. Also one has the bonus of having the timing displayed for each take.
Lastly, the system cries out for a ganged Master record level control and smaller separate presets. Thus recordings, particularly from mics, can be faded down or controlled without interchannel level changes. The presets would allow the master to be operated over a decent range of rotation. A final thought — if the presets were controlling the feedback in the mic preamp, then the circuits head-room would be automatically tailored to suit higher level inputs. If a separate mixer were used via the line inputs then no doubt all this would be taken care of.
A pair of Calrec CM652D cardioid capacitors were tried. These via their power unit are unbalanced, which is ideal for the PCM F1 mic input. The mics do, however, easily allow 50 metre plus lead length due to the low source impedance and the high signal levels available. One has to be in the capacitor mic class to take advantage of the PCM system — alternatives, at higher prices would be the AKG C451s or Neumann KM84s.
Peak levels must be properly controlled in digital systems! Analogue tape systems squash slowly — nevertheless the writer feels that too many people drive analogue tape too hard, so there is a relative rise in distortion and an HF compression. Digital has no HF compression problem, right up to the top signal limit. But take the level above that for which conversion is available and the system cannot cope. Actually the PCM F1 appears unique in this respect. Its error correction arrangements seem to deal with short term excursions if one makes them!
The PCM F1 has an excellent LED metering system. This has a 70 dB dynamic range (revealing to see traffic noise showing on classical sessions in London) covering -50 to +20 dB. The words 'over' are lit up if this is exceeded. A feature of the scaling is an expanded warning area. It pays to consider minus 15 dB as one's working zero, especially with mic inputs, and then one should never need to trespass into 'over'.
It seemed a waste of time to do extensive technical tests—tones sound like the oscillator itself. Especially 10 kHz and above — the absence of modulation noise behind the tone is a revelation. No unsteadiness on a 3 kHz tone could be heard — a notorious frequency to show up wow and flutter. The response was 10 Hz to 20 kHz at all levels.
Subjective impressions — exceptional cleanness and precision stereo imaging of recordings. Especially from a decent pair of mics. Other sources are simply replayed complete with whatever quality or faults they already have.
Tape running costs — Reel to reel at 15 ips probably works out at about £15 per hour whereas with the Betamax tape we used the running costs of the Sony PCM F1 seem to work out at about £8 for 3 hours!
Part 1 (Viewing) | Part 2
Gear in this article:
Feature by Mike Skeet
Previous article in this issue:
Next article in this issue:
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!