Home Recording With Digital (Part 1)
Sony PCM F1
Article from Home & Studio Recording, December 1983
Introducing the Sony PCM F1 recorder.
Most classical vinyl LPs are digitally mastered these days. These are generally just two channel stereo recordings as the relatively simple microphone techniques used suit the deriving, on site, of the stereo master. Often just a crossed pair of mics are used or at most, a pair augmented with a few more pairs of single spot mics. The recording systems used, employ either a separate Analogue to Digital (A to D) processor, feeding a video recorder or an adapted specialist reel to reel recorder, with the necessary A to D electronics. Editing is usually achieved with the aid of further digital processors, although with some of the reel equipment, the inclusion of analogue tracks allows 'razor blade' editing.
The object of the 'professional' recording digitally was initially just to get the improved mastering quality. The finished result could only then end up on a vinyl LP or the cassete equivalent. This is apart from broadcasting organisations, like the BBC, who regularly broadcast from their digital masters. Now we have the Compact Disc (CD) available to the consumer, allowing the benefits of the digital process to be applied all the way to the listening end. This in itself should increase the number of records that are digitally mastered and that can remain, in the 'digital domain' as it is called.
In the non-classical music field, there are very few equipped with digital multitrack recorders, so it will take some time for the all digital domain idea to get a foot hold in this area. Mind you, from the comments often made about multitracked productions that are released on CD from analogue masters, the recording industry needs to look at its techniques as CD has shown up all sorts of defects that vinyl LPs had previously concealed.
The capital cost of all these various bits of pro digital gear is very high, but such is the way with technological advances that prices will soon come tumbling down especially with the wider markets found. It's still early days for digital in the Home Studio scene, but there is one remarkable system, still two track I'm afraid, with a hidden flexibility and a performance equal to or even ahead of the pro gear. It is the Sony PCM F1 system, which uses domestic video recorders and is in fact being used in the pro field in place of another earlier Sony system, costing ten times as much! Getting away from the video tape storage will be the next step. Can we look forward to a Revox D77? Even now we hear from Japan that special versions of the compact cassette are around in prototype form with specifications in the F1 class!
When one looks at the specifications of a good digital system, one could be forgiven for assuming that one is reading an amplifier spec sheet. Three of the parameters are 'beneath measurable limits'. These are wow and flutter (speed consistency), print through (pre and post echoes) and erasure of previous recordings. All other parameters measure way below those of any analogue recorder, and hence the term 'transparent' is applied to systems such as the Sony PCM F1. See Figure 1.
In my view it is absence of tape hiss, modulation noise, lack of high frequency compression and rock steady speed that subjectively shows up digital's virtues. Allied to this is the fact that tape costs are much lower than reel to reel at 15ips ½ track, as any reputable video tape can be used without any deterioration in performance or the need to be concerned with the bias or EQ! Further there are no 'copying losses' in making digital to digital copies nor for that matter, if one works using two systems (or more correctly one and a half, as I shall reveal) and does the copying in analogue form for overdubbing purposes. With multiple overdubs it is still possible to outdo such activities done on home multitrackers, be they cassette or even reel to reel.
There are just a few basic aspects of the Digital encoding process of an audio signal, that need be considered. It's all based on measuring the (amplitude) height of the audio waveform, at any one moment, encoding the height information in Binary Code form, storing the code and on playback, decoding it, to recreate the waveform in all its intricacies. There are demonstrably far less changes to the shape of a complex wave after passing through the Sony, than with any reel to reel at 15ips, let alone cassette!
In doing this encoding and rebuilding, several factors emerge. One has to look at (sample) the waveform at sufficiently short intervals to detect the nuances in the changes of waveform shape that the presence of high frequency harmonics dictate. To achieve an upper frequency response limit of around 20kHz, sampling needs to be done at least twice this frequency. In practice, with the Sony PCM F1 working with a PAL SECAM video recorder it is done at 44.1kHz.
When the waveform height is measured there has to be sufficient resolution to avoid errors in the reconstruction. These 'errors' will in fact manifest themselves as noise - called Quantization Noise. Telephone systems (yes British Telecom!) have used PCM connections between UK towns and cities for a decade or more. The international standard for such systems is to use 8 bits (binary digits) in the Binary Code. This results in 256 (28) different levels being accurately detectable - not enough for music. A world standard for excellent music performance, of 16 bits has thus emerged. The CD system is based on this. There are then a staggering 65,536 points of accurate resolution when the waveform is reconstructed and a noise performance of over 90dBs below the systems' peak.
How is stereo handled? Simply by dealing separately with each channel in turn and appropriately switching at the replay stage. How can the wow and flutter performance be so good at such slow video tape speeds? Well, of course the rotating head system used for video actually involves a much higher reading speed than the linear tape speed suggests. But there is more to it than that. The information coming off the tape goes into a temporary store and is very accurately clocked out. It is a continuous process and provided the information blocks enter the store in sufficient time then there is no failure of the read out. Overall there is a 10mS delay through the F1 system.
The F1 system has extensive error correction electronics. This is to overcome information dropout due to detail imperfections in the recording medium. There is also a redundancy system used where information is recorded more than once in an interleave manner. Dropout removing one set of data will be of no consequence. Also there is interpolation, where 'guessing' is applied to keep the system going. In common with other digital systems, if dropout is outside what redundancy and interpolation can correct, then muting occurs. At least there isn't the gradual worsening of performance as there is with analogue. It's all perfection or nothing!
The Sony PCM F1 processor is a very small product considering the complexity it packs. There is a separate mains power unit or the processor itself may be powered by an internal NiCad battery, chargeable from the mains unit. The F1 is the same size as Sony's SL F1 portable Beta video recorder, and this is the unit I would recommend. Other Beta's are usable, as are VHS units. I have no experience of the latter but there seems to be a problem of 'run back' at each stopping point. Recording and playback are excellent of course but it's all not so neat and easy to use.
Figure 4 shows the basic 'ins' and 'outs' of an F1. If you can already couple a cassette deck into your system, then you can easily substitute it with an F1. Phono connectors are used for the line in and line out, with impedances and levels similar to a cassette deck. But, in practice, as one records with a digital system set more below its peak level than one would with a cassette deck, there is perhaps the need for a bit more output level although this can be compensated for in the equipment it is connected to. The video connections are also by phono socket. Notice the low pass filters, pre-emphasis and de-emphasis before the A to D and D to A sections. There is also pre-emphasis before the metering circuitry. Notice also the connection of the metering circuitry at the output. This is of significance when we come to consider the overdubbing usage next month.
The metering itself is excellent, but like all metering systems, it needs interpreting. It is peak reading, as it is vital that the finite peak point where you run out of quantizing levels is not exceeded. Actually the system just limits if one exceeds this, there are no nasty noises.
The LED display is not a column type but incorporates the unique idea of just having the highest LEDs lit, apart from the several seconds retention of the highest peak reached. Additionally 'peak hold' can be switched in, so that at the end of the scaling two red 'over' legends light up, if such has occurred.
In use, it is best to have the nominal operating level set at around the minus 20dB indication. As this is approximately -16dB in practice on all the processors I've come across, one is not working at too low a level.
Even this is still in fact some 74dB plus above the noise! But there is a good reason for this which is tied up with the pre-emphasis mentioned earlier. There is approximately 7dB of pre-emphasis at 10kHz with respect to 1kHz. The metering has to show this, as the digital crunch point is that much nearer at high frequencies (HF). Therefore it will depend on how much HF there is in the signal and on the level of the peaks one allows above the nominal operating level you have chosen. Initially you will be puzzled by the liveliness of the reading when, for instance, cymbals are played or there is audience clapping. It all came home to me at the last night of the Proms in the Albert Hall with an F1 and a Soundfield mic just how much HF energy there is from the Promenaders at 4 metres above their heads!
Remember though during recording, that one is not doing the usual analogue trick of getting the mean level as high as possible, with the inevitable HF compression and high intermodulation distortion, but merely letting the dynamic range have room to realistically rise. I bet many who use an F1 to record for the first time will feel it strange that the peaks come through without the compression of analogue that we have become so used to.
As mentioned earlier, if a short duration peak does pop across the line one does have a fine limiter! And it still sounds OK. On playback the overload segment only shows if the digital copy out switch is operated. The over indicator, it seems, is triggered by all 1s or 0s in the digital code.
When making digital copies it is best not to just link two video recorders together, although in fact this will work. Better to use the digital copy out facility on the F1 and benefit from a fully error-corrected feed. The copy out socket, when the switch is at normal, doubles as a parallel feed to an additional video recorder should a security copy be needed. This is common practice in the pro recording world even when working with the large cumbersome U-Matic recorders some operators use.
The best way of connecting up whatever inputs and monitoring system one is using is shown in Figure 5. Notice that the monitoring is from the F1 line outputs. This has the advantages that one gets automatic selection of playback when the playback button is pressed and when recording, the signal being monitored has gone into the video recorder and out again - not on and off tape you understand but as near to it as possible. Hence one is reassured that the connection is complete and barring a faulty tape, all is well.
If we have whetted your appetite for digital mastering - that was the object. Next month we look into the F1 system and show how with one processor and two video recorders, one can overdub in analogue form with results exceeding even reel to reel multitrackers!
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!