Sony PCM F1 Digital Audio Converter Page 2

Dealing with digital criticisms
I shall deal now with a few other criticisms which have been directed at all digital recording systems:

(1) "It changes the timbres of musical sounds." It doesn't. Accuracy of instrumental timbres is in fact one of its strongest points, for—unlike disc playback—the system is totally free from any mechanical resonances whatsoever. The sound is absolutely uncolored!

(2) "Digital fouls up massed-string tone." I did not get to record a large orchestra, but can vouch that massed-string tone from analog- (or direct-) mastered discs is reproduced exactly.

(3) "Digital erases the 'airiness' from musical sounds." Again, PCM copies of "airy"-sounding discs proved that it does not. Since the system's upper limit is 20kHz, this is strong evidence that that "airiness" is not a function of frequency response out to 30kHz and beyond.

(4) "Digital's anti-aliasing filter causes extreme phase shift." This is very true. Would you believe 180° of rotation within the audible range? Would you also believe that the system rings furiously at around 17kHz (see fig.2), produces prodigious amounts of ultrasonic garbage above 20kHz (although at very low levels) and generates several spikes of ultrasonic energy, with the worst one at 100kHz. The PCM-Fl. does all of these things. Horrible, huh? Well, it looks horrible on the 'scope, but dammit, there is absolutely no indication from its sound that anything is amiss at all.


Fig.2. A 2kHz squarewave shows horrendous ringing at around 16kHz. It was not audible.

And I don't care what you tell me about the audibility of all these things. Granted, they do seem to affect the sound of reproduction from discs (with analog equipment), but there has been mountinq evidence that disc sound is unique, in that things which affect it dramatically have little or no effect on taped or direct-feed (from microphones) signal sources.

You see, when It comes to a recording system, I .In mil give two hoots in hell how "good" it sounds. (The ultimate, in-the-room reproduction of music is quite another matter, though. See "As We See It" in this issue.) The only legitimate criterion for judging a recording system is how closely its playback approaches the sound of the original input signal. If it replicated the original perfectly, and you don't like the sound, blame the original signal, not the recorder. If you want a good cloning job, you accept the warts along with the glamor.

Much of the widespread criticism of digital recording compares it, unfavorably, with analog tape and Direct-to-Disc analog. Analog does this, digital fails to, etc, etc, etc. So, I reason, if analog is capturing things that digital is losing (or distorting), those fidelity losses should be audible when digital is used to copy one of those fabulous analog recordings, and the analog original is used to compare the digital copy.

Those of you who are already shaking your head in disagreement are urged to try for a moment to think with that head. This is an exercise in the simplest form of logic: If analog recording X retains A, B, and C, and digital recording Y reduces or eliminates A, B and C, then digital copy Y of analog original X should have less A, B and C than does X. If you can fault that reasoning, you weren't paying attention. Re-read the preceding until you can accept its truth, or go get your fortune told!

Well, it turned out that Y was indistinguishable from X in virtually every instance! No loss of ambience, no distortion of musical timbres, no loss of "air" at the high end (fig.3). The only difference that was ever audible was a very slight improvement in the characteristic stringy sound of massed violins playing loudly, when using a moving-coil cartridge up front. I suspect that this latter was because, in filtering the input signal above 20kHz, the digital copy was relieving the power amplifier from having to cope with all that ultrasonic garbage that wide-range cartridges spew out along with the program material.


Fig.3 Audiotracer frequency response curve of the PCM-Fl's playback (5dB/vertical div.).

In other words, the PCM-Fl is the ultimate copying machine. You can feed it signal from any tape or disc and get back what is for all intents and purposes a perfect replica of it. (It will not, unfortunately, make your phono or cassette deck any better.) In this respect, it wipes out any analog recorder money can buy, as has been demonstrated by several people who have tried cross-copying (analog to digital, digital to analog) to and from even the most expensive professional analog decks. Obviously, though, the PCM-Fl's strongest suit is live recording, with the best microphones money can buy.

With the longest available Beta tape (L-830), you can get up to 5 hours of recording on one cassette (up to 8 if you tape on VHS) at the slowest VCR speed, or 3 hours and 20 minutes at the slower Beta II speed. If you pay list for your cassettes, this works out to a bit more than 8 cents per minute (16 cents at the faster speed), for a degree of recording fidelity that is otherwise unequalled.

Incidentally, the only advantage of using the higher tape speed is that, if a tape dropout (momentary signal loss) occurs, it will be of shorter duration and thus more easily filled in by the unit's error-correction circuitry. The error correction is ingenious, by the way. Instead of laying all of the digitized pulses on the tape in consecutive temporal order, they are interleaved so that the pulse series representing each sample of the signal is spread out along the tape instead of occurring in a single burst. Thus, if one part of a pulse burst is lost for a moment, there is usually enough of that burst recorded adjacent to the drop-out to allow the original sample to be reconstructed with a fair degree of accuracy. Instead of the entire signal disappearing during the dropout period, there is only a very slight, momentary reduction in its fidelity for that period.

I mentioned earlier that a switch on the PCM-Fl allows you to select 14-bit or 16-bit encoding. Most of our tests were done at the 14-bit setting, but those two extra bits can buy you about 5dB more of signal/noise ratio. The cost, however. is rather less effective error correction, with a corresponding increase in the audibility of any dropouts that might occur. On the other hand, we used the slowest (Beta III) tape speed and the 14-bit encoding for a total of about 6 hours of recording and never once heard any evidence of a dropout.

Then there's the matter of overload. Unlike analog tape, which goes more or less exponentially into overload distortion, digital overload is an all-or-nothing affair. When overloaded, the first PCM-1 could emit a pistol-shot bang that was equally threatening to nerves and to speakers. The PCM-Fl appears to have a couple of dB of headroom between the point where the record level shows red and that where signal distortion sets in. When it IS overloaded, the sound becomes filthy-dirty, but there is no bang. (The record-level indicators are weighted to favor high frequencies, where overload of the medium is more likely to occur.)

A note about copying. Duping from one tape generation to another has always been standard operating procedure with analog recording. Most pop and rock recordings in fact are the result of many duplicating steps, in which layer upon layer of sound is built up to create The Creation. But analog tape copying has always involved substantial losses in sound quality—accumulated distortion and scrape-flutter sidebands, increased flutter, and snowballing hiss. Wider tape tracks, scrape-flutter "filters," higher recording levels, and sophisticated noise-reduction techniques have helped to minimize the losses but have never eliminated them. Digital tape is far superior in terms of all of these performance parameters, but it too has a certain amount of noise and distortion, both of which could be expected to increase each time a copy is made—if the conversion from digital to analog and then back to digital is done for each copy generation. But there's a sneaky way around this.

Digital's noise and distortion come from the analog circuits. If the signal is kept in digital form for the copying, neither noise nor distortion will increase from original to copy. There is provision for this in the PCM-Fl. A copy output, which can be switched on from the front panel, allows the digital signal from the tape to be fed to the copy machine without undergoing D/A and then A/D conversion. This, in other words, gives new meaning to the term "D-to-D."

And since (as it was described to me) the copy signal uses the off-tape pulses to trigger new ones, even dropping in error-correction ones when necessary, the copy made in this way should be absolutely indistinguishable from the first-generation original. Thus, even though split-second editing is not possible with PCM-Fl tapes (as it is with analog), a second VCR makes it possible to do such rudimentary program editing as the elimination of bad takes and rearrangement of the good ones, without any loss of signal quality.

I could wax enthusiastic over this for several more pages at least, but enough is enough. My excuse for devoting as much space to this as I already have is that I am convinced that the availability of a recording system like this, at the price, is the most significant thing that has happened to perfectionist audio since the stereo disc. If it does indeed "freeze" the state of the art, I cannot imagine a better state in which to freeze it. It will not of course make decent-sounding discs any easier to cut from digital masters, but it may eventually make this unnecessary.

This is, quite simply, an astonishing piece of hardware. I am going to buy one (and its companion SL-2000 Betamax recorder) if I have to auction off my mother to do it!

Sony Electronics Inc.
16535 Via Esprillo
San Diego, CA 92127

Archimago's picture

Excellent reporting from JGH at a time when vinyl was still the primary audiophile medium!

And he only used mainly 14-bits resolution back in the day.

thethanimal's picture

Did JGH read my comment on the MQA thread? ;-)

tiagoramossdg's picture

between posts. I thought of writing something about it, but did not want to seem paranoid.