Sony PCM F1 Digital Audio Converter

In 1978, when I reviewed Sony's first audiophile-type PCM-1 converter, I earned the undying scorn of a large segment of audiophilia by reporting that, on the basis of a rather short testing period (which did however include some live recording), I was unable to hear anything the matter with its sound. Four years later, but after substantially more testing, I am obliged to report the same thing about the PCM-l's son, the PCM-Fl.

Before I continue to dig myself deeper into this controversial hole, though, I think a little background might be in order. Most self-proclaimed audio perfectionists hate digital audio with a passion that treads a thin line between irrationality and outright paranoia. A few critics have attempted to justify their fear and loathing with seemingly reasoned arguments that the sampling rates and encoding resolution (number of bits) are too low. Others, with somewhat more justification, have expressed misgivings about the new digital technology "freezing" the state of the art, and thus preventing any further, incremental "improvements" such as have characterized the entire history of analog audio from Edison to Electrocompaniet.

My personal feeling, expressed here countless times in the past, is that the way something is done is of far less importance than how well it is done. If theory tells me that something has to be bad and it isn't, then I am not going to sit sullenly in the corner muttering "Well, it's doing those bad things all right, but I just haven't learned yet to listen for what it's doing badly." So, all considerations of bits and samples notwithstanding, I am hereby warning those digiphobes among you that this is going to be a rave review.

Most critical listeners, including those (like myself) who have continued to keep an open mind on the matter, have agreed that there are certain things unmistakably wrong with the sound of the majority of digitally-mastered analog discs. A nagging doubt has however been raised by those few discs, notably from Sonic Arts, which do not betray their digital origins. But the number of digital audio's shrillest critics who have ever heard digital tape (or the new compact-disc system) can probably be counted on the fingers of one hand. And the audible difference between the digitally sired discs and "the real thing" is a hard, cold fact which cannot reasonably be denied any longer. Digital tape can be incredible!

Testing
First, let's lay a few ghosts and deal with another. The sampling rate for all non-professional digital recording systems is 44.1kHz. Each alternate sample feeds an alternate stereo channel, so each channel is sampled 44,100 times. What this means is that the system grabs and preserves an instantaneous fragment of the analog audio signal every 44.1-thousandth of a second. Many people who pride themselves on their technical knowledge will swear to you that such a system cannot possibly reconstitute a 20kHz sinewave.

For them, we reproduce here an off-screen photograph of the PCM-Fl's playback of just that: a 20kHz sinewave (fig.1). The waveform shown was recorded at a level 10dB below nominal zero. You may take my word (and that of John Bau of Spica Speakers, whose time and equipment were volunteered to run the instrument tests) that the waveform did not change visibly at levels from zero down to the soup. More about the soup later, though.

621Sonyfig1

Fig.1. A 'scope trace of the PCM-Fl's reproduction of a 20kHz sinewave. The wave shape did not vary at any recording level from zero to noise.

Ahah, say the purists: A 20kHz sinewave is one thing, but music isn't sinewaves. How does digital reproduce transients? The quick and flippant answer to that is: Unbelievably Well (see Sidebar 1). Better than any analog recording including direct-to disc. In fact, several audiophiles I've talked to who have recently heard true digital for the first time have complained that it seems to reproduce transients too well—a tribute, perhaps, to the fact that we have too long relied on transient exaggeration from speakers and amplifiers to compensate for what was being lost through analog recording.

Sound Quality
One of the live tapes I made was of a chamber orchestra that included a harpsichord. Using only two microphones for pickup, and playing back with the tweeters set for proper string sound—an amalgam of unctuous sweetness with a small amount of resinous bite, the pings of the harpsichord were clearly audible even when all the other players were digging in hard. This combination of sweetness and razor-sharp inner detail is almost impossible to attain with analog sources, and I have never heard it accomplished as well as it was with this recording system. In truth, digital transient reproduction, at least from the PCM-Fl, is absolutely hair-raising when first heard—indistinguishable in fact from what comes right out of the recording microphones.

Which brings me to a daunting detail: Input signal. Sony included with our PCM loaner a so-called demonstration tape of a "Jazz" group. It turned out to be four instruments—vibes, drums, bass and piano, plus an embarrassingly wobbly voice on some numbers. My reaction was that it was a ludicrous effort to show off a system with the performance capabilities of this one, but it did reveal one thing: The primitiveness of my own recording equipment.

The crystalline liquidity of the sound from that tape was simply astonishing, and the comparison with the live tapes I made (using what I considered to be topnotch microphones and input-matching devices) was a little discouraging. Here, in other words, was a recorder that was better than most professional-recording input equipment. (Only Sheffield's latest Direct-to-Discs can match the PCM's liquidity, and I'm not sure they can equal its transient capabilities.)

Another accusation aimed at digital is that its signal distortion may be fabulously low at high record levels, but increases drastically as the level goes down. We could find no evidence of this with the PCM-1, although a British publication reported having measured an increase from 0.01% to 0.22% when recording level dropped from Zero to –60dB. As they pointed out, this is far lower than the distortion ever measured from any analog recording, at any record level, and it is unlikely that distortion at 113dB below Zero level would be terribly audible.

Many digital foes have also claimed that digital "chops off hall reverberation." I was prepared to believe that one, because the system does have a "floor," below which no signal can be recorded. But, try as I might, the only way I could find any evidence of that floor was by cheating.

I PCM-taped a selection of wide-dynamic-range audiophile recordings that had lovely hall ambience on them, and direct comparisons with the originals showed no loss whatsoever of any ambience. I taped them again at reduced volume levels, in hopes of dropping their ambience level to the PCM's floor, and failed to hear any difference until I had the loudest parts reading –40 on the record-level meters. Then I could hear, not only ambience loss but loss of the quietest musical information too. And because I ran out of system gain with the volume cranked up full, I had to use headphones with an external amplifier to follow the signal all the way down to inaudibility in order to hear the noise-floor losses happening.

This doesn't seem to make sense. With an available signal-to-noise (floor) ratio of around 85dB (using 14-bit encoding—the PCM-Fl offers a choice of that or 16), and the system playing at peak levels of 100dB, the noise floor and hence the ambience cutoff point should be 15dB above hearing threshold—quite audible. Yet when I played back a live piano recording, miked extremely closely (from inside the piano), at 100dB-peak level, no floor was detectable. I could even hear the rustling of clothing between recorded segments.

So, another ghost laid to rest! I cannot hazard a guess as to why the PCM-Fl does not seem to abide by the accepted rules here, and can only conclude that the figures we have been tossing around when we speak of dynamic range in music must be way off the beam.

COMPANY INFO
Sony Electronics Inc.
16535 Via Esprillo
San Diego, CA 92127
ARTICLE CONTENTS

COMMENTS
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.

X