Meridian's Bob Stuart

Meridian's MCD CD player was perhaps the first audiophile-quality player to be introduced in the high-end market. I met with Bob Stuart of Meridian at the Winter Consumer Electronics Show in Las Vegas, in January, 1986 (footnote 1). My first question was about the name of the company he runs with industrial designer Alan Boothroyd:

J. Gordon Holt: Meridian in England is called Boothroyd Stuart, right?

Bob Stuart: Yes, the company is called Boothroyd Stuart, Limited, and the trademark is Meridian.

Holt: What is your function at Meridian? Sales, engineering, design, or what?

Stuart: Well, up until quite recently I did virtually everything: I ran the company, designed the products, handled sales and promotion. The company grew—it started in 1977—out of a design partnership of Alan Boothroyd and myself. Alan is an industrial designer; his specialty is the appearance and the mechanical aspects of the products—the way they look, the way they're constructed, and so on. My part is the electronics end of the design.

Holt: You are solely responsible, then, for the sound of Meridian's products?

Stuart: That's right.

Holt: I'm sure you're aware of how some members of the audio community feel about digital—the widespread opposition to CD. You must have known that any high-end electronics manufacturer that produced a CD player was going to earn the scorn of a lot of perfectionists. At what point did you decide it was worth doing anyway?

Stuart: The CD has never encountered as much resistance in the UK as in the States, so we didn't see where producing a player would be damaging to our reputation. We got involved with Compact Disc in rather a curious way, though. As an established high—end manufacturer of amplifiers and speakers, we were always concerned about the source material used with our products. And with phono units, we had the same problems that everybody has—lack of consistency. No two ever sounded alike. No matter how carefully we worked over refining the sound of our speakers and electronics, we had no way of knowing what they would sound like in a dealer's demonstration, because every dealer used a different phono unit. Everyone had his own favorite, and they all sounded different.

Holt: Couldn't you have recommended the phono equipment to be used for your demos?

Stuart: We could have, but no one would have paid any attention. They would have used what they preferred or whatever they had on hand. So when we first heard about CD we were very excited, because it looked like the way of ridding ourselves of those source material inconsistencies. But we were very disappointed with the sound we heard from those first players. I bought one of the first-generation players, a Sony CDP-101, and while it certainly showed that we now had a consistent program source, it was a consistently poor program source. We put it to one side. Actually, we had been doing a little bit of work with digital before that. We got interested in the Sony PCM-F1 system, and made some modifications to it and got it to sound quite good. But I could never understand why Sony had opted for linear encoding instead of logarithmic.

Holt: Why logarithmic?

Stuart: Because at high signal levels, each decibel of level change involves a much larger increment of signal change than is involved at very low signal levels. Linear encoding provides more resolution at high levels than is needed, and not enough resolution at low levels.

Holt: Yet the linear encoding seems to work quite well.

Stuart: Well, yes. We did some very interesting listening tests with that PCM-F1 system—tests which showed that if you didn't go onto tape with the PCM signal, but just processed and deprocessed the signal through the PCM adaptor, the digitizing introduced so little degradation that one was very hard-put to hear the difference between the original signal and what came out of the processor. In fact, one of our tests, at a seminar involving about 50 of our UK dealers, was done under very carefully controlled conditions, setting levels to within 0.01 dB and using computer control to do the switching. There was no statistical evidence to indicate that the difference between the original signal and the digitally processed one was audible at all.

Holt: So this was the thing that convinced you that . . .

Stuart: That it was possible, yes. It proved to us that you could store and reproduce musical sound well enough in digital form to provide a consistent and high-quality program source. It proved to us that the digitizing process was satisfactory, and that the audible problems we had heard from PCM tapes and from the Sony '101 were due to other causes, which should be remediable. That's what encouraged us to start investigating the weaknesses in existing error-correction systems and some of the odd things we were observing about the interactions between CD players and preamplifiers. Like the fact that, with CD sources, tubed preamplifiers generally sound better than solid-state ones. Preamplifier designers will tend to optimize their designs for one input, usually the phono one, paying less attention to the high-level inputs.

Holt: I've noticed that CD players are often at their worst with very wideband preamplifiers, designed for HF response out to 250kHz and beyond. Was this a problem with the Meridian preamp?

Stuart: No. We've never felt it was wise to design for an upper frequency range greater than the program material could provide.

Holt: In other words, you use bandwidth limiting.

Stuart: Yes.

Holt: Yet Meridian was, if I recall correctly, the first preamp manufacturer to incorporate a specially filtered CD input.

Stuart: I believe so.

Holt: If you were already filtering the range above 20kHz, what additional filtering did you feel was necessary for CD?

Stuart: We didn't limit the HF response any more; we just gave it a steeper cutoff slope, to reduce further the ultrasonic interference that remained in the signal after processing by the CD player's own filters. Remember that, at that time, none of the players had digital filters; they had very steep analog filters. Even so, the 44kHz pulses were only about an octave removed from the topmost audio signal, so it took an 8-pole filter just to bring those down to 48dB below the maximum signal level. That wasn't enough, though. Any distortion at all in the electronics would. . .

Holt: And even the best electronics have some distortion, as slight as it may be.

Stuart: Yes, of course. And that distortion would create intermodulation effects between the signal harmonics and the ultrasonic energy, and the difference tones would extend downward into the audible range. Our CD filtering merely did more effectively what the CD player's filters were supposed to do.

Holt: What prompted you to turn your attention to the CD player itself?

Stuart: Well, it became obvious to us that much of the trouble with CD sound was due to inadequate filtering of that ultrasonic material. We discovered that, in just about every player, Japanese or European, the digital part was being pretty competently dealt with, but the audio parts looked as if they had been designed by computer engineers who had no understanding whatsoever of analog audio topology. That's how we got into it; we rebuilt the audio stages of a CD player, and we were amazed by how good it sounded. Of course it probably helped that we chose to start with the best-sounding production deck.

Holt: If you had chosen somebody else's deck, which perhaps had much worse sound than the Philips, and then reworked the analog section in much the same way, do you think you could have ended up with something as good as you now have?

Stuart: No, because I think an essential point of the one we used is the oversampling D/A conversion. If we had used a one-to-one, which is what nearly all the Japanese players were based on, it would have posed a nearly impossible filtering problem. I don't think it is possible to build an inaudible 22kHz filter in the analog domain. The real advantage of the 4-times-oversampling is that it allows most of the filtering to be done digitally, and this eliminates the ringing and severe phase shift problems which are inherent in very steep analog filters.

Holt: My only complaint about both of your players is the deck itself. It's very slow, it does not allow you to do audible cueing, and you can't even call out a specific band without standing there and pressing a button over and over. You can access band six of an analog disc in a tenth of the time it takes to get there with the MCD or the MCD Pro.

Stuart: That's true. Those things are clearly relevant, but they're not very important to someone who just wants to sit down and listen to a recording all the way through.

Holt: Perhaps the reason they irritate me so much is because, as a tester, I often have need to access specific bands of a disc. . .

Stuart: Like track 99 on Denon's test record.

Holt: That's the pink noise band. You must have been reading the magazine.

Stuart: We use that disc too. But you know, for the music lover, it doesn't matter. The slow access time can even be an advantage; it gives you time to sit down before the music starts.

Holt: I think I still prefer to sit down with a remote control unit and choose my own moment for the music to start.

Stuart: One of the other major things we did in the player, besides redesigning the audio stages, was to use a faster integrator circuit and better topology in the power supply. And we changed the grounding to conform to good, sound analog engineering practice. Then we isolated and improved the power supply regulation to the various parts, and added better coupling caps in the analog circuits. Then somewhere along the line we noticed that the sound was very different when you put two discs into the player together. And it was different if you put the player on the floor, on the carpet. In theory there was no reason why that should happen, in digital. We found what the problem was: power supply modulation.

Holt: From the overworked servo system?

Stuart: Exactly. You see, the tracking servo consumes more current than any other part of the entire player, and if for any reason it was forced into constant action, it would deplete the power supply voltages to the rest of the player, causing the analog circuits to change their distortion characteristics.

Holt: What would cause this to happen?

Stuart: Any up-and-down motion of the disc surface, due to outside vibration, or even to the rotation of the disc itself. The second disc minimizes both.

Holt: So it has nothing to do with defocusing of the laser beam, or with stray light bouncing around inside the disc material, as claimed by some other manufacturers of "CD stabilizers."

Stuart: No. We examined the bit stream when this problem was showing up, and found that it was intact. In fact, with most discs, the error rate is very low, and nearly all errors are fully corrected.

Holt: Your MCD Pro has two lights on it to indicate corrected errors and interpolated (uncorrected) errors, and I have yet to see either of them go on.

Stuart: They do, occasionally. But the interesting thing is that, when the yellow light flickers, indicating a corrected error, you can't hear it at all. But whenever the red light flashes, for an interpolated error, you do hear it.

Holt: What do you hear?

Stuart: A click.

Holt: That's it? No distortion?

Stuart: No, just a small click, rather like that from a disc. Anyway, we did the work, and we discovered quite clearly that what was happening was that the rotating disc often goes into violent up-and-down vibration while rotating, and the servos are so successful in tracking—we found they would track to within half a micron on a one-millimeter warp or wobble!—that the power supply would show really serious fluctuations. That affects the analog stages, as one would expect. The thing that was extremely curious was that, when we started to separate out the power supplies, we found that those fluctuations were also interacting with the digital part.

Holt: Oh? Could you explain how?

Stuart: Yes. The power supply was jittering the master clock.

Holt: Changing its frequency?

Stuart: Yes. Causing a small FM modulation. And, in a PCM system, a small error in sampled time amounts exactly to an amplitude error. One least significant bit of amplitude is equivalent to 200 picoseconds of time. That is a very short period of time. A quartz crystal oscillator is extremely stable in the long term, and can be stable in the short term from wave to wave, provided that the oscillator itself is stable. But if you think about it, an oscillator is in fact not a digital component. The oscillator is followed by a voltage divider, and digital dividers are subject to jittering. The exact moment the divider triggers is a timing function, and if the timing is off, the output pulse will not correspond in amplitude to the digital code.

Holt: And this causes an amplitude modulation of the output signal.

Stuart: Yes (footnote 2).

Holt: This problem was solved in your first player, the MCD?

Stuart: Yes, by simply improving the power supply to that particular chip. In the Pro, we went even further with a separate power supply, with its own transformer, for the audio section.

Footnote 1: A more recent interview with Bob Stuart was published in September 2006.—Ed.

Footnote 2: Variations in the voltage on the supply rails to the D/A converter would also have this effect.