Tim de Paravicini: Dialing-in the Original Experience Page 2
JA: I understand from talking to some manufacturers at this show that American audiophiles, at least, are not buying as many turntables as they used to.
TdP: They're not buying as many, no. But there are still people buying them. There are still manufacturers of turntables in America.
We continued our conversation with a discussion of Tim's modification of the industry-standard Sony PCM-1630 for CD mastering. While he leaves the digital section alone, he basically rebuilds the analog circuits so that they minimally degrade the sound. (Martin Colloms's Stereophile review of the '1630 in Vol.10 No.5 mentioned that the analog sections of the processor are based on relatively ordinary NE5534 op-amp circuitry.) I asked Tim what CDs had been transferred to digital with his modified 1630.
TdP: Well, all the CDs put out by Charlie Records, See For Miles, and Affinity Records in England, quite a number by Dire Straits, Cat Stevens, Bob Marley, the Wailers, and Lincoln Kwaze Johnson—Island's catalog...And we did Kavi Alexander's CD [Arturo Delmoni performing Bach, Kreisler, and Ysaÿe, Water Lily Acoustics WLA-WS-07CD].
JA: What do you see as the major problems in digital replay and recording?
TdP: We are stuck with a system that has already been formatted by Sony and by Philips 10 years ago. We are well down that road and it's going to be awfully hard to change it. I think perhaps in five years' time, a second, up-market system, like HDTV in television, will happen in digital...That has got to happen. But at the moment I don't think the vested interests wish to have another completely new format.
JA: I could be cynical and say we already have a low-resolution, mass-market medium, which is CD, and an up-market, high-resolution medium, which is the LP.
TdP: That is exactly what I am trying to say. And the LP is still not a finished art. I'm still doing development. Other people are still doing development on cartridges and all the rest of the chain. Because the LP's an analog system, it's not bound by the same limitations of, for example, saying we will have 525 lines, take it or leave it. The parallel is that analog film, Kodak film rolling through a shutter, is actually holding its own. And in England a lot of the newsreel companies are going back to their Arriflexes, because there's a lot of creativity that they can do in film, in celluloid, that they can't do in video.
JA: And if you think that the analog medium, film, offers ten times the line resolution of good video...
TdP: And film is going through continuous improvements. They can improve the definition and the contrast and everything in film without upsetting the compatibility. They can still put it on the same projector. That is the whole crux of it. I think vinyl still can be a viable medium, but it's a matter of how it's [distributed]. The difference is that film and video in the studio industry are used by professionals, so they haven't got to worry about the distribution. Whereas with vinyl and CD, we have to deal with trying to distribute it, to disseminate it, as against keeping it in-house. I mean the same argument can be applied to magazines. Why aren't Hi-Fi News or The Las Vegas Times sold as microfiches? Why don't they get down and print magazines this small and sell people magnifying glasses?
JA: The answer is that the centuries-old format of the magazine is more portable, more easily accessible.
TdP: That's my argument. We should keep with things that are acceptable in human terms. I mean, vinyl is not practical for jogging but there is a time and a place for enjoying music. I don't go jogging when I'm in a concert hall—I may go jitterbugging but I don't go jogging!
JA: Tell me how you set your priorities when you designed EAR's The Mic.
TdP: Was it transparent? Did it capture the whole sound picture? A lot of the traditional microphones, for example, don't have enough bass response. When we're listening to live music in a theater, there is a dimension of the theater that with our eyes closed we could recognize where we were sitting, in the Royal Albert Hall, or the Queen Elizabeth Hall, or Barbican, or wherever. If you're familiar with the buildings, it doesn't take five seconds to realize which one you're in. Most of this is the low-frequency signature, the fundamental reflections of the building, the time decay of the building, and the low-frequency pattern—they all have to be phase coherent for us to recognize it.
JA: The English writer Rex Baldock years ago said that it was in the infrasonic region that the hall signature imposed itself on the music.
TdP: Precisely. And human beings use vibration of the body, as well as what we hear directly through the ears, to build the sonic picture. The tradition that we only hear down to 20Hz is rubbish. We actually perceive sound all the way down to 3Hz. Below which the body cannot experience vibration of any description; the body has designed itself so that we don't hear our own legs when we're walking, or running. The traditional concept should be rewritten to say "Experience of sound extends to 3Hz." But the traditional mike manufacturers tend to consider themselves doing well if they get a bottom E on a bass guitar. Commonly 42Hz. And they say that's fine.
The problem is that those microphones lose some of the information that should be there. So trying to understand the microphone's job is a compound picture. Do we hear? Do we feel what it's supposed to be doing? Even if some consumers haven't got reflexes to experience it all. That's irrelevant. As long as we can produce and inscribe into the finished product all that we can, then it allows those who are in a more privileged position to at least share the experience more validly.
JA: Does that mean that when you cut records, you don't roll off the extreme lows?
TdP: We cut records all the way down to sub-5Hz. So it's a matter of whether the guy's turntable has the correct arm and the correct resonances, and the damping on the arm, whether it retrieves the bottom end accurately or not—that is not for me to judge. As long as it's been inscribed on the record, I know that it can be recovered with the right equipment.
JA: Did you choose to use quite a large Milab capsule to capture that low-frequency information?
TdP: No, the capsule is moderate size. The reason for using the rectangular capsule is that it's more akin to a ribbon microphone, to a line source. Stereophony, bounded by walls, requires two pure line sources to reproduce the stereo picture. If the microphones are effectively line sources—in fact, my next-generation microphone will be even more of a line source, it's going to use a capsule some 2" tall and 5/8" wide—all the HF coloration is reduced in the horizontal plane and that gives us better localization in the horizontal plane. And the same way the head has temporal response—in other words, we hear certain frequencies above the head and in front of the head—I want the microphone to do that so that we can convey more of the dimensions of the sound that we are trying to record. So that we can position the orchestra in three dimensions. Left to right, as well as up and down and back to front.
JA: That's a real area of contention. Most theoreticians say that stereo imaging is only two-dimensional: side to side and front to back.
TdP: Well, we seem to manage to hear very effectively in the real world where a sound has come from pretty well all points of the compass. In other words, if somebody—whisk—comes up to you from behind, you know he's there.
JA: But there is some ambiguity in height perception. It is hard to distinguish whether you are hearing a soundsource directly under your feet or above your head.
TdP: Oh, there are areas of ambiguity, yes. So humans have the ability to turn their heads and try to zero in. But there is [a small enough degree of] ambiguity to at least make our choice error-minimal. We tend to turn our heads in the right direction in the first place.
JA: And that adds rate-of-change information which resolves the ambiguity.
TdP: Yes. But the more we can get that property into the recording, the more we will then have a recording where the head is stationary. In other words, it will have no more and no less realism than what we would experience if our head if we were sitting still in the recording environment. We would go to a holographic experience where we could say, "Let us dial ourselves into the original experience."
JA: To return to the mike, a lot of your design philosophy seems to be very traditional. The use of tubes, and particularly the use of transformer coupling.
TdP: Yes, because I have to be practical and use the mike in a conventional environment where there are long cables and where cables have to be of practical dimensions. It is no good me using cables that are 2" in diameter. And I use a transformer to be able to give me an output that will transmit down a 100 ohm cable correctly...A transformer can achieve consistently high common-mode rejection, it has the ability to withstand high superimposed noise signals...Active balanced circuits have weaknesses which transformers don't have—if designed correctly. The problem is traditionally what I call "wound" transformers that have been compromised too severely.
JA: It's another black art.
TdP: It's a black art. Yes. Likewise, tubes have dynamic properties that transistors don't easily have. And tubes are very rugged; they may have microphony, they may have other disadvantages, but at the end of the day they are extremely robust beasts. And they're not significantly any less efficient than transistor circuits. After all, power is power. You can't squash the watt, as I put it.
JA: A 100W tube amp is around the same physical bulk as a good 100W transistor amp.
TdP: Because heat is heat. I should also say that because tubes are made on a mechanical process, they're actually remarkably consistent.
JA: If you can get the skilled operators to make them—I understand that the Chinese company that's using the old M-O Valve company tooling had a lot of consistency problems in their tubes because of the lack of experience of their operators.
TdP: Experience, yes. But I'm talking about an all-things-being-equal condition.
JA: You do design with transistors for some of your third-party projects, Musical Fidelity, for example.
TdP: Yes, yes. I have no particular personal preference for any medium. Whether it's transistor or tube, each can do its job well. But through EAR I choose to purvey tubes, the same way that Rolls-Royce still purveys large cars and doesn't make small cars. But what I did for other people is less stringent.
JA: You also appear to be a traditionalist in amplifier design in that you reject the use of a lot of voltage regulation. Both your mike preamp and your G88 preamp that you introduced at this show (footnote 4) use traditional power non-regulated voltage supplies.
TdP: Because a good building doesn't require props to keep it up. I take the same approach: if you design the circuitry properly, it doesn't need prosthetic aids.
JA: But can't you see the temptation for designers when a 20-cent, 3-pin chip buys them 50dB of hum rejection?
TdP: Yes, but if I can already achieve that, why do I have to bother? It's as simple as that. And putting a regulated supply on is not going to give me any benefits that I don't have already. It's going to give me disadvantages, because it's going to make the product more susceptible to fragile power sources. Whereas when I have an unregulated system, the product is relatively immune from mains disturbances.
JA: I noticed yesterday you turning your preamp off while it was playing an LP without there being the usual speaker-destroying bangs...
TdP: I can have mains crap-outs for up to a couple of seconds without there being any tangible problem. The mains can crap out or brown out, and the preamp won't have a problem. Whereas with a lot of regulated circuitry, there is going to be a cutoff point before the parachute fails.
JA: And then it fails catastrophically.
TdP: Yes. I prefer to build equipment that, if all else fails, will still limp. So if you're in a recording session and there's a glitch in the mains, at least the recording will still go through okay. I don't want to be downed because of that one problem. The same way a lot of people are going back to cars with simple, mechanical engine management systems rather than the modern mega-electronic management systems because there are less things to go wrong. As a result, they're more likely to be able to limp home [when things go wrong]. And if an electronic management system fails while you're [passing], for example, you're a dead duck.
It is better to have a system that is less prone, long-term, to problems—but it has to be designed correctly to cope with the short-term problems. Regulation is only a device used by conventional circuits because they conventionally are too sensitive to their power supplies. I try to design the beast to be relatively immune to the supply. Now one of my tube-amp competitors puts regulators willy-nilly everywhere. But some of those regulators are actually noisier than the prime regulator they have at the beginning of the supply. So they have achieved nothing. It is just a sales argument: "We have 10,000 regulators in this product; it must be better." "More" is not necessarily "better." That is my argument. That's the whole point.
JA: I understand your G88 preamp also doesn't use any capacitors in the RIAA EQ.
TdP: I'm using inductors to generate the RIAA network, passively. There is no feedback. The reason is that all capacitors have different sonic signatures, which has been written about ad nauseam by many people over the last 10 years. An inductor is wound with copper wire and is easy to wind at consistent linearity that allows me to get away from the problems of voltage distortion that capacitors have. It allows me to produce a low-impedance network to keep the noise floor of the preamp down. It allows me to generate a good accuracy beyond the two ends of the audio band. And it also allows me to provide a preamp with a good overload margin at high frequencies so the clicks and spikes of the records do not overload and do not intermodulate.
Provided it's done correctly and shielded correctly, you don't have any hum problems. The inductors are used in a balanced configuration, but I'm not going to go into all the intimacies of how it's achieved. That's still proprietary. But I am formally stating that I am the first to do it.
JA: Do you use a transformer for the moving-coil input?
TdP: Yes, I still use a transformer. I'm using what is in fact the HEAD sans packaging in the preamp. [A transformer] is still the most efficient way of converting one high-grade signal to another level. I'm also using balanced topology all the way through the preamp so the phono amplifier ends up with a balanced output transformer.
JA: Do you have favorite tubes for each application? There appear to be two factions in the US. They appear to go for either the 6DJ8 or the 12AX7 and their derivatives.
TdP: I don't have any particular favorite tubes. I use both. It's horses for courses. But the 6DJ8 never should have been used as an audio tube. It was originally designed by Mullard as a television front-end tube. It was designed as an RF amplifier and an RF oscillator to operate at 200MHz. It's a frame-grid tube, consequently it was never designed to have low mechanical noise, it should never have been used for audio. Because of the fact that it was used in televisions to operate on the 90V supply, it is a high-current triode by comparison to the 12AX7. But it's only fit to do service in certain parts of the audio circuitry. It should not be used carte blanche all the way through. I use it in my preamp, but only in the high-level balanced sections since it does a good job of driving the transformer.
JA: One tube you never see is the EF86 pentode (footnote 5). I only ever recall seeing it in the front end of the Radford ST25 power amp.
TdP: The EF86 is a very good tube for preamps. It's an incredibly linear tube. It has the gain, it has low microphony, it doesn't need DC filaments to make it work quietly, because it was designed to operate with a center-tap 6V filament. In those days , DC filaments were prohibitively expensive, because of the cost of low-voltage, high-current rectifiers. It was designed for low-level microphone or photomultiplier amplifiers, projector sound and so on. So it had to have all these requirements designed in, it had to have a mechanical construction that minimized microphony. Quite the opposite of the 6DJ8, which was designed to do a certain job of work a different way.
I would use [the EF86] but for the fact that it is only produced in one country now. All the original Mullard stock is effectively finished, so from a production point of view it is not a viable tube to use. The problem was that people weren't using the EF86 in the '60s because they'd gone to transistors. Mullard didn't really see the need to keep it in production because it was only service-replacement volume, which was minuscule since tubes tend to have a life of the order of 100,000 hours. Power tubes have lifespans in the thousand hours or low thousands, whereas some of the small-signal tubes can have lives quite easily in the hundreds of thousands of hours—people don't realize that tubes can very nearly equate with transistors as far as longevity is concerned. I would love to use it. But the reality is that if you're going to produce more than one piece of equipment, forget it. It was a beautiful tube but unfortunately the uncertainty of buying it precludes it in modern-day manufacture.
JA: When you designed the Luxman 3045 in the '70s, you used an exotic NEC pentode wired as a triode. A 1990 tube designer, however, is pretty much forced into using the commonly available tubes.
TdP: You can always produce a one-off product with whatever you wish. You can produce a one-off using Western Electric 300Bs but you can't produce it in volume. Unless you use something that you know you can buy in five or ten years' time, it is really foolhardy to try to tread that path.
JA: This being the beginning of the '90s, will we still be playing LPs at the beginning of the next century?
TdP: I hope so.
JA: Do you think there will still be plenty of tube-amp manufacturers around?
TdP: Yes. We haven't seen the end of tubes yet. I mean, they were developing quite miniature tubes in the late 1960s for hearing aids, and the British Aircraft Corporation in England used micro-miniature tubes in their early missiles. It's just that when transistor technology came along, tubes got pushed to one side. In fact, a re-analysis of the tube is happening in computers. They're developing chips with cold-cathode tubes.
JA: Each chip carries many little microtubes, each about 20µm [half a thou'] across...
TdP: ...which operate on very low voltages. It can only get better. If we had devoted as much attention to tubes in the last 15 years as has been devoted to semiconductors, I think tubes would have been in a much more interesting state today.
Footnote 4: EAR's audiophile products are distributed in the US by EAR USA, Inc., 1087 East Ridgewood Street, Long Beach, CA 90807. Tel: (562) 422-4747. Web: www.ear-usa.com.
Footnote 5: My 1966 edition of the Allied Electronics Data Book lists the 6267 as the US equivalent.