Interviews

Ed Meitner is one of those rare individuals who charts his own course in audio product design. From his platterless turntable of the mid-1980s to his new Intelligent Digital Audio Translator (IDAT, reviewed elsewhere in this issue), Ed Meitner's products have been distinguished by original thinking and innovative engineering. Although not all his designs have been commercially successful, in each he has attempted to advance the state of the art by rethinking fundamental principles.

Ed is also pursuing an ambitious project that would radically change the way recordings are made. It began when he recorded an electric guitar through a 10" guitar-amp loudspeaker and was dismayed that it was impossible to even come close to capturing and realistically reproducing this apparently simple sound through another 10" speaker. This experience launched his investigation into why reproduced sound is never mistaken for live music, a quest that may result in a radically new recording technique.

As the designer of the LIM Detector, the jitter measurement instrument used in the research for January's "The Jitter Game," Ed is also at the forefront of quantifying and analyzing jitter in digital processors.

During a visit to Museatex Audio, where Ed is principal engineer, I talked with him about high-end product design, music reproduction, and the future of audio. I began by asking him how he got involved in audio and electronic design...

Ed Meitner: I started off with a company called Olive Electrodynamics, making recording-studio consoles with Wayne Jones—the early days of recording-studio consoles. We were the first ones with VCA [Voltage Controlled Amplifier] controls in consoles—this was 1972—and with logic switching circuits. That's how I got started in North America. Then I was involved in Amber, the test-equipment manufacturer. Then I joined Museatex.

Robert Harley: How did you make the jump from test-equipment design to high-end audio?

Meitner: The test-equipment customers would call me up continuously, asking, "Should we do the circuit this way or that way?" So I said I might as well do it myself.

I also saw a perpetuation of old ideas. You get out the RCA tube manual and look at a few schematics, add a volume control, put in a few switches, and plug it together. It wasn't doing anything where you could say you advanced the state of the art just a little bit.

And having had all this previous experience with VCA switching and volume controls, I figured it might be a good idea to have a product with a remote volume control and be able to ship it to Los Angeles and have it work two years later. We all know what happens to products in corrosive environments. That was the beginning of the first preamp.

Harley: Your focus now seems to be on digital audio. Do you spend most of your time on digital?

Meitner: The LIM Detector [the jitter measurement instrument] is very much analog, even though it has some digital chips in it. I think that analog and digital are very closely related. If we look at power-supply rejection, slope modulation, Phase Locked Loops—where is the dividing line between digital and analog? Unless you're strictly doing software, you're doing digital. But if you're doing hardware, you find out very quickly that you're doing both analog and digital.

So it's not a switch. And the complexity of the analog circuitry within a digital audio system is obviously getting smaller. So it looks like you're doing more digital but you're really not.

The nice thing about digital is that it has shown us what we really need on the analog side. We know we don't need amplifiers with 100MHz bandwidth and circuits that go into infinity, and all sorts of ridiculous stuff like 5000V/µs slew rates. We have a clear target now for analog. It makes it more fun; it makes it more rational. But the design approach to digital isn't very different from analog.

Harley: You seem to be one of the leading audio designers addressing the subject of word-clock jitter in digital products. How significant is a product's jitter performance to its overall sound?

Meitner: I think it is one of the very, very significant parts, if analyzed properly. But just as a random term, it's meaningless. We could have a jitter figure that's relatively high and the system would sound very good. But if that jitter isn't analyzed, we don't know the frequency components of that jitter and we have no idea what we're dealing with.

My involvement with jitter started when I identified it as a problem. But I found out I had to build my own test equipment to measure it, think about how it gets generated, and how to get rid of it. This will be a lifelong involvement for me.

Harley: What you're saying is that it's the frequency distribution of the jitter that matters, not the overall amount. A single jitter figure doesn't tell the whole story.

Meitner: Absolutely not. It may be your target, but it doesn't tell you much at all. If the jitter is in the MHz region, it is absolutely meaningless to the audio components. If it happens in the hundreds or thousands of Hertz in the audio bandwidth, then it is a serious problem. Conventional test equipment will not tell you where the jitter is and how it's correlated.

Harley: This led you to develop the LIM Detector.

Meitner: That's correct. It reveals numerous problems and effects that otherwise one could only guess at.

Harley: And C-Lock, your jitter-reduction circuits, were developed in response to what you saw with the LIM Detector?

Meitner: C-Lock was simply a way to eliminate part of the jitter problem. If properly applied within the system, it can eliminate a lot of it.

Harley: What are the audible effects of jitter?

Meitner: I hear it as noise modulation on single sonic events. On a massed presentation I hear it as a lack of depth, as a lack of impact and dynamic contrast. On low-level passages I hear it as noise. It also produces a restriction of left-to-right image width. Overall, it's a smaller picture, a more constricted picture that, instead of fading into black, fades into some kind of grey. Over time I get a headache and don't want to listen anymore. That's what I hear.

When you look at what jitter does to the audio, it does most to the lowest frequencies and least to the highest frequencies. Because you accumulate many more sample points, the sensitivity to jitter is greater at low frequencies. It's wrong to look at the high-frequency part of the spectrum for trouble.

Harley: Do you think the effects of jitter are the primary reason many listeners prefer analog to digital?

Meitner: Probably. It's best assessed over a long period of time. Do you want to listen more or do you want to turn it off? Fatigue has a lot to do with it.

Harley: You've said that if a digital processor had a perfect input receiver—no jitter—the processor would be completely immune to audible differences between transports, interconnects, and other factors that now affect CD playback quality.

Meitner: That's correct. There may be some problems that are transport-related—crystal oscillator instability or vibration-induced problems that generate low-frequency events that are difficult to filter out after the fact. This is one of the reasons we made our C-Lock circuit in the transport as well as in the receiver—to make sure both ends are covered.

Transport-related jitter is particularly nasty because it produces a lot of asynchronous events that have nothing to do with audio. You get your tracking-servo information, focus information, servo-motor speed control, and LSI-related jitter components coming out of a transport. They are totally non-correlated with the audio, so they stick out during playback. I've heard numerous times that cleaning up the drive [transport] is 80% of cleaning up the system. And having seen the effects of jitter, it's probably true.