Lucas van der Mee

Apogee Electronics recently celebrated 20 years in the business of digital audio. These days, 20 years in business is quite a long time. The following is a conversation with Apogee's lead analog designer, Lucas van der Mee. We talked about what makes a good converter, the future of digital audio and how Apogee's philosophy is making digital converts even out of some of the most hardcore analog studio engineers.

I know that you've worn a few different hats in the music biz. Tell us about that and how your previous experience led you to Apogee and influenced your current work.

Well, obviously I always had a great fascination for music and electronics — permanently looking for improving audio quality, especially making sound reproduction more musical. What my live sound experience taught me is the "no bullshit" approach. Stuff has to work under any condition at any time. So, making an occasional compromise in favor of compatibility and reliability is in the end, a much better decision. My studio days in Holland at Wisseloord were the schooling days of working and dealing with digital audio, which was at its infant stage in those days. I learned through practical experience and studying the books what digital audio could and could not do. The reason why I was attracted to it was the impulse response, the transients. I could never get that from analog. I was always disappointed how the drums came back from analog tape. You know, you set up your console, you feel every beat, it really hits your ear — and then you hear it back from tape and it is all muffled. Unlike many other engineers (US engineers especially), I loved and preferred digital and took the artifacts (harsher sound, thinner bottom end) for granted. Bob Clearmountain [ Tape Op #84 & #129 ] introduced me to Apogee when he was working in Holland and was I assisting him. His Apogee AD-500 broke down and I fixed it for him. So that was my first contact. I really liked the converter and its philosophy. It was all about being portable and sounding good for rock and pop music, unlike the DCS and Prism converters, which I was already familiar with. They were too "clean", too sterile to my taste. Talking to the people at Apogee for tech support and acquiring equipment for our new mastering studio established a relationship and I started to do some part time work for them, like AES shows. Next, I decided to move to Los Angeles to start working for Ocean Way. They hired me for my digital expertise and I could not have made a better choice. For the first time I could hear what real good analog equipment sounded like. The ATR-124 was a killer, analog tape multitrack that not only had great warmth and bottom end, it had the transients too. Talking to many engineers who had sworn by analog, I now had a much better idea of what they did not like about digital. This knowledge piqued my interest in converters, recognizing that they were the weakest link in the digital chain. During my Ocean Way days I still worked part time for Apogee, and my first project on the side was an attempt to improve the sound of the AD-8000 (make it more analog). That became the SE. When Alan Sides (Ocean Way's owner) decided to sell half of the studios on Sunset Boulevard, I took that as a sign for a change of career. After many requests from Apogee, I finally agreed to become a full time design engineer. I felt I was ready at that point. So to answer the question in a short way, live mixing taught me that things need to work. Wisseloord studios in Holland showed me the power of digital audio and my experience at Ocean Way revealed how good analog can sound. My mission is to combine these three.

That certainly is the magic that fans of Apogee have come to know and love. But what I wonder is, at the heart of your boxes is an integrated circuit made by one of a handful of companies. Any of your competitors can go out and use the same chip. So how do you add that Apogee touch? How did you address the things that Ocean Way engineers disliked about digital?

Have two chefs make tomato soup and you are guaranteed that the dishes will taste different. Part of the secret is the ingredients, of course, but way more important are the quantities and the way it is prepared. A bit too much salt, or a little too much heatand the soup is spoiled. Designing a converter is not much different. You pick a part as the major ingredient and then your experience and your creativity decides what will be around it. The sound of a converter is not only determined by the converter chip, the analog circuitry around it is just as important. What kind of analog filtering is used, what types of op amps are implemented, what the power supply is like, what are the clocks driving the chip like, the PCB layout — all these factors affect the sonics and (to a certain extent) the specs. Over time I have learned to know the sound and behavior of many active and passive parts. So, besides trying to design the proper circuitry for the application from a technical point of view, I am also making artistic decisions of what would be the best sounding solution for that specific design. If you open up our units, it is possible you'll find an array of different op amps in the same audio path. Yet in another design, I could stick to one type. There is no particular rule or method — it differs for every new design. I should add that this is also the result of the continuous redesigning I do for every new product. I am always trying to find new inroads, improvements. I tend to recycle very little from old designs and I think it shows. We are getting better and better and more cost effective at the same time. The major complaint of engineers who prefer analog over digital is that digital is lacking in the lower midrange (no energy) and the widened (flat) stereo image, with a virtual "hole" in the center. Harshness was a very famous, other complaint, of course. But that has been pretty much addressed by any quality converter company. The former complaints, however, (no energy, flat, etc.) are a different story. I feel Apogee is one of the few companies to address these issues very successfully. Our current line has that same analog feel as tape. A lot of the guys I talked to in my Ocean Way days now use my converters and love them — some even prefer them over tape. How cool is that?

Wow! That must be the ultimate compliment. One thing you didn't mention is dither. Apogee uses aproprietary dither algorithm calledUV22. What's the story behind that?

UV22 has been available for over ten years now. It has seen one major revision about five years ago, in which a few things have been optimized and improved. This version has been labeled UV22HR. The story behind it goes way back in time, before I worked for Apogee. The major brain behind it is Jerry Goodwin, a professor in mathematics in Berkeley. The concept was born to avoid the solid noise floor that normal (triangular) dithering or noise-shaped dithering adds to data. It basically smears or conceals the information at the bottom of the dynamic range. UV22 is a different approach — instead of the full bandwidth noise it uses a concentration, or clump, of noise around 22 kHz. The purpose is to modulate the lowest bits instead of filling them. It is sort of comparable to the principle of using bias on tape to eliminate the non-linear behavior of the magnetic coating. Kind of anecdotal is the story how UV22HR came alive. The whole art of this technique is to create a noise that is a 100% random within that narrow range — no repetition allowed, since that will create idle tones. In order to create this perfect, "controlled" randomness we had every personal computer in the office run sequences at night for about half a year. All the results were being collected at the end of the week and sent to Jerry to analyze.

Crazy! So what did Jerry do with all that data?

He had written some applications to analyze the data. Probability analysis was, of course, his major interest.

Besides dither, another important factor in converter performance is jitter. What kind of artifacts will you hear in your audio if you have bad clock jitter? How has Apogee tackled the issue?

Jitter is very crucial for converter performance. It can make or break a converter. The whole concept of digital audio is based on amplitude versus time. So you want both to be as accurate as possible. Amplitude is determined by the linearity of the chip, time by the system clock. Jitter is something you continuously have to fight and reduce. Every transmission line, trace, cable, adds jitter. There is no such thing as "no jitter" as some companies claim. That is the same as saying that a certain analog signal does not have noise. It is always there. The only things you can do are to keep it low and reduce it, making sure that what remains is outside the audible range. There are several tricks, besides keeping your clock paths clean and as free of interference as possible. A high quality, Phase Locked Loop [PLL] is the first requirement for clocking to an external device. We have spent a lot of time and effort in designing the optimal PLL. The C777, used in the Big Ben and the AD-16X and DA-16X converters, is the latest development. That is a PLL with the low jitter performance of a crystal, but with the flexibility of a VCO-based PLL. In normal speak — it follows everything. Which is very important because the realities of the outside world are not perfect and that is what we need to deal with. Another thing to avoid is beating clocks within the same unit. The results of that can be really nasty. In our design everything is synchronous. Which is also the reason why we do not make a converter card for inside the computer — it is full of beating clocks. Some of the audible results of jitter are a flatter soundstage, a certain fuzziness in the top end and a less pronounced center — in other words, everything we have come to hate about digital audio. So the goal needs to be to attack it vigorously and make it as inaudible as possible.

Indeed... That's interesting about the beating clocks thing. I had never thought of that. Another thing we hate about digital audio is that when you clip it, it does not sound friendly, whereas with analog, most people actually like the natural compression you get when you slam the tape a little (or a lot!). Some Apogee products have a 'soft limit'. Is this just to reduce the chance of getting overs, or is it designed to simulate the way tape behaves?

Well, it is a bit of both. The problem with digital recording, and this was especially true in the 16 bit days, is that you want your recording levels as hot as possible to keep distortion low. For most signals that is no problem, but when the material has a very wide, dynamic range, like percussion for instance, it is really difficult to get a proper level. The peak information is so much higher than the average level. So, some sort of an instant limiting would be welcome. The concept of soft limiting is based on the experience we have with recording to analog tape. Tape, as we know, is way more forgiving to overs than digital. In digital, an over is hard clipping, while tape approaching the maximum recording levels results in saturation of the magnetic material. This creates a much gentler type of distortion to our ears. It basically rounds it off instead of clipping it. That is the specific phenomenon we try to simulate with soft limit.

I see. Now, on the subject of 16 bit versus 24 bit — how many bits are we really able to achieve these days? Do you see that improving? Are 32 bit converters anywhere on the horizon?

When it comes to single chip solutions, we see dynamic ranges of 120 dB and over, for the best ones, which translates to roughly 20 to 21 bits — pretty close to where we want to be. Keep in mind that most analog circuitry does not have a wider dynamic range than that. The dynamic range of an orchestra is about 90 dB — our ears can handle levels up to 135 dB spl (which is the pain threshold). So if we manage to make true 24 bit converters, we have pretty much covered the dynamic range we need. Any higher bit rates only make sense for math (like processing, digital mixers). A 32 bit converter would have a dynamic range of 192 dB. That means the ratio between the lowest and the highest signal level in volts would be, rounded off, 4 billion. So, say your lowest level is one microvolt — in that case you will need power rails of 4000 volts to cover the entire range. Ouch.

Ouch indeed. I'm not sure that converter would pass UL. What about sampling rates? We're at 192 kHz right now, about five times the theoretical rate needed to encode 20 kHz (in a perfect world). Is there any point in going higher?

I would like to answer this one wearing three different hats: The technical engineer: Personally I have not been convinced that we needed to go to 96 k. To restrict the bandwidth to 22 or 24 kHz may have been a little too tight, so a sample rate of 60 kHz would have been ideal. Unfortunately, it is too late for 60 kHz now. I believe the extended bandwidth does not really make that much of a difference, it has more to do with anomalies of the brickwall filter as required for reducing aliasing noise. By moving the sample rate up, we can move up that filter and so its side effects will be less audible. I think that is what is being perceived as the audible improvement. To my experience, in our converters the difference is negligible, as we are pretty good at filter design. But it does make sense to run at a higher sample rate for digital processing. Digital EQs can sound better at a higher sample rate.

The recording engineer: Do we really need the extended bandwidth? The art of recording is to make something that sounds good within the limitations of the medium. None of the transducers are perfect — a microphone is imperfect, a speaker is imperfect, a converter is imperfect, etc. We end up with something that attempts to replicate an image of reality, but in reality, does not come close at all. Anyone can immediately hear the difference between an instrument and the best possible recording of that same instrument. So, we have to accept that the medium is not perfect and therefore it is permissible to make something sound as best as possible on that medium, while accepting the limitations. Recording engineers have done that for decades. Just like a good photographer, the best pictures are made understanding the art of leaving things out. It is better to not capture everything, it is more important to get the main object in full focus. For instance, leaving the extremes of the bandwidth out is a trick many engineers use. You really don't want 50 Hz on a vocal, and the same is true on the top end.

The philosopher: I have learned there are a lot of things we don't know. I have found myself, many times, saying that certain experiences cannot happen because I could not explain them from a technical point of view. The first responses during the development of the Big Ben master clock come to mind. The reports I received of an improved soundstage sounded like nonsense to me until I heard them myself. And it was not subtle. I had to go back and find out why this happened instead of trying to deny it. That is a bit of a problem that comes with the job. Technical engineers have to trust in logic and reason, which can make them a bit stubborn, hard to convince — in some cases they start to believe they are godlike in a sense, they think they know everything. Thanks to my hands-on experience, I know that there is more than just logic. Our ears do not lie — our perception is actually really good. The things an experienced audio engineer can hear can never be discounted, even if it is not explainable. The task for a technical engineer is to find out why it is perceived and what are the main contributors. We are continuously learning new things, so we may not know exactly why certain credible people do prefer 192 k. There is still a lot of work to be done.

That's fascinating. Your philosophical answer reminds me of the now late, great Robert Moog, whom I know you are a big fan and collector of. He had often spoken about unexplained, yet very real connections and experiences people have had with machines. Has Moog been an influence and/or inspiration on your work in any way?

Yes, in an indirect way. I was 12 years old when I saw part of an Emerson, Lake and Palmer concert on TV. I remember that I did not quite get the music, I was too young, but the sounds Keith retrieved from that huge telephone switchboard blew me away. My future was suddenly clear to me. I wanted to know how that thing worked, down to its innermost part, and I wanted to be able to play it. I saw the perfect symbiosis of my curiosity of electronics and my love for music. So, I got myself lots of books on the theory of electronics, read literally everything I could find on synthesizers and took piano lessons. At age 15, I knew every knob, every function of the Minimoog, without having ever touched one. They were way too expensive and I did not know anyone who had one. All this also created the interest for recording, and everything else that had to do with technology and music. My interest in synthesizers declined dramatically with the preprogrammed keyboards like the DX-7s and sampler keyboards. They were just "sound jukeboxes" to me. Not only did I like the sound of the Moogs, I also liked the esthetics — the very nice woodwork, his choice of knobs and layout, etc. His passion for Theremins I found very interesting too. He just kept on trying to improve this "impossible" instrument, catering to a very small audience. I like that kind of quirkiness. Bob Moog definitely put a stamp on the music of the last four decades, not so much by being an innovator, but by being a very good listener to what people and musicians want and need. That quality I am really taking to heart.