Introduction
My recent descent into the madness of outboard digital RedBook transports gave rise to the obvious question: What technical parameters contribute to the differences in sound between transports? If there was one man who could answer said question, it would have to be Pierre Lurné of Audiomeca. His firm is the only audio boutique operation to ever attempt the in-house design of a RedBook transport from scratch. Audiomeca's digital creations are famous for not using over-the-counter transports, not even one as dear and well-regarded as the famous diecast Philips PRO2. The only thing good enough for the Mephisto IIX player and transport? Pierre's own disc-spinning mechanism.

An e-mail introduction from your pesky audio journalist sent out a probe. Would Monsieur Lurné be willing to shed some light on these matters? I was advised that due to the necessary hard separation between analogue and digital -- or work and home in his case -- the Lurné residence lacked a computer unlike the factory. Would I mind transcribing handwritten notes assembled after hours in his home that is surrounded by the old roses he and his wife enjoy cultivating? If not, I should expect some 20 pages in the mail in a few weeks. Could I then apply my questionable Teutonic grasp of the English language to his words? I might need to remove French pecularities that had crept into his text. That certainly was the very least I could do to repay Pierre's kindness. What follows is the outcome of this cooperation: French ingenuity, German labor - there's worse recipes for success...


"Why do CD transports sound different if all they do is read 1s and 0s? Audiophiles agree that transports sound different from one another but for non-engineers, it's hard to fathom what exactly is responsible for those differences."


Basically, a CD transport is made up of a CD mechanism [CDM] and a dedicated electronic servo. This system must optically track the disc, extract the digital data and send them to the DAC stage for conversion into an analog signal. Despite these being the "all-electronic" times, there are only a handful of CDMs commercially available. Even famous companies have little choice but to use ordinary mechanisms produced in large quantities by volume vendors. As a simple comparison with older CDMs proves, modern variants have not improved but are getting worse. The usual objective of our industry at large is low cost and high profit. This goal mandates technical simplifications. Those lead to poor quality but high quantities. As a rule, the associated electronic circuits correct for the majority of mechanical problems. Alas, nobody knows exactly what's inside those servos and dedicated drive/correction chips, where they perform well and where they fail. In the end, peripheral cures for innate problems can only be second-best. A true from-the-ground and fully optimized CDM of uncompromised performance does not exist to serve the luxury, audiophile and high-performance professional markets. Those firms who design digital products for these markets must instead rely on inferior parts of restricted performance which, if you think about it, is particularly harmful. After all, the CDM sits at the very source of a digital system, prior even to a DAC.
Audiomeca's goal is one of musical excellence regardless of cost. This means extreme attention to detail from the very beginning of the audiophile chain. That goes counter to the perception that if there was a need for better transports, their effect on sonic quality would be hard to quantify. "Perfect Sound Forever" and Shannon's Theory explain that digital quality does not matter. As long as 1s and 0s are properly distinguished, the digital signal can be reconstructed precisely. Contrary to the analogue medium of vinyl or tape, the famous "bits is bits" motto seems to proclaim that the digital medium is impervious to treatments, transfers, storage or defects. Regardless, audiophiles continue to say that "every transport sounds different". Presently, the primary reason has been identified as jitter. No doubt the future will elaborate on this theme. Time always adds new discoveries.
Audiomeca Model 2.0 used in Mephisto IIX


The plague of jitter causes time-domain perturbations. Virtually everything creates jitter. For example, the light refracting from adjacent tracks creates parasitic jitter even before the laser beam ever reaches the diode receptors. This jitter phenomenon is still not fully understood. A well-informed technician could counter that the data buffer (basically a reclocked memory) manages these problems satisfactorily. That's wrong. The bandwidth of the buffer is too limited. And why be so concerned about error correction after the fact? Why not avoid errors in the first place? Unfortunately, superior mechanics cost a lot more than the cleverest of circuits - hence the proliferation of reclocking DACs, DSP interpolation etc. Needless to say, if the information is already corrupted, all that sophisticated number crunching comes too late.


Let's suppose that jitter can be lowered to ridiculous nano levels or even virtually eliminated. The Swiss company Anagram does this very well. Shouldn't this make for perfection? It should but it doesn't. Zero jitter would be ideal if the data itself wasn't already corrupted. The ideal servo only works on incoming data. Ditto for the ideal DAC. It works only on incoming data to convert what it receives. Both are incapable of recreating what they didn't receive to begin with - unless you used magic. Back to the real world where this kind of magic doesn't exist. When the incoming data is flawless and correctly processed, we have maximum quality (though in practice not total perfection) at the output. When finger prints or scratches prevent complete capture of all data, corrections are performed which at the very least adds more jitter. Should the encoded data be too badly damaged, it invokes a deeper layer of error correction that inserts a line where there was a gap of information. Fortunately, this second degree of error correction is rarely prompted.
Remember that the original signal undergoes a variety of digital treatments and transformations between the microphone and DAC recorder. At the input of the first transition and at everyone that follows it all the way into the pressing plant and back to your CD player, there are data correction/manipulations. Each treatment step relies on what it receives from the one preceding it. If the data is corrupted at the input, it will remain corrupted at the output no matter what happens in-between. Still, Shannon's Law is respected as an unalterable truism. Alas, reality is more complex. How and why can the data be corrupted? First off, mathematicians have already demonstrated that there are exceptions to Shannon's Theory although those are impossible to explain in simple terms. Just imagine what happens when a high-frequency signal is sampled at the exact moment of crossing the zero axis: No signal!


Electronics engineers know that integrated circuits/chips suffer from specific problems and limitations. That includes servo chips and DAC chips. Can we really believe that the rapid stream of 1s and 0s beneath the laser is unaffected by vibrations, by parasitic light refractions, by endless processing and calculations, by encoding errors during the mastering and pressing and duplication stages? What portion exactly makes up the music signal? What data is really servo instructions and decoding flags? In short, this is what we call the corruption process.


We shall see why a CDM generates its own errors and why servo corrections cannot be perfect. But now you understand that well before the signal ever gets to the laser reader, errors have already been introduced. How much more degradation on the way to the preamp should we tolerate? The simple fact that servos, buffers and anti-jitter mechanisms are necessary proves that the mechanical, electrical and optical aspects are flawed. Why would they require their own correction mechanisms otherwise?


One can appreciate that even little improvements at every step along the way become accumulative and thus audible. The servo relaxes. Jitter is reduced. Data remains respected. After all, the strategic position of a CDM mirrors that of a turntable. At the very source of the chain and unlike a musical instrument, it must remain utterly neutral and track the disc without additions or omissions. With a turntable, neutrality walks a fine line between pleasantly low colorations and overdamping and becomes the core trait of a proper source for a complete sound reproduction system. [Excerpted from Audiomeca's J1 turntable brochure 1989 and retold by Geoff Husband of TNT Audio.]
I designed my first turntable and pickup arm in 1968. Eleven years later, I launched Audiomeca to manufacture high end products for my own label and for other brands as an OEM supplier. In the early 90s, we had to switch to digital but our analogue products remain in our catalogue to this day. I often say that "Digital is a passion but my true heart belongs to Analogue". After 25 years in the business, I'm told my reputation is one of paying meticulous attention to every detail of design while scrupulously following the laws of Physics and -- I hope -- providing good customer service. Naturally, audiophile opinion about the subjective performance of Audiomeca products will diverge. But I believe nobody could seriously criticize our design integrity or the applied engineering we garnered from university schooling and a lifetime of experience.
44mm platter from Audiomeca
30mm standard platter


25 years after launching the company, we're still working on better solutions. I believe this speaks for itself in an industry where companies come and go like butterflies. Though I had long considered work on my own dedicated CDM based on my experience designing turntables, I did not want to begin such a project without a true man of the art. Though it took a long time to identify the right person and then convince him to work with us as a consultant, we had the good fortune of working with Monsieur Rene Boonen from Philips Hasselt in Belgium who spent 10 years designing the CDM4 and CDM9. The latter is considered by many to be the best industrial CD drive mechanism ever made. It was at the heart of our first transport, the Mephisto 1 (since discontinued).


After 5 years of cooperation and with many thanks to Rene for his expertise, patience and friendship, we built a large prototype to conduct lengthy measurements that would help us understand the mechanical and optical phenomena associated with such a device. The first commercial iteration of this prototype became the Audiomeca High Precision Read-Out Mechanism Model 1.0 and was used in the Mephisto II CD player and transport. Today's Model 2.0 provides the mechanical core of the upgraded Mephisto IIX. Other drive models including DVD drives will follow. When we remember that one human hair is thick enough to be imprinted with 50 CD tracks, it becomes obvious that even so-called micro vibrations become a limiting factor of mechanisms that include moving parts. After all, anything mechanical that moves creates vibrations. With the minuscule size of encoded CD data, even barely measurable vibrations disturb proper tracking and place high demands on the corrective servo.


Let's follow the data from the Philips/Sony RedBook CD through the transport to its digital output and consider a few select examples at every stage. For more than 25 years, the so-called RedBook has been the official specification which defines the CD standard. Quoting from The Absolute Sound, issue Spring 1986 and as written by Philip Greenspun and Charles F. Stromeyer III, "Awai's president, Heitaro Nakajima and one of the original proponents of the CD while a vice-president of Sony, put it best in a letter explaining why such a low sampling rate was chosen for the CD: "We had to decide on the cutoff point for the sampling frequency, and we also had to take economic and production factors into consideration. The CD system was not designed as the ultimate level sound reproduction medium. However, this does not mean that we feel any need to change the present standard.""

To this frank and reliable declaration that is born out by millions of RedBook CDs on today's retail shelves, the same authors added: "One of the problems with CD software is the indifference of consumers and record companies to sound quality. In the early days of high-fidelity sound, companies like RCA and Mercury took great care in making realistic recordings to satisfy a demanding public. As reproduction technology has become less expensive, the quality of the records from the big companies has fallen dramatically. When the record companies began marketing cassettes, it was expected that the product would be used in car and portable stereos. Consequently, cassettes are normally duplicated at high speed (x 64 over real time) and have severely rolled-off high frequencies. To everyone's surprise, more than 50% of classical music recordings sold in 1984 were purchased on low-quality cassettes. This has convinced record company executives that most consumers are not interested in sound quality.