with spurious noise, again making the zero crossings less defined with respect to both amplitude and time. External vibrations, or a warped CD, or wow and flutter in the rotational transport can make the zero crossings jitter or wander in time from the exact periodicity they're supposed to have.
      There's abundant anecdotal evidence that the quality of this analog input signal from the CD, of this the eye pattern, does indeed affect the ultimate sound heard from your CD player. Let's now discuss several different kinds of evidence, probing into the reasons why the eye pattern quality could make a sonic difference.

-- Fluid Treatments and Cleaners

      There are many fluid tweaks for CDs on the market, and virtually all of them make a significant sonic difference (not necessarily wholly for the better, but the remarkable point here is that they do make any difference at all) - and the only thing they can be doing is altering the quality of the analog input signal eye pattern. These tweaks include felt pens for the CD rim, and cleaners and polishers for the main CD reading surface. These CD treatments leave little or no added mass behind, so they cannot be significantly altering the servo current draw transient factor discussed above. All that they can be altering are the reflectivity characteristics of the CD, which of course would alter the characteristics of the analog eye pattern waveform.
      Most of these fluid tweaks that we have evaluated are not wholly positive in the sonic changes they wreak. That is, the sound gets better in some respects, but worse in others. Typically, when we try treating a new CD with these fluid tweaks, we find that the midrange gets better (more musically natural, with better inner detail) but the treble gets worse (fuzzier, sometimes dirty, and defocused, with poorer articulation). But for now the point is that virtually all the fluid tweaks do indeed make a difference, which demonstrates that changing the analog eye pattern input signal does indeed change the ultimate sonic output from your CD player.
      In our lab research, we have been probing the limits of CD resolution, trying to extract the ultimate in music information from the 16 bit CD format. We have discovered further evidence that the quality of the eye pattern does indeed make a sonic difference. Some examples are worth discussing here, because they illustrate: the surprising amount of music that can be extracted from an ordinary 16 bit CD; the surprising vulnerability of the digital format to incredibly subtle analog influences; and the incredible resolving power of human hearing (especially when teamed with a very high resolution system like our lab research system).
      First, we researched fluid treatment and cleaning techniques for brand new CDs, and we developed a technique that gave us incredible sound, sound in a whole higher league than we had ever heard from CD before (even brand new CDs may still have mold release compounds or other contaminants on their surface). Interestingly, the sonic performance of various techniques correlated very well with the final reflectivity they left on the CD. We independently compared various fluid techniques by gauging the raw, gross reflectivity of the CD after treatment with each technique. Some fluid techniques made the CD a much brighter reflector of light (you can use sunlight) than other techniques (the brightest reflectivity also implied the best cleaning of the CD). And then it turned out that the techniques yielding the brightest reflecting CD also yielded the best sound.
      Conversely, we found that those fluid treatments which left behind some sort of deposited film (e.g. waxes or polishes) degraded some aspects of the sound -- in spite of the supposedly magic properties that this deposited film had, to purportedly benefit CD sound. These deposited films reduced disc reflectivity, compared to an ultra clean disc, and they also worsened a new CD's sound in some ways (though, in some other sonic aspects, some changes they wrought were improvements). In other words, the only way, for a fluid treatment to yield truly better CD sound without concomitant degradation, was to clean the disc very well and then leave nothing behind. This maximized disc reflectivity and also yielded the best sound of all, with no sonic downsides.
      This correlation, between brighter reflectivity and better sound, strongly suggests that the quality of the eye pattern is indeed an audible factor in the ultimate signal put out by a CD player. Also, slight improvements in reflectivity produced major improvements in sound, strongly suggesting that the CD medium is very vulnerable and sensitive to this external analog influence of reflectivity. The cleaning technique that produced the best reflectivity also produced the best sound, and it was stunningly better than we had ever heard before form the 16 bit CD medium - more transparent, more musically natural, more vividly sparkling, and even more dynamic!
      Unfortunately, the great sound did not last long. After about 2 minutes of playing, the sound from the freshly cleaned CD started deteriorating, back toward its normal self, losing color (becoming grayer), more veiled, less dynamic. Why? We soon found the problem. The freshly cleaned CD, as soon as we started playing it, became a piece of plastic rotating fast enough to attract very small particles of normal room dust, which we could see upon closely examining the CD. There was enough room dust in the air inside the CD player so that, after about 2 minutes, the fast rotating plastic disc had attracted enough to noticeably decrease its reflectivity, and this reflectivity decrease had produced a degradation of sonic quality. Although this was a sonically disappointing outcome, it constituted another scientifically valuable piece of evidence. The sound had improved when we got the CD super clean and maximized its reflectivity, and now the sound was deteriorating when attracted dust decreased the reflectivity. Further scientific confirmation that the quality of the eye pattern is sonically important.
      It's a little amusing and tragic to pause and realize how vulnerable the supposedly bulletproof digital medium really is, and how little we have achieved despite years of technological innovation. In the old days when vinyl LP was the only medium, dust was a serious concern. Most of you remember how the sound would get gradually fuzzier and dirtier as a fuzzball built up on your stylus during the playing of one LP side, even if you had meticulously cleaned both stylus and disc at the beginning of each side. In the beginning, we fought this by putting clear plastic covers over our turntables to keep dust from settling while playing records - that is until we discovered that these covers with their large surface area resonated with vibrations (including air vibrations picked up from the loudspeakers), thereby degrading the sound right out of the starting gate, even before any dust settled. Then we tried dust bugs that would clean the LP while it played - until we discovered that these additional tracking devices made wow and flutter worse, thereby also degrading the sound even before any dust settled. In the end, most of us just learned to live with an open turntable and settling dust. But now, we see that the digital CD is also sensitively vulnerable to dust. And it spins about 10 times faster than an LP, so it attracts a lot more dust. Perhaps the next step, for ultimate CD playback, will be a CD player in a hermetically sealed box, with air filters, and a special remote pump that does not introduce foreign mechanical vibrations into the CD player environment. Obviously, it's helpful to scrupulously clean each CD just before each playback, just as you used to do with LPs. You might also try a static discharge gun on CDs, after cleaning and before putting the CD into your player.
      Incidentally, in our research we also found that the second best sounding cleaning technique did not attract as much dust, and thus gave us better sound over the longer run. This second best technique employs the Kodak lens cleaning fluid sold in the UK (apparently, the same fluid is not available in the USA from Kodak due to environmental concerns). This Kodak cleaning fluid is also available worldwide through Trackmate (based in Ireland), who package it in a convenient felt tip applicator as their CD cleaning fluid.

-- Play It Again, Sam

      Another example we discovered in our research also demonstrates how extraordinarily sensitive and vulnerable the sound of the supposedly robust CD digital medium is to the quality of the eye pattern, and how its actual fragility is again similar to the analog LP. We found that when the same section of CD's musical track was played a second time, shortly after the first playing, it does not sound as good the second time (or the third, fourth, etc.). The subtle sonic degradation is similar to that we heard with the lower reflectivity of inferior cleaning techniques: less color, grayer sound, less transparency, veiling, loss of vivid sparkle, and poorer dynamics (especially on attack transients). So a reasonable inference is that the metalized layer of the CD does not reflect quite as well if it is impinged with the laser a second time within a short period after the first playing, thereby producing a slightly inferior eye pattern, which in turn produces slightly inferior sonic performance. This sonic degradation is noticeable only on a very high resolution system, and of course is most noticeable if the CD has been well cleaned to obtain maximum reflectivity and optimum sound for the first play through of a given track section.
      Over a period of time, whatever caused the degradation relaxes and stabilizes back to its original status, and then the previously played section of the CD is once again capable of its best sound. How long is this relaxation period? It seems to be a period of many minutes, perhaps an hour, though this is difficult to verify, given that human hearing memory is not perfectly accurate over a long span (and given that two instances of the same CD do not have the same sonic quality).
      Incidentally, we conducted this music repeating test within the first minute, so that the results would not be biased by settling dust. And then we also observed the same effect even after some dust had settled (though of course the sonic change was less dramatic). We would quickly repeat just a short musical section, within and at any phase of the dust settling problem (thereby rendering this phenomenon constant), and we could still observe the sonic degradation upon the second playing (this sonic degradation remained with subsequent repeats as well).
      Why should a CD's reflectivity be worse for a second playing of a given section soon after the first playing? In our everyday macro world, we colloquially think of a passive reflective surface as simply reflecting back the light that impinges upon it, and this model suggests that reflectivity should not change. But if we examine the situation more critically, at a close up atomic or molecular level, new possibilities for a plausible hypothesis emerge.
      First, consider just how light is reflected, especially by metals, which are generally the most effective reflectors of light. Metals have a special and unusual atomic structure, where the atoms share a common lattice rather than being bound into molecules, and where electrons are effectively shared among atoms. These shared electrons move so freely in the lattice among atoms that we call them free electrons. Free electrons make metals good conductors of electricity - and also in many cases good reflectors of light. Simply speaking, here's what happens. Some free electrons are located deep within the lattice of a metal, but others are flying around right at the surface we will be using as a reflector of light. The ones flying around right at the surface set up a strong negative electrical field right at the surface. Photons of light coming at this surface of the metal hit this electrical field set up by the free electrons near the surface, and they bounce off this electrical field, as if they were repelled by it. When photons of light bounce off a material's surface instead of penetrating into that material, we say that that material is reflecting the light.
      Thus, reflection by a metal depends on the free electrons at the surface. If anything happens to change the number or distribution of free electrons that are at the surface, or to change their characteristics or properties in some way, then the reflectivity of the metal will be altered (and probably degraded). What could happen to cause such changes? Well, for instance, these free electrons at the surface might be hit by a sudden, quick, strong burst of incoming light, say from a laser. If this first quick hit of strong light changes the characteristics of the surface free electrons in any way, then any subsequent hits of strong light might not reflected as effectively. Therefore, the first quick hit of strong light would receive the benefit of maximum reflection, thereby producing the best possible eye pattern quality, but subsequent hits of strong light hitting the same small area shortly thereafter (due to our replaying the same musical passage again) would not be reflected as strongly, thereby producing an inferior quality eye pattern and inferior sound from the same CD section.
      What characteristic of the surface free electrons might change, and how might this come about? Recall that the electrical field set up by the surface free electrons repelled the incoming light photons, and pushed at them to make them bounce back whence they came, effectively reversing their incoming momentum. But, as Newton noted, every action has an equal and opposite reaction. The electrical field from the surface free electrons pushed at the incoming photons, but the equal and opposite reaction is that the photons also pushed at the electrical field and thus at the surface free electrons. When the first quick hit of strong light comes along, from the laser quickly passing over that particular square nanometer of CD aluminum reflective surface, the surface free electrons are sitting at the surface there, able to generate maximum reflectivity by having the strongest electrical field. But that first strong hit of light that gets reflected also pushes back at the electrical field and at the surface free electrons, effectively pushing many of these free electrons away from the surface and down into the atomic lattice. Thus, if a second incoming hit of strong light comes at that same square nanometer of the CD (that same musical moment) shortly thereafter, there won't be as many free electrons at the surface, so the electrical field won't be as strong, so the reflectivity won't be as high, so the eye pattern quality won't be as good, so the same musical moment won't sound as good played the second time.
      Eventually, the free electrons in the metal lattice will redistribute themselves back to an equilibrium, whereby the full supply of free electrons will be at the surface once again, reflectivity will be at a maximum again, and that moment of music will again sound great when played the first time. How long might this redistribution take? It might take longer because the aluminum (or gold) is vapor deposited on the CD as a very thin film, which means that there is not a great reservoir of surplus free electrons deeper in the metal that could readily force their own way (or force other free electrons by massive repulsion) back to the surface.
      Furthermore, electron travel through a metal is actually a very slow process. The chances of any given electron getting very far are slim, even if we wait an eternity, because it bumps into or gets trapped by countless obstacles in its path. So instead, let's look at what we call electric current, which is the average drift caused by one electron bumping into the next, and causing that second electron to move onward, billiard ball style, say back toward the surface where it can restore the earlier maximum reflectivity. How fast does this billiard ball drift propagate through metal? Also very slowly. As we discussed in IAR issue 11, it takes about 17 hours for electric current, or the propagation of electron billiard ball impacts, to travel from your power amp to your loudspeaker. Thus, it might take considerable time, say many minutes or even an hour, for the free electron distribution to re-establish its earlier strength at the surface.
      Note that the phenomena we're discussing here are all transient in nature. If you measure the average reflectivity of aluminum continuously bombarded by strong light over a period of time, it will seem to be constant. But the transient reflectivity, when first hit by a strong light, might be higher for a split second. And the laser's light only hits a CD's pit edge for a split second, so it is the transient reflectivity that counts for the first time a given moment of music is played. If you then replay that same section of CD shortly thereafter, you're now dealing with the average reflectivity of aluminum, which might be lower.
      A second (additional) possibility is that the surface free electrons, being suddenly pushed back into the metal by the quick strong hit of light, cause a disturbance among the other electrons and the nuclei of the metal's lattice, and that this disturbance causes some of the electrons to drop down to a lower quantum energy level. If so, new photons of light would be emitted by the metal. Thus, the extra transient reflectivity of the metal that we hypothesize for the first quick hit of incoming light, might in part comprise new photons of light being emitted by the metal itself.
      A third possibility concerns the incoming photons of laser light that are absorbed by the metal instead of being reflected back. Aluminum and gold are, even at their best, only about 90% reflective (presumably this measurement is a long term average). This means that 10% of the incoming photons are not reflected, but instead penetrate into the metal layer. Once inside the metal, these photons would naturally cause disturbances within the metal's lattice, perhaps causing some of the metal's free electrons to drop down to a lower quantum energy level, thereby causing some photons to be emitted by the metal, thus raising its transient reflectivity at first. But, after the first quick hit of strong light lowers the energy level of these free electrons and causes photons to be emitted, then these electrons will already be at the lower quantum energy level when subsequent hits of strong light come in, and they won't want to go any lower, so there won't be as many electrons available to move to a lower energy level, and thus there won't be as many photons emitted.
      A fourth possibility is that the 10% of photons which do penetrate into the metal cause some sort of change in the internal energy levels and equilibria, such that the surface free electrons are simply no longer as effective in setting up that electric field to repel the next incoming hit of light. It might take many minutes or even an hour for the internal energy levels and/or equilibria to gradually pull themselves up by their bootstraps to former levels, especially when only a thin film depth of metal material is available as a resource to draw upon.
      Still further possibilities relate to the clear plastic covering the reflective aluminum. One possibility is that the plastic absorbs enough heat from the laser to slightly deform the plastic at the pit edges, so they are not read as cleanly and sharply for a second pass of the same area. This would degrade the quality of the eye pattern for the second, third, etc. passes. Plastics are blessed with a memory, so eventually the deformation would disappear, and then that section of the CD would sound fresh again.
      Another possibility is a little more complex and subtle. The energy of the laser passing through the clear plastic might subtly change its molecular arrangement, not enough to physically deform the plastic, but enough to temporarily change the clear plastic's index of refraction slightly. Now, it turns

(Continued on page 57)