Stop Digital Madness! Professor Reilly responds part 2
To account for the observed speed variations, I have postulated that, once the microcrack is generated, moving dislocations pile up against it, thus storing energy in its immediate vicinity. Here thermal energy is of primary importance. Experimentally this is corroborated by the temperature relationships discussed earlier. As the dislocations collect (established in electron microscope micrographs), the area adjacent to the microcrack becomes elastically deformed. When the energy builds up sufficiently, it is suddenly released in the form of little shock waves called acoustic emission (AE). AE is typically a high-frequency phenomenon of 50kHz or more.
For those readers unfamiliar with acoustic emission, I suggest reading ASTM Special Technical Publication 505, Acoustic Emission (American Society for Testing and Materials, 1971), and A.E. Lord, Jr., "Acoustic Emission" in Physical Acoustics, Vol. II (1975). The hypothesis is, given the high density of microcracks, AE bursts are continually emitted, causing minute tremors in the bearing. Cumulatively these tremors lead to speed variations and the attendant musical smear. Detection of AE is a logical next step in providing further support for this theory.
The ways in which the graphs were printed and discussed in Stereophile are filled with grave inaccuracies. First, these graphs on p.14 were published without asking my permission. The graphs were sent to Mr. Archibald and Mr. Holt as part of an informational package to help them comprehend the research. Never would I have permitted the names of the manufacturers to be used. The graphs are representative of dozens of models of turntables, irrespective of brand. Secondly, the labeling of axes and the pertinent data printed on each graph were omitted in the reproduction, leaving the graphs unintelligible. The horizontal axis should be labeled "Time," with each two divisions equaling one second. The vertical axis carries a label of "Percentage Variation From 33.3 rpm."
Thirdly, the discussion in the text does not correspond to the graphs. As part of the testing, the speeds of new turntables were measured before, during, and after their initial playing of digital recordings. Graphs 1, 2, and 3 are part of such a sequence. Consider graph 1. This represents a tiny segment of the "before" measurements on a new 'table simply rotating. The deviations observed are not fluctuations at all but merely reflect the limits of resolution of the high-precision tachometer used to measure speed. Correctly interpreted, the graph says that speed was constant, and thus it remained throughout this phase of the test.
When a digital record was played, graph 2, the speed varied considerably, the fluctuations becoming more pronounced in magnitude and frequency of occurrence as time passed. After 20 minutes this phase was terminated for fear of permanent damage to a particularly expensive turntable. Immediately thereafter an analog record with large dynamic range was played. The speed measurements, illustrated in the left-hand portion of graph 3, show that residual effects remained; there are some variations. These, however, tapered off over the next 45 minutes until the 'table, still playing the same music, returned to constant speed which was then maintained to the end of the test (right-hand portion of graph 3).
Every test of this nature on various new 'tables has produced the same results. After usually 4-10 hours of digital-record playing, the fluctuations persist whether digital or analog records are used.
When a 'table experiencing speed variations was exposed to cooling currents or lower temperatures the fluctuations would abate. Graph 4 comes from a sequence demonstrating such behavior. In this particular sequence, the 'table was simply turning. At higher temperatures earlier in the test, the speed was quite aberrant. With a cooling current, the fluctuations moderated (graph 4) until, in this case, they finally ceased. These results have been duplicated on many 'tables.
These graphs were generated by using the average speed of the 'table as the baseline. For clearer presentation of the data we are now using the speed with the greatest frequency of occurrence as the baseline. With this alternate graphical representation, the fluctuations still appear pronounced but reduced somewhat in magnitude and usually fall on one side of the baseline.
As the reader will note, we are using test gear which is nontraditional for turntables. The musical smear which originally prompted this study is associated with high-frequency phenomena falling outside the range of conventional audio measurements.
The correlations existing between the many experiments and the theoretical postulations are of such a magnitude as to warrant further investigation, not immediate, out-of-hand dismissal. Correlation is the very nature of original research into areas heretofore unexamined. From experimental data certain relationships emerge. Hypotheses developed to explain these observed relationships lead to predictions which are then tested in the laboratory—and so it repeats. Such is the manner in which this study has proceeded. The recurring marked consistencies between laboratory results and my theoretical predictions are certainly grounds for further research.
I feel very strongly that Neil Levenson's reputation has been unnecessarily compromised in this article. In truth, audio writers of his caliber and perceptiveness are extremely rare. For months he was entirely skeptical of my postulations on the origins of speed inconstancy. However, he closely followed the research developments and read extensively. (A very thorough bibliography related to this research will be found in his column in the March/April 1986 Fanfare.) Far from casually endorsing my work, he spent months critically assessing it, looking for flaws, and trying to disprove it. His deep commitment to music and to upholding its values should be emulated by more audio commentators.
In responding to "Stop Digital Madness," I felt that a coherent account of the research was of far greater importance to the reader than addressing each inaccuracy separately. Hence, given the limitation of the space allotted me, many of the article's erroneous details have gone uncorrected.—Judith Reilly)