A New Standard in Turntable Speed Constancy (June 1978)

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting




[Authors: George T. Gillies, Rogers C. Ritter, and Robert T. Rood]

The last decade has seen rapid advancement in turntable technology. Some of the models commercially available have featured precision belt-drive systems, direct drive motors with various types of closed-loop feedback schemes, and a variety of other innovations.

An important goal in engineering turntables to high standards is to reduce the wow and flutter levels to some point well below the limit of human detection.

Rotating turntables, however, also have interesting uses outside the field f sound reproduction, and one finds at for certain scientific applications, eel speed constancy of the turntable must be between 10,000 and 10,000,000 times better than the finest commercially available turntables.

There has been interest in developing such rotational apparatus in our physics laboratory at the Univ. of Virginia for a host of applications over most of this century, and recently we have succeeded in driving turntables at speed constancies never before achieved [1].

Our specific purposes in developing such instrumentation are three-fold:

1) We wish to do a laboratory experiment in which we can unambiguously test the theories which predict that the force of gravity gets weaker as time goes one]; 2) another experiment (currently underway) is one in which a study of the motion of rotating cylinders would tell us whether or not matter is being creed in the universe [3], and, 3) it appears that we can test the earth's rotational speed fluctuation and wobble, measure latitude fluctuations, and study other geophysical phenomena by observing the forces acting on small masses placed on an almost totally constant-speed turntable.

Description of the System

The turntable we are currently using for these experiments is shown in Fig. 1. It is a 95-lb. brass disc that has been mounted on an air bearing with no mechanical contact between the bearing surfaces. Instead, the rotor of the bearing floats on a cushion of air approximately 0.0001-inch thick. The turntable (moment of inertia 500 lb.in.^2 is mounted on a 500-lb. granite block which rests on damped pneumatic springs which sit on an 8000-lb. granite block. This, in turn, sits on four damping isolators, each consisting of eight steel plates separated by ribbed neoprene pads.

A new drive for the table consists of a special motor with certain similarities to, but many differences from, commercial direct-drive motors. It consists of 50 coils facing inward (see Fig. 2) from the stator which "write" semi-permanent poles in a 10 mil steel strip around a 12-inch diameter aluminum rotor. The coils are in series with alternate ones reversed. As the rotor advances 1/25 of a turn, the drive signal has completed one electrical cycle and the motor operates synchronously. The main difference between this unit and an ordinary synchronous motor is that the partial "rewriting" of the poles each 1/25 turn ends up in an average pole position and strength being set up on the rotor.

This greatly reduces the effects of pole asymmetry in position and strength and largely confines short-term fluctuations to the synchronous frequency which is much easier to filter out in the feedback system.

The drive signal has been generated from a function generator for wave-shape testing, but the precision source is a sine wave derived from a local oscillator phase locked to an atomic clock. This low-level signal is fed through a 200-watt commercial power-supply amplifier (Kepco Model BOP365[M]) to drive the 15-ohm coil load. Figure 3 depicts the drive system schematically.

Other Drive Schemes

We have studied many different drive systems over the last three years, some more exotic than others, and a number of them are interesting and potentially useful. Two schemes on which we did only preliminary work are: 1) two permanent magnets were mounted on the bearing rotor, bracketing one such magnet on an arm from the stepping motor drive, and were aligned for repulsion, so that a "soft" magnetic coupling between the motor and table absorbed the sharp jumps at every step of the motor and 2) an eddy current drive we developed consisted of permanent magnets on an arm from the drive motor being rotated near a conducting disc on the table. This last provides the requisite "soft" drive but puts considerable demands on the feedback system. Both of these drives were ultimately replaced by a mechanically coupled stepping motor drive which will be discussed later.

Consideration of Speed Constancy

For audio purposes wow and flutter are of prime .concern with absolute speed accuracy secondary, as few human beings can detect 1 percent variations occurring over a period of several minutes. For broadcast purposes the duration of a selection may be important in which case absolute speed accuracy is needed. In our research all of these factors are important, particularly the latter, and our methods of measurement may seem unconventional with respect to present industry standards.


Fig. 1--The turntable, air bearing, and optical lever are shown here. They are mounted on two stages of vibration isolation hardware, each stage being composed of a massive stone block supported by appropriate spring-like couplers. Some of the electronic components of the drive and speed sensor are shown in the background.


Fig. 2--The 50 motor drive coils are shown mounted on the steel stator, and the motor-drive plate is centered among them. The motor power amplifier is in the foreground.

In non-technical form the standard terms are defined [4]: " 'Wow' usually refers to cyclic deviations (in the motion of the medium) occurring at a relatively low rate, e.g. a once-per revolution speed variation of a turntable," and " 'flutter' usually refers to cyclic deviations occurring at a relatively high rate, e.g., 10 Hz." Commercial turntables quote figures for both of these as low as 0.03 percent and occasionally lower. The long-term accuracy of direct-drive turntables is usually governed by the accuracy of the drive oscillator, much as our own system. Typical quartz oscillators are good to about one part per million, i.e. to 3 sec. per month, which is also the accuracy of good quartz wristwatches.

For audio purposes such accuracy is sufficient but when one is studying a force like gravity which may change by as little as one part in 100 billion per year, speed drift must be much lower.

Wow and flutter are somewhat less important since such effects average out in a long (one to two month) measurement provided they are small enough in magnitude, less than 0.000001, so that they do not generate effects in the signal through non-linearity.

Performance Measurements

One basic measurement of the turntable was in the undriven mode, determination of the Q or quality factor of the table. This was found from the slope of the exponential decay of the periods in coast-down tests and has the value of about 3000 at 33 1/3 rpm. (This number varies by about 10 percent over a wide range of speeds.) This means that the decay time of the turntable is about 860 sec., i.e. the table speed falls to 37 percent of its initial value in this period of time. It will coast visibly for more than an hour.


Fig. 3--The local oscillator output is phase locked to the WWV reference signal. After digital division and reshaping of the waveform, the resultant signal is amplified and used to drive the motor coils.


Fig. 4--The optical lever sends a light beam towards the table and receives a reflected light pulse from it once per revolution. The timing trigger provides a uniform pulse for starting and stopping the counter. The counter output, the rotation period length, is then printed on paper tape for later analysis.


Fig. 5--Typical data depicting the small fluctuations in the driven turntable period are shown here. Each plotted point represents a single measured turntable rotational period.

The method of measurement used to obtain the coast-down results and others described below was to measure the periods of rotation. A block diagram of the method is shown in Fig. 4. An optical system called a "Jones Lever" uses a light source and split silicon photo-diode with appropriate lenses and slits which generate a pulse upon the passage of a mirror mounted on the table. A fast counter times pulse intervals. In principle, the timing resolution can be made as low as 0.00000001 seconds with this method, but with our simplified version the rotational periods were timed to about 0.000002 second. For a speed of 15 rpm this limits the measurement of fluctuations in the rotational periods to no better than 5 parts in 10 million.

Careful measurements of individual periods have only been made on the old, unfedback, stepping-motor drive.

Figure 5 shows a typical set of measurements of individual periods at 15 rpm for about 200 periods. The standard deviation of these is about ± 5 parts per million and none exceeded ±8 parts per million. Thus the wow is less than ±8 parts per million or 0.0008 percent.

Long-term variations were measured by averaging over times corresponding to 10, 100, or 1000 periods, thus bypassing the timing limitation mentioned above. This was done automatically by setting the 8-digit electronic counter to time and average these decade multiples of periods. In good runs the variations were a few parts in 10 billion over 1000 revolutions. That is, the average speed variations over 1000 periods were about 0.00000003 percent at 15 rpm.

When reaching into regions of precision such as this, the reference clock for timing and for the drive must be much better than the usual temperature-controlled quartz oscillator. We used a local quartz oscillator phase locked to a radio signal from WWV. Due to atmospheric variations the received WWV signal varies over short time periods, but this is averaged over long-term measurements: Thus our stability over short periods, e.g. an hour, was set by the local oscillator, this one having frequency fluctuations which were typically 0.0000000001; longer periods would have a precision traceable ultimately to the atomic clock at the National Bureau of Standards in Washington.

The measurement of flutter at levels of interest to us is a more difficult problem. As yet, we have no numerical results though we estimate that flutter with the old drive was comparable to the wow and will be better with our new drive.

A system for measuring flutter is now in the early test stages. It consists of the following: Laser light is diverged to a 3-inch diameter beam and passed through two 12,000-line transmission type encoder plates, one stationary, and the other rotating with the table.

The transmitted, fluctuating light is then converged onto a silicon photodiode and the resulting electrical signal is amplified and analyzed.

Although noise and alignment have prevented us, thus far, from using this system near the limits of its capability, we can, obviously, expect ultimately that Fourier analysis of the signal will lead to a very sensitive measurement of all flutter components up to about 10,000 Hz before aliasing sets a limit.

Our vibration isolating system is very effective in that the residual vibration levels are below our ability to measure them. With a Mechanalysis Model 350 Analyzer and its #544 probe, we find an average vibration level of about 10 micro-inches peak-to peak on the floor, but internal instrumental noise completely overrides any vibration on our table at a level of about 0.5 micro-inch rms.

Future Work

If vibration ultimately is still a problem, we will install a feedback system to correct this. It consists of three plates separated by two sets of stacked piezo-electric crystals. The upper "sensing" set senses remaining vibrations and puts out a signal which is amplified and fed back out of phase to the lower "motor" set which cancels out the vibrationally induced motion.

In one such system at The International Bureau of Weights and Measures, Sevres, France, the vibrations are reduced by a further factor of 100. Clearly, such measures as the above are not needed for everyone's stereo turntable, but for those of you interested in having a setup like this in your den, the price tag is roughly $30,000. Finally, such a turntable needs a suitable cartridge and tonearm. In a future article we hope to report on the laser-optical pickup system we have conceived. Finely-focused laser light is guided to and servoed onto the record grove; then position-sensing detectors produce the two components of the stereophonic signals. A microprocessor sorts out the signal from record imperfections (the noise) by spectral analysis and provides appropriate control for the light beam as well as processing the signal. Such a system will be most useful if it can function with current records, and it seems likely that it can. The overwhelming advantage of such an arm is the delicacy with which it treats your records. The "stylus force" is only 0.000000000000000000000001 gram.

1. G.T Gillies, Development of a Constant Speed Drive, Masters Thesis, University of Virginia, 1976.

2. R.C. Ritter and J.W. Beams, "A Laboratory Measurement of the Constancy of G," in On the Cosmological Variation of the Gravitational Constant, ed. P.A.M. Dirac and Leopold Halpern, (in press).

3. R.C. Ritter, C.T. Gillies, R.T. Rood, and J.W. Beams, "A Proposed Dynamic Measurement of Matter Creation," Nature 271, 228(1978).

4. Reference Data for Radio Engineers, International Telephone and Telegraph Co., (H. Sams and Co., Indianapolis, 1968), section 28, page 21.

(Source: Audio magazine, June 1978)

Also see: Phonograph Reproduction --1978: part 1 (May 1978)

= = = =

Prev. | Next

Top of Page    Home

Updated: Wednesday, 2017-06-28 13:32 PST