.

von EMORY G. COOK (Cook Labs, 139 Gordon Place, Floral Park, L.I., N.Y.)

Causes of noise modulation in disc recording and how they may be overcome

.

Due partly to the influence of the FM (=UKW) program and partly to the competition of various new forms of sound recording (Magnetophone), current standards of performance in the disc recording field are improving rapidly. The newer "British phonograph records", together with the advent of feedback recording heads, improved disc materials, shapes of styli, etc., have all served to make possible the use of wider range reproducing playback facilities.
.

However, widening of frequency response and improving of all forms of distortion uncover from time to time certain residual "inherent" defects of systems, recording or otherwise. For example, the appearance not long ago of higher quality loudspeakers met with harsh words at first in some quarters because when they were attached to existing systems already in operation by the customer they drew aside the curtain on all forms of high frequency trouble, cross-modulation, non-linearity, and even parasitics.

The speaker itself was blamed. These defects had theretofore been buried in the much greater shortcomings of the previous loudspeaker or concealed by its limited frequency response, or both.

In the same way, shortcomings of the established disc recording methods are gradually being exposed. One of the most glaring deficiencies in disc recording (which incidentally has a certain parallel in magnetic wire or tape) is the matter of noise modulation
.

It is a matter of common knowledge with experienced disc recorders, that one of the prime essentials to the obtaining of a quiet groove is the correct mounting in the cutter of the cutting stylus.

The stylus set screw must of course come to rest directly upon the milled flat of the stylus shank in order that the stylus be pointing straight ahead. Otherwise, the stylus will be twisted in the mounting and a noisy cut will result.

Yet a glance at Fig. 2 will show that operation at an angle of twist is exactly what the stylus must do when engraving a signal in the groove. When the stylus is at the point in the sine wave shown at (a) it has no way of knowing - as far as noise is concerned - that it isn't being operated in a dead groove with twist (b).

The common method of measuring residual or surface noise in "lacquer disc recording" is to cut an unmodulated groove at the diameter and rpm indicated, insert in the playback circuit a high-pass filter to remove rumble and hum, and observe the resultant meter reading referred to "zero."

As a practical matter the character of the sound thereby reproduced will be not unlike thermal agitation - random noise. But in this case the frequency spectrum of the noise is such that about 70% of the energy content lies between 3.000-10.000 cycles (if we were to double the speed to 150 rpm for instance, and use an appropriate reproducing head, the upper limit would naturally be extended).

Unfortunately, this method of measurement is not necessarily a true index of performance of the stylus and lacquer, so far as surface noise is concerned, for in a way similar to that of intermodulation between two differing program frequencies, the actual reference noise level as measured by the dead groove method above, is modulating by the program itself.

Unlike intermodulation, the noise modulation is up, i.e., when a cycle of program comes along, the noise in the sloping part of the groove is higher than it was in the unmodulated groove (see Fig. 1).

The increase (Steigerung) in noise may be in some way a function of the slope of the modulated groove (without regard for sign), but not necessarily so. It depends a lot on the stylus itself. The actual value of noise modulation is conveniently expressed in decibels, and is defined for the purpose of this discussion as being the number of decibels increase in noise obtained when a 35° slope occurs in the modulated track (see Fig. 2).

No mere butterfly chasing is being undertaken here in the matter of noise modulation. The writer is certainly not in the process of waving gleefully aloft a rare, microscopic, but otherwise utterly inconsequential specimen for examination by fellow entomologists. (Der Satz war gut - ich habe ihn nicht übersetzen können)

Any impulse on the part of the reader to suspect so must be reconciled with the fact that "before and after" listener tests made with regular styli and anti-noise modulation styli are hard to argue down.

A "before and after" change so large in proportion that it is plainly distinguishable to the untrained ear stands out like a red nose in the measurements and instruments department, and can hardly be explained away any more easily.

It has already been well established in the art by numerous sources working in different ways, that cross or intermodu-lation is a more accurate index of listener reaction than amplitude or wave form distortion.

It often does happen in reproducing sound systems that the phenomena which cause amplitude distortion also cause intermodulation at the same time; and since distortion content was at first easier to measure, this was used as a standard of performance.

Just as intermodulation between various program frequencies runs a close parallel to listener reaction in the overload level department, so noise modulation closely parallels reaction in the matter of effective noise.

Although a conclusive probe into the matter of listener reaction has not been concluded, enough information is nevertheless at hand to warrant a few arbitrary statements concerned with the subject of the listening public vs. noise modulation.
.

• 1. When peak intermodulation is less than 4%, frequency response reasonably flat to at least 8-10 kc and the whole playback channel including speaker is "clean," the effect of noise modulation is laid bare in its character as a residual trouble.
• 2. With any given set of circumstances (i.e., type of program, rpm, diameter, etc.) the use of high frequency preemphasis, since it produces sharper angles of groove excursion than a "flat" response, increases noise modulation, although it decreases the noise.
• 3. When shellac pressings originally recorded with no high frequency content over 5,000 cycles are played back into a channel flat to 15,000 cycles, the impression is given of distorted high frequencies. In listening tests it is easy to imagine that the recorded program contains components above 5,000 cycles even when such is known not to be the ease. Modulation of the high shellac surface noise by the lower frequency signal produces this illusion. Indeed we do after a fashion have high frequency material present, in that the components of surface noise over 5,000 cycles heard in the wide-open playback system are being modulated or cross-talked at program frequencies.
• 4. The listener reaction ordinarily associated with plain noise is perhaps more correctly identified with noise modulation. There are some indications already that noise modulation, depending on a variety of circumstances, is the more important of the two factors.

.

For purposes of the present investigation an arbitrary (beliebigen) stylus Factor of Merit has been devised which, although open for appropriate modification and change from time to time as the occasion demands, is nevertheless a signpost leading to an ultimate disposal of the problem.

Die Berechnung dieses Faktors sähe wie folgt aus:

Factor of Merit = (Number of db between conventional noise and 10cm/sec) -40 dividiert durch Noise modulation (db)

In order to give an idea of the range of Factor of Merit values, figures in the neighborhood of 1.0 are acceptable, 2.0 is excellent, and 0.5 is very poor.

Obviously, the relation is shaky and starts to break down for values of unmodulated groove noise level noisier than -45db.

Fortunately, it appeal's that it is somewhat easier to manufacture a cutting stylus with a good noise modulation figure if the unmodulated groove value demanded in the application is not too severe. It is much easier, for instance, to make a stylus with only 6db of noise modulation if it is a -50db stylus than if it is a -60db stylus.

The method of measurement used to establish the noise modulation figure for any given set of circumstances may be of two fundamental types, static and dynamic.

The static measurement is easy to make, and as is so often the case with easy things, is not particularly reliable. It consists of turning the stylus 35° away on its own axis from the conventional position (Fig. 2(b)), cutting an unmodulated groove, and observing the increase in noise between this groove and an adjacent groove made with the same stylus without twist (Verdrehung).

The direction of twist must be tried both ways to correspond with the positive and negative velocities of the cycle in the groove. The unreliable feature of this measurement appears to be that since the stylus is operating under questionable conditions when twisted, the slightest imperfection in either it or the lacquer will hang up on the leading edge and stay there, causing a pessimistic reading, whereas in the dynamic method the stylus is continuously modulated or cycled as it would be with program, and particles which would be hung up under the static condition are thrown off.

The dynamic method of measurement involves selecting a freciuency which when operated to fill the groove to capacity will produce approximately 35° of slope as measured through the microscope. Naturally the freciuency necessary to produce this slope will depend upon the diameter and the rpm, i.e., linear speed, of the groove. It is not surprising that the same stylus will show slightly different values of noise modulation at different linear speeds depending upon the dimensions of the various facets, etc.1 The frequency necessary to produce the required angle will ordinarily be below 2,000 cycles so that in playing back the modulated (and noise modulated) groove .......

Hier fehlen die Seiten 11 und 12 der Ausgabe 12/1947.

und damit ist Ende.