By Robert E. Berglas [ University of California]
Before describing the system I have worked out for an accurate measurement of Wow and Flutter using digital counting techniques, some attention should be given to the weaknesses inherent in existing methods.
Present methods do not lend themselves to a central criterion of scientific investigation: namely, reproducibility of data. Given the art as it exists, published results are more a function of the particular test instruments, than the actual matter under review. Printed Wow and Flutter specs simply cannot be reproduced unless the manufacturer's set-up is known (and his test equipment happens also to be on the user's bench). Recognizing this embarrassing fact of engineering life, the industry has attempted to bring some method to this madness by hitting upon a mind boggling variety of "weightings" and a priori suffixes. In sum, these are equations which lop off equipment eccentricities and, unfortunately, "fudge" results.
To be more specific, consider for a moment the numerical manner researchers devised for specifying degree of speed variation-of course, the cause of Wow and Flutter. The unit of measurement became RMS percentage of speed variations. And, yet, the RMS notion is no more than a convenience--
a) to transform a periodically changing quantity into a steady quantity, and, b) a convenient average value that subverts more than it explains.
Fundamentally, Wow and Flutter is frequency modulated (FM) program material; and the RMS concept is, at heart, amplitude modulation (AM) of an AC voltage. It simply will not do to transform peak-to-peak voltages into "peak-to-peak percentage speed error." It may be convenient, but it is also a fudge beyond the definitions we are working with.
Another in proportionate conversion takes place within existing Wow and Flutter meters--i.e., the conversion of AC to DC (to drive the meter). The DC voltage in these instruments is generated by a "detector." The problem here is the same as was for the RMS concept-namely, that this is not "detection" in the true sense. Detection refers to the recovery of amplitude modulation from an RF envelope.
Wow and Flutter components, however, change the frequency of the incoming signal, and converting to DC is inappropriate and not proportional. In the process of making the meter happy, investigators have fundamentally confused AM with FM.
A New Digital Wow and Flutter Method
Analog devices such as meters are fine for measuring voltages; but they just cannot contend with rapid fluctuations in frequency.
With recent advances in digital IC propriety devices, we have at our command, and for the first time, accurate and precise systems for frequency determination. Indeed, my friends at Texas Instruments developed their remarkable TTL digital IC line largely for counting purposes, where, often, the input would be variations in frequency.
The new digital method for measuring Wow and Flutter that is described here, is made at once simple and incredibly accurate by pressing into service the latest generation of Frequency Counters, built around TTL logic. With this instrument and a suitable stable test signal, frequency modulated variations in turntable and tape recording gear can be measured with repeatability and authority. The procedure described is simple, easy to perform, and, to tell the truth, a remarkable experience for the investigator.
Wow and Flutter components will frequency modulate a test signal, and if we monitor resulting output tones with a Frequency Counter, we will not see displacements from the standard input signal.
Anticipation and an understanding of the phenomenon serves to prepare one for the actual procedure. Wow and Flutter give off specific frequency variations: Wow is known to affect frequencies under 6Hz; while Flutter exists in a higher range, up to 250Hz.
Further, these frequencies modulate very often both above and below the test signal (known as positive and negative speed variations). Let us say we have a 1000Hz test signal, either on a record (say, the CBS BTR-150, or the one put out by Stereo Review), or recorded on a test tape. Moreover, let x equal a possible wow component of 6Hz or less; and y equal flutter existing at up to 250Hz.
Then, if we monitor the playback on either a turntable or tape machine, we will get the following interesting phenomena:
Wow .... 1000±x;
And
Flutter .... 1000±y;
The readout usually starts off, obviously, with the 1000Hz reference.
But, then, with transients which repeat themselves with periodicity, the modulated compromise is flashed out as the instrument is toggled by variations in count. We can now calculate the true percentage of speed variation, manifest as percent FM:
[(1000± x)- x / (1000± x)+ x] x 100
Flutter determination is similar.
Choice of Frequency Counter
The Heath Company today puts out some of the most advanced and inviting Frequency Counters on the market.
They are designed around TI's TTL logic IC's, and a first generation Schlumberger unit was used by the author.
A word of explanation should be noted about other Counters. There is some difference between the general purpose Counter/Period Counter, and one designed specifically for high-speed frequency work.
For one thing, the readout should be non-blinking, and reflect both a continuation of count and change only when the input's frequency varies. The unit, then should be latched. We do not want the readout to reveal a count up or count-down. With latched decade counters, the display continues to put out a given frequency while a new one is being counted. With the general purpose Counter, the count period is a divided down timing interval of the clock frequency.
The input shaping circuitry show some differences, too. Counters/Period Counters use a comparator input which is best suited for reducing noise or zero-crossing, while the other kind of unit uses well designed amplifiers and a precision Schmitt Trigger. The latter is better suited for rapidly and idiosyncratically changing frequencies.
Readers who try out the above test procedure will probably notice what appears to be either a mis-cue or falsely recorded as FM of the least significant digit. The "± 1" flicker of this digit is inherent to most Frequency Counters, and should not be noted down as a speed variation. (It occurs when the count gate opens and a plus-or-minus one-count ambiguity exists.)
Present specs are far from perfect. The DIN is not really an improvement. RB's idea is an interesting one but it has snags too. -Ed.
(Audio magazine, Jun. 1973)
Also see:
Build a Poor-Man's Wow & Flutter Meter (Jun. 1983)
Cassette Deck Transport Problems (Sept. 1974)
Logic Control of Tape Recorders (Apr. 1973)
Happy 25th Birthday, TRANSISTOR! (Jun. 1973)
= = = =