A year or so ago, a friend gave me a little box he found at a Boston-area electronics swap meet. It was labeled "Dolby B decoder tester" and with it was the article (reprinted below) from a long-ago issue of the Boston Audio Society monthly journal.
In essence, the article describes a potential sonic problem with most consumer Dolby B decoders.
Have to get technical here to try and explain what goes on. These decoders operate by sampling the high frequency (HF) content of the incoming audio "stream" and automatically "rolling off" the HF of the incoming audio when there is no "information" there. In this way the audible effect of tape hiss is decreased. The decoder senses this information by changing the incoming AC audio information to a DC control voltage that varies the gain of a amplifier in the later audio path to increase or decrease the HF. Problem is how this DC voltage is detected. To keep the costs down for consumer equipment, most B decoders used "half-wave" detectors which only sense the "positive" portion of a music signal. Because music is naturally asymmetric, controlling the HF response of the entire signal based on what's happening on "only" the upper half of its waveshape will change things.
Bottom line is that there CAN BE AN AUDIBLE DIFFERENCE to the high frequencies of music coming OUT of a consumer (not professional) Dolby B decoder depending on the absolute phase of the music going IN to the decoder. Unfortunately, assessing the audible effect is not easy - you have to devise a way to reverse the phase of each channel's signal before the decoder - then do an "A/B".
Charles
To "B" or Not To "B"
Inverting polarity can create problems beyond those correctable by simply inverting the signal an additional time. Inversion can cause gross level errors in signal processors using half-wave detectors, integrators, or any other control circuits that behave asymmetrically. For a sine wave, the error is predictable, for the energy is symmetrical about the xero-axis. It may be inherently "corrected" if the circuit was designed-using sine (or symmetrical square) waves as test and calibration signals. Yet most music, in fact, most sounds are asymmetric.
Dolby-B is a typical problem system. A simple, but quite representative, music signal shape is a ramp waveform of 2 Khz with a rise time of 100 microseconds and a fall time of 400 microseconds. There is a 6 dB peak-to-peak level difference at -30 dB relative to Dolby level when such a signal is inverted in the record mode. It would not be a serious problem if a reciprocal change incurred in the play mode, but a couple of variables conspire to prevent this from happening much of the time: the tape recorder may invert the signal, and there is no convention of correct polarity within the different circuit elements by different manufacturers of Dolby-B units.
In fact manufacturers may not even be self-consistent. The Advocate 101, the Advent 100 and the Advent 100A all differ from each other. In the record mode, the Advent 100 and the 101, if fed a positive-going signal, provide a positive-going signal to the integrator and a positive-going output to the tape recorder. The Advent 100A, on the other hand, provides a negative-going signal to the integrator and an inverted output to the recorder. In playback, the 101 again provides a positive-going signal to the integrator and a non-inverted output, provided the tape recorder is non-inverting. The 100 however inverts the integrator's signal, yet gives a non-inverted output. The 100A does not invert the signal to the integrator, but does invert the output. But the 100A has already inverted its record signal, so again assuming a non-inverting tape recorder, the integrator really is seeing an inverted signal, and the unit is giving a non-inverted output; it is decoding correctly.
This is a confusing situation, and it is easy to see why manufacturers of Dolby-B's are confused too. Unfortunately, your tape's sound will become even more confused when a friend plays it back through a Dolby of a different color. Perhaps this is why Advent 100's sometimes remove high-frequency detail along with tape hiss and why Dolby-B encoded tapes do not always travel well. The differences (with a total of seven possible combinations) occur not only between products once made by Advent, but randomly between those of other purveyors of Dolby B systems as well. The Tandberg Dolby-B, for example ...
Then there is the problem of pre-recorded tapes. The cassette medium is perhaps insufficiently imprecise that one need not be concerned with minor frequency response and envelope tracking errors (if 6 dB is minor?). But how are Barclay-Croker tapes encoded? Fortunately for the recording industry, neither Dolby-A nor the various DBX systems have these problems.
Another consideration might be that before proposing such a problematic system as Dolby-B for FM broadcasting, standards should be adopted not only for all elements of Dolby-B circuitry but also for all parts of the FM signal chain. The standards would have to include the exciter, transmitter and the FM tuner, its multiplex system, and of course its Dolby decoder. It might be far simpler to adopt a 50 microsecond emphasis/de-emphasis standard and forget Dolby-B entirely. As Boston-area listeners know, state-of-the-art FM broadcasting by at least two local stations hardly requires Dolby-B encoding.
Scott Kent (Massachusetts)