A Short History of Transmission Audio Processing in the United States

A Short History of Transmission Audio Processing in the United States

Robert Orban, San Francisco, CA

In the early days of broadcasting, the primary purpose of transmission audio processing was to protect the AM transmitters of the time from damage due to modulator overload. Simple peak limiters using variable-mu tubes in a push-pull configuration were employed. Because the gain-control signal was, in essence, mixed with the audio signal, these early vacuum-tube devices required careful balancing to cancel “thumps” representing feedthrough of the gain-control signal into the audio. Dynamic range control was effected through careful manual gain-riding — in classical music broadcasts, the “compressor” was a skilled operator reading the musical score and using it to anticipate the required level adjustments. To this day, no one has invented a more subtle or effective method of compression!

Later, simple compressors were placed upstream from the limiters in situations where the budget did not permit skilled manual gain-riding. These compressors were not gated and could exaggerate noise objectionably.

In the Region 2 countries, 75µs pre-emphasis is used in FM and television sound transmission. This pre-emphasis is up 17dB at 15kHz and can cause severe over-modulation if its effects are not controlled. The obvious solution – placing a wide-band peak limiter after the pre-emphasis filter – proved unsatisfactory because high-frequency overloads would cause severe spectral gain intermodulation: cymbal crashes would cause the sound to literally collapse. The Fairchild “Conax” (originally designed for disk cutting) was often used to ameliorate the problem. This device divided the audio into two bands with a 1kHz crossover and applied pre-emphasis, clipping, and high-pass filtering to the upper band. The high-pass filter reduced the difference-frequency intermodulation caused by the clipper, yielding reasonably acceptable sound.

“Modern audio processing” could be said to derive from the work of the design team at CBS Laboratories in the early 1960s. Their “Audimax” (mispronounced by generations of engineers as “audiomax”!) was a gated wideband compressor that successfully eliminated the noise-breathing problem of earlier compressors. The “Volumax” was a clipper preceded by a limiter with a moderate attack time. The moderate attack time prevented the unit from punching holes in the program, while the clipper controlled the peaks that the preceding limiter did not catch. The “FM Volumax” introduced a high-frequency limiter to control overload due to the pre-emphasis curve. This high-frequency limiter was a program-controlled 6dB/octave shelving filter placed between the limiter and clipper. Once again, a moderate attack time was used and the overshoots were controlled by a final clipper. The “Dynamic Presence Equalizer” measured the ratio of midrange energy to wideband program energy and applied midrange equalization as necessary to correct the midrange spectral balance of the program.

In the early 1970s, Dorrough Electronics introduced the “Discriminate Audio Processor” (“DAP”). There were versions for AM and FM. The DAP divided the audio spectrum into three bands with gentle crossover slopes and compressed each band independently. The bands were recombined and applied to a clipper with a very “soft” transfer characteristic. The DAP greatly reduced spectral gain intermodulation by comparison to its wideband predecessors. Additionally, many engineers adjusted the three bands for different gains, using the device as a dynamic program equalizer as well.

In 1975 Orban Associates introduced “Optimod-FM.” This unit combined compressor, limiter, high-frequency limiter, clipper, 15kHz low-pass filters, and stereo multiplex encoder into one box. This greatly reduced the possibility of misadjustment of the processing chain. The unit’s 15kHz low-pass filters were non-linear filters without significant overshoot, and therefore permitted higher average modulation by comparison to the linear low-pass filters used in the stand-alone stereo encoders of the time.

In 1977 Orban Associates introduced “Optimod-AM.” This unit contained a high-slope receiver equalizer to pre-compensate for the highly rolled-off radios of the time, and also included an 11kHz low-pass filter to ensure that the unit complied with the occupied bandwidth requirements of the 1978 FCC Rules. It also introduced the distortion-canceling clipper, which substantially reduced difference-frequency intermodulation distortion caused by clipping.

In the late 1970s, Circuit Research Laboratories introduced a processing system for AM whose most important novel features were a phase rotator [the Kahn “Symmetra-Peak” being a fore-runner] prior to processing (to make voice more symmetrical, reducing clipping distortion), and a subsonic equalizer after final peak clipping to pre-distort the output waveform of the processor to compensate for low-frequency tilt in the plate-modulated transmitters of the time. Compensating for this waveform tilt enabled the better transmitters to be substantially louder by eliminating a factor that would otherwise increase the peak-to-average ratio of the modulation. Although intuitively inobvious, using a phase rotator to purposely eliminate the asymmetry in voice proved to be far more effective than the older “polarity follower” [Pacific Recorders AM “Modulimiter”] circuit. The older circuit preserved any natural waveform asymmetry and switched its output polarity such that the side of the waveform with the higher peak level modulates the carrier in the positive direction.

In the late 1970s, a number of manufacturers made “composite clippers” designed to be placed between the output of the stereo encoder and the input of the transmitter. These controlled the peak modulation of the composite stereo signal unambiguously at the expense of introducing harmonic and intermodulation distortion throughout the stereo baseband. Many “hit-format” broadcasters thought that the increased loudness achieved by these devices justified compromising the spectral purity of the baseband. Eventually, the FCC judged these devices to be in violation of the FCC Rules of the time if they caused the instantaneous 19kHz stereo pilot tone injection to be less than 8% modulation. In essence, this meant that the pilot could not be clipped and must be injected after the clipper. In 1982, Modulation Sciences introduced a composite processor that did this, thereby performing to the letter of the FCC Rules.

In 1982, Orban Associates introduced the “Hilbert-Transform Clipper” as part of its Optimod-TV processor for stereo television. The “Hilbert-Transform Clipper” was later adapted for use in shortwave as well.

In general, transmission audio processing in the 1980s refined and built upon the revolutionary developments of the 1970s without introducing any radical novelties. Each manufacturer, for example, has a proprietary technique for producing non-linear overshoot-free low-pass filters for FM and television applications. Several manufacturers (including Inovonics and Circuit Research Laboratories) introduced programmable processors whose subjective setup controls can be changed by remote control to match the programming of the moment.

In the 1990s, the field must be considered “mature.” As in every other area of audio, digital signal processing (DSP) is likely to eventually supplant analog circuitry. As of this writing, Orban, CRL, Valley International, Gentner Electronics, and Audio Animation have introduced transmission processors in which all processing is done in the digital domain. [The Valley, Gentner, and Audio Animation units are no longer manufactured.] If properly designed, such a processor can be readily reconfigured in milliseconds to change almost any aspect of its topology, such as the number of bands in its multi-band compressor. Subjective setup control settings can be stored and later recalled by local clock, remote control, or computer to daypart processing. The processor can readily generate test and signalling tones, facilitating tests of the transmission system and the generation of EAS alert tones.

In a digital processor, achieving sound quality equal to or better than its analog counterparts requires a marriage of art and mathematical design more rigorous than anything in the genesis of its analog ancestors. Many common analog processing functions (such as clipping) are much more difficult to do competently in the digital domain. However, digital also presents the opportunity to do things unachievable in analog, and digital’s overwhelming advantages will ultimately manifest themselves as clearly here as they have elsewhere in the audio processing arena.

Copyright 1992 Robert Orban. All rights reserved. All trademarks are the property of their respective companies.