The "Prophetic" Processing History
Yesterday, Today and "Beyond"


Remembering Jim Somich
By Frank Foti

Jim’s last review on RW Online at 02.14.2007
for Sangean HDT-1 HD Radio

Go To Part #2

This series of articles re-published from
issues of February to August of 2007 original on PDF format.

With this issue we start a new series of articles on audio processing. While this article starts with a lot of History, the true focus of the series is the current state of the art and where we are headed. Jim Somich takes us on this guided tour.  

Jim Somich’s career included positions as a major market Chief Engineer, Director of Engineering for a group owner, and as the designer of a number of products, including the FlexiMod FM Processor of his own MicroCon Radio System company.
(Radio-Guide February 2007)

Yesterday, Today, and Tomorrow
by Jim Somich

The author would like to thank Barry Mishkind, Bob Orban, Frank Foti, and Corny Gould for their invaluable assistance in the writing of this series.

Part #1 

From Peaks to Power
Audio processing has its roots in the need to prevent over-modulation. In the 1950s, which some consider to be the birth of modern day audio processing, it was also used to automatically ride gain. But today’s broadcasters depend upon audio processors to do much more than just ride gain. Is this good or bad for broadcasting?

Audio processors are now almost completely controlled by one or more microprocessors. From wideband leveling to multi-channel, stereo-generating, look-ahead limiting and diversity delay capable, in many ways, they are the heart of a modern broadcast station. The importance of the arcane field of broadcast audio processing is evident by the extensive work done in the field since the earliest days. From PROGARS through Uni-Levels and Sta-Levels into the Audimax years, processing was evolving into an artform to be mastered by a select adventuresome few. This is not merely an historical article, but to be prepared for the acutely competitive future, we had better know a little about the past and a lot about the present. This series will be a quick look at the past, a lingering look at the present, and a glimpse into the future of broadcast audio processing, written by someone who has “been there, done that.” The processing business will change radically over the next few years and, in a decade, you will not even recognize it! I think back to the 1980s and a young engineer I hired who had a certain gleam in his eye. He did not have much experience but his enthusiasm was almost boundless. Little did I know that he would go on to become a true rock star in the processing world. This article is for you, Frank.

One thing is for sure: there will be new players on the scene along with many of today’s superstars. I, for one, cannot wait to hear what they accomplish. While some actively attempt to re-create the sound of the tube processors, the leading designers already are looking ahead to new horizons of audio control. The names are legendary and few: Emil Torick, Mike Dorrough, Bob Orban, Greg Ogonowski, Eric Small, Ron Jones, Steve Hnat, Donn Werbach, Glen Clark, Jim Wood, and Frank Foti. Each had a vision of the way radio should sound and made it happen. What a heritage: Audimaxes, DAPs, Optimods, Aphexes, and Omnias. As processors became more sophisticated, creative engineers and programmers found ways to use processing to develop a unique sound for their station. Some felt that a “wall of sound” with virtually no dynamic range was the way to snag listeners. Others felt that smashed audio only resulted in lower TSLs. The controversial world of modern audio processing was born.

Some exceptional engineers, finding their hardware lacking, began to modify boxes to perform tricks never anticipated by their developers. Another group went one step further and designed their own custom processors from the ground up. In many cases, these custom boxes became products literally built in garages and sold to the industry. Over the years, successful audio processing has not been the province of large companies or groups of engineers. In most cases, a single lone engineer with a vision struck out on his own to capture his aural imagination in a “magic box.” But who will be the gurus of tomorrow? And what will they have to work with? They are out there, working in the trenches. Guys like Scott Incz, Corny Gould, and John Burnill have dreams that just might come true. It is not as easy to hotrod a DSP processor the way one could change component values in an Optimod or a DAP, but I know one thing for sure: there will be a new generation of processing gurus and they will have new ideas. They will do things with their boxes that we have not even dreamed of.

In the beginning, there was no audio processing. AM radio stations kept the  modulation levels low – perhaps 30% on average – and used the technique of “manual gain riding” to avoid over-modulation. That meant a live person sat with hishand on the knob, trying to anticipate what was to come next. The practice was reasonably successful, but hardly efficient. Failure to properly anticipate a spike in audio level often resulted in the transmitter overloading and dropping off the air – or worse. Even the behemoth WLW 500 kilowatt transmitter was operated in the mid-1930s using only manual over-modulation control. The transmitter logs are replete with descriptions of outages caused by modulation peaks. Most were brief, but some notes indicated blown up capacitors, tube failures, and other problems that took longer to repair.

Gain riding was an art, and practiced diligently by the studio engineers of the 1930s, 1940s, and 1950s. When I started at WGAR, in 1959, almost all gain control still was done manually. By then there was a GE BA-5 peak limiter at the transmitter for over-modulation protection, but one of the primary duties of a studio “engineer” was to ride the gain. There was not a compressor in the entire studio plant. The chief had installed a chart recorder in master control that made a permanent record of outgoing level to the transmitter every minute of every day. Each morning, one of his stops was at the chart to check up on the gain riding of the engineering staff during the past 24 hours. Each engineer was held completely accountable for his shift. But, with the advent of post-Television radio broadcasting with its combo operation, fast-paced shows, short jingles and multiple elements, the need for an automatic form of gain riding became inescapable.

The first peak limiters came to market in the mid-late 1930’s. RCA introduced the 96A in 1936 and this may be the very first commercial peak limiter to hit the market. Western Electric introduced the 1126A in 1939. But you could hardly call these boat anchors audio processors. They were basically mundane tools to eliminate over-modulation, pure and simple. The PROGAR (PROgram GuARdian), developed by Al Towne at KSFO, San Francisco in 1935 really was the first known audio processor: a combined intelligent compressor (automatic gain control - AGC) and peak limiter. But it took more than ten years for Towne to patent it, sell it to Langevin and bring it to market. Then something exciting happened. Peak limiting became much more sophisticated. Even today we would have to agree, it was ahead of its time.

Early peak limiting kept the transmitter on the air

The PROGAR brought automatic gain control and peaking limiting together

In 1947, General Electric introduced the BA-5 delayline peak limiter and it took the broadcast industry by storm. It was cleaner than anything that came before because it used a feed-forward limiting scheme and a delay line to “give the audio a change to catch up to the bias generator.” How clean was it? Back in those days, NBC had an iron-clad rule that they would only use RCA equipment. Yet, they bought BA-5s and removed any evidence of GE manufacture. They repainted them RCA umber gray and added the RCA meatball. Magically, the new “RCA” peak limiter was born! Every NBC O&O began sounding much better thanks to some “midnight engineering.” GE did not rest on their laurels either. They continued their dominance of the peak limiter market by introducing the BA-6 in the early 1950s and the BA-7 in 1957. These boxes were really unique. To make a long story short, the input audio modulated an RF carrier and all peak limiting was applied to this carrier. After demodulation, the audio was fed to the transmitter. Many processing artifacts were eliminated by this scheme, but it was an absolute bear to keep in alignment and it took two engineers or one bodybuilder to wrestle one into a rack!

Throughout the 1950s, many FM stations eschewed audio processing entirely. I remember one old-timer studio operator telling me that “you really couldn’t overmodulate the FM.” Later, some FM stations installed a Fairchild Conax preemphasized clipper to tame the pre-emphasis, but that was all. In fact, it was not at all unusual in those days to watch the modulation monitor “pin” on muted trumpets even when using a (conventional) peak limiter. There was quite a way yet to go in developing effective FM processors.

The most important part of a compressor or peak limiter is the gain control element. During the early years, this was usually a tube. The PROGAR used a 6L7 heptode tube. The 6L7 was designed as a variable mu (amplification) tube – that was the purpose of the extra grid. All tube compressors and limiters functioned by mixing a DC control voltage with the audio at the grid of a variable mu tube. These  amplifiers used push-pull operation so the control voltage could be effectively canceled at the output. This reduced the “thumps” that were common when these boxes got out of balance due to tube aging.

There were a few remote cutoff tubes designed before the GE-6386, but this tube became the rock star of the 50s in audio processing. It was the basis of the GE Uni-Level, Gates Sta-Level and Level-Devil, and the CBS Audimax. The 6386 was a remote cutoff dual-triode, which made it ideally suited to push-pull gain control operation. A remote-cutoff tube has a grid that is wound in a nonlinear fashion and this gives the tube the unique characteristic of reducing its mu with increased signal levels. This was a valuable characteristic in a compressor or limiter. Conventional sharp-cutoff tubes tended to operate with substantially more distortion and artifacts. The Gates Sta-Level was a straightforward compressor, using the 6386 as a gain control element. A 6AL5 dualdiode was used to rectify a sample of the output from a pair of 6V6 tubes operating push-pull. This DC control voltage was fed, via an R/C time constant network to the grids of the 6386 tube. The circuitry was very similar to the GE Uni-Level, which preceded the Sta-Level by a year or two. The Gates Level Devil added a level dependent expansion gate that released about 10 dB of expansion when the input level was above “noise level.” This gate did not work very well and resulted in a lot of “sucking and wheezing.” However, those were the humble roots of intelligent audio processing.

The Gates Sta-Level, a solid performer used well into the 1970s


The Volumax 400. Together the Audimax and the Volumax
were known affectionately as “The Maxx Brothers.”

The year was 1959 when CBS Laboratories introduced the Audimax I. Designed by Emil Torick and marketed specifically as a “gain rider,” the Audimax I made no  pretense to being an “audio processor.” Yet, the unique design of the Audimax ushered in the era of audio processing. The Audimax I was also the first broadcast audio processor to be sold on a 30-day trial basis. A broadcaster could submit a purchase order for a unit and put it on the air for a month. If they were not happy with the sound they could return it at the end of the trial period with no questions asked. I am sure they got a few back, but the vast majority of users were quite satisfied with this box. I believe the original price was around a kilobuck (in 1959 dollars) and most of those who gave it a try became true believers in the Audimax concept.

The Audimax I

Just like in the Uni-Level and Sta-Level, the 6386 dual-triode was used as the gain control element, but with enhancements. A “platform mode” of control kept the Audimax gain constant over a 6 dB gain platform. This resulted in a lot less “busy-ness” in the sound. Should the input audio move outside the platform range, gain was quickly readjusted to define a new platform. The Audimax II quickly followed, which added an adjustable noise gate that froze the gain when input level fell below a user adjusted threshold. The Audimax II-RZ added a “return to zero” function which did just what you might think. It slowly returned gain to the zero gain point during periods when the input audio was below the threshold of the gate. The Audimax really started something and it transitioned into solid state versions and later added a biaseddiode peak limiter called the Volumax. The development of the Audimax was indeed a megaevent in the history of audio processing for broadcast. The “Maxx Brothers,” Audimax and Volumax, ruled for a decade and continued to be in demand for another ten years after that! Truly a remarkable record. In its later years, Thompsen acquired the Audimax line and produced a thin, one rack-unit version of the Audimax and Volumax.

There were several modifications applied by engineers who just could not accept the parameters that were fixed in the units. Most of these were attempts to speed up the release action, but there were many others. It seemed like every creative engineer had his own set of Audimax and Volumax tweaks. Basically, General Electric and CBS Laboratories ruled the roost when it came to state-of-the-art audio processing in the 1950s and 1960s. But nothing stays the same – and things were about to change big-time in the 1970s.

Go To Part #2