Aeranti-Corallo raises interference concerns at Italian World Radio Day event
Bob Orban has spent decades solving the audio and broadcast industry’s most pressing technical and creative challenges. In this Q&A, we delve into those that mattered most and the solutions that helped shape the industry.
RedTech: When you started in the early 1970s, what was the biggest problem you saw in broadcast audio that needed fixing?
Bob Orban: Compared to a theoretically optimum solution, overshoots in FM transmitters’ low-pass filters were causing uncontrolled peak modulation that forced engineers to reduce average modulation by 3 dB or more to maintain legal carrier deviation. This loss of average modulation limited signal coverage and caused unnecessarily noisy reception in weak-signal areas.
Many major market broadcasters were using what I considered inartistically large amounts of audio processing to make the best of this situation, but it was compromising FM’s “high fidelity” potential.
RedTech: The Optimod 8000A put you on the map in 1975. What challenge were you trying to solve with that product?
Orban: The 8000A incorporates non-overshooting nonlinear low-pass filters that permit 3 dB more average modulation. Additionally, the most popular audio chain of the time had a total harmonic distortion of well over 1%, whereas the 8000 has a full order of magnitude less. Because the 8000 incorporates the entire audio chain from the console (or STL) output to the composite input of the FM exciter, multiple, quality-compromising audio transformers were not needed. Moreover, the gain relationship between the elements in the 8000’s audio chain (compressor, high-frequency limiter, peak limiter and stereo encoder) is fixed. In contrast, in traditional chains, these gains were often adjusted non-optimally. Finally, the 8000’s audio processing is straightforward to set up: input levels are set using the unit’s meter and overload indicator, and the only control affecting the compressor characteristics is a release time adjustment. This simplicity guards against maladjustment.
The most significant thing I did was to disrupt the FM audio processor market with a radically different concept.
RedTech: The pursuit of loudness became a defining challenge in radio broadcasting worldwide. How did you approach that problem?
Orban: My initial engineering and marketing goal with the 8000A was to “bring high fidelity back to the FM band.” Inevitably, some broadcasters added processing ahead of the 8000A, which sabotaged its clean, easy-to-listen-to sound.
My first response was the 8100A. It incorporated our first multiband compressor, but with a novel band-coupling configuration that allowed continuously variable adjustment between quasi-wideband operation (to preserve source frequency balances) and full multiband operation (which could yield more loudness and consistency). The non-overshooting low-pass filters offered improved high-frequency headroom and lower distortion. Most importantly, the 8100A implemented our distortion-canceled clipper very cleanly by using the primary 15 kHz low-pass filter for delay compensation—a game-changing upgrade from the 8000A’s simple clippers.
But program directors wanted it louder, so pre-processors again appeared ahead of the 8100A. A cottage industry developed to support pre-processing by replacing the 8100’s gain control card.
My response was the XT2 six-band limiter accessory chassis. Interleaved into the 8100A circuitry, it repurposed the two-band compressor as a two-band AGC and craftily used the existing distortion-canceling filter circuitry to cancel distortion in individual limiter frequency bands. The XT2 was easy to install. As an integrated system, it shared the same advantage as the 8000A: minimization of potential setup error, and it enjoyed great market success.
RedTech: Making the leap from analog to digital processing was a major turning point. What made that transition so difficult?
Orban: Discrete time signal processing presents two main challenges: aliasing distortion and filter frequency warping. In digital signal processing, filter design equations are complicated by additional trigonometric terms that warp the DSP frequency response compared to its analog counterparts.
I was an expert analog designer but knew little about digital signal processing. Around 1980, I acquired several textbooks and started studying DSP. I set a “thesis” goal for myself: developing a computer-aided design program that would provide minimum-error recursive DSP approximations to analog filters, which were needed to implement FM pre-emphasis (among many other applications). I succeeded; we have used this program extensively and advantageously since then.
Aliasing distortion is easy to predict and control when the processing is linear (e.g. filtering). However, DSP peak limiters and clippers are troublesome because they are highly nonlinear. They introduce harmonic and IM distortion that can alias down into the audible frequency range, causing unpleasant inharmonic distortion.
In our first DSP-based processor, the 8200, we addressed the problem by oversampling the clippers by four times (to 128 kHz), which we chose after listening to 1x, 2x, and 4x oversampling. However, we were getting heat from a competitor regarding aliasing, so I developed and patented a method of anti-aliasing clippers, which we first introduced in the 8400 and subsequently used.
RedTech: Can you share an example of a product challenge where the solution wasn’t what you initially expected?
Orban: When developing the Optimod-AM 9000A, I had assumed that I could continue to use simple clippers as I had for the 8000A. However, the main goal of the 9000A was to implement pre-emphasis, equalizing the frequency response of the 1970s AM radios, which was typically down 3 dB by 2 kHz — poorer than a landline telephone. The pre-emphasis caused a significant high-frequency boost in the midrange, and applying this signal to a conventional clipper caused severe intermodulation on vocal sibilance, with “s” sounds turning into “f.”
My solution was the distortion-canceled clipper, which low-passes the clipper-induced distortion at 2 kHz and subtracts it from the main signal to eliminate clipper-induced distortion in the range where the receiver has a flat frequency response. It exploits the receiver’s high-frequency rolloff to minimize the audibility of the remaining distortion. Successful subtraction requires matching the delays in the main signal path and the 2 kHz low-pass filter, which we achieved in the 9000A with a bucket-brigade delay line and in later products by using the delay of the low-pass filter in the main signal path.
One test track was Jefferson Airplane’s “White Rabbit”: “one pill makes you larger, and one pill makes you small.” The “s” on “small” always distorted, but when played through the 9000A’s newly breadboarded clipper, the distortion vanished like Alice’s rabbit. It was the most dramatic subjective breakthrough I had experienced in my career — one that was unforgettable.
Additionally, the 9000A incorporates a psychoacoustic masking model that continuously estimates the non-cancelled distortion’s audibility and constrains the clipping depth accordingly. I later revisited this principle (with a very different implementation) when I developed the MX limiter for Optimod-FM 8600 around 2010.
RedTech: You’ve been involved in industry standards committees for decades. How does that collaborative work differ from solving problems on your own?
Orban: I originally designed solo so that I could work quickly. Even when we hired more engineers for our DSP initiative, I designed and subjectively tuned the algorithms with just a few collaborators (most notably Greg Ogonowski), facilitating fast decisions.
Committee work is slow. There are often strong, contradictory viewpoints (particularly during the development of the National Radio Systems Committee NRSC-1 standard), and working toward consensus requires technical arguments, persuasion and trading off of participants’ interests. I heard, understood, and helped balance conflicting positions, which broadened my own thinking. Ultimately, NRSC-1 was an exercise in technical and economic politics where no one got everything, but everyone eventually accepted a compromise forged in a spirited exchange of expert opinions.
Other committees facilitate industry education. I was recently co-editor on an AES committee that created a Technical Document (TD1009) to elucidate the causes of poor dialogue intelligibility in many modern productions and to offer solutions. We assembled subject matter experts from psychoacoustics to consumer electronics and poured their collective expertise into a coherent and clearly presented document that industry people and even consumer enthusiasts could use productively. This took about two years.
Our customers seek more integration, such as incorporating ratings encoders and streaming encoders within the processor’s hardware platform.
RedTech: How has the shift to software-based audio processing changed the kinds of problems you’re solving?
Orban: We have offered “software-based processing” ever since Optimod-FM 8200 — DSP chips run software. Our previous hardware DSP-based processors all utilized Motorola (now NXP) 56000-series DSP chips, which were programmed in assembly language by in-house experts. While I contributed heavily to the algorithm design of the 56k-based processors, the first Orban product whose DSP I personally coded is Optimod-PCn 1600, which runs natively in x86 processors and is coded in high-level language. I also coded the XPN-AM DSP. My specialty is algorithm design, so the hardware platform on which software-based processing runs has had relatively little effect on my work.
The other current trend is distributed or “cloud” computing. The challenges here are reliable processor containerization and low-latency input/output, but this is irrelevant to signal processing algorithms.
Additionally, our customers seek more integration, such as incorporating ratings encoders and streaming encoders within the processor’s hardware platform. While placing the ratings encoder optimally in the signal path requires touching the processing algorithms, implementing the rest of the features requires a somewhat different skill set and is capably handled by others in our engineering team.
RedTech: Looking back over 50 years, which challenge are you most proud of solving? And what problems are you still working on?
Orban: The most significant thing I did was to disrupt the FM audio processor market with a radically different concept. When we started Orban, CBS Laboratories dominated with their Audimax/Volumax combo. They had excellent marketing and a well-established brand. It took more than an incremental improvement to dislodge them. We were fortunate to work with Eric Small in technical marketing, as he specialized in making major-market stations loud. He appreciated how the 8000A could improve processing and had the connections to place units with top stations. The “3 dB edge” was more than a marketing slogan — we backed it by rigorous measurements using the rack of equipment that Eric had assembled as a processing consultant.
Our second major advance was the 8200. While a few DSP-based broadcast processors predated the 8200, the 8200 was the first one capable of displacing analog processing while adding digital’s inherent controllability, repeatability, and preset support. It was the first DSP-based broadcast processor to achieve commercial success, with approximately 5,000 units sold. Its LESS-MORE control was an important innovation, allowing even non-experts to tune the processing successfully.
My most important algorithmic invention was probably the distortion-canceled clipper (including its multiband variant). It turbocharged processing for loudness, and its principles are almost ubiquitous today.
There are some other processors and algorithms that I am fond of. Optimod-HF 9105A, with our “Hilbert-Transform Clipper,” became the industry-standard processor for international shortwave broadcasting and remains popular among amateur radio operators. The half-cosine interpolation composite limiter (introduced in the 8400) is our take on the composite clipper, providing its modulation-control advantages but with lower subjective distortion and full stereo separation. The MX limiter, developed for the 8600, builds on distortion-cancelled clipping while elegantly controlling audible distortion using a psychoacoustic model.
As for “what’s next?” — well, a peek into the secret laboratory would destroy its secrecy! But I encourage readers to keep watching Orban. There’s plenty more to come.
This story originally appeared in the RedTech special edition, Solutioneers 2025/2026. You can read or download this edition for free here.
You can view and download all RedTech magazines and special publications for free here.
These stories might interest you
The Solutioneers 2025/2026 builds resilience for a changing industry
Thai Government Radio upgrades FM transmission with Orban Optimod 5950
