Orban’s first audio processor for FM broadcasting, the Optimod 8000, has turned 50. To mark the anniversary, Orban is sponsoring this series of interviews of Bob Orban in conversation with Radio World Editor in Chief Paul McLane.
In Part 1, Bob talked about his initial interest in audio, the development of an early stereo synthesizer and parametric EQ, the launch of Orban Associates and the introduction of the 8000.
Here is Part 2.
Paul McLane: What was the radio processing marketplace like as the Optimod came in?
Bob Orban: CBS Labs had come in and disrupted the market. Prior to that, if you wanted a compressor or limiter, you bought it from your transmitter manufacturer, such as a Gates Sta-Level or Level Devil, or an RCA BA6A.
CBS Labs developed innovative technology, including the idea that the final gain control element in the limiter should be just a clipper, and the limiter should have a relatively moderate attack time, typically around 2 milliseconds, so you didn’t get pumping and hole-punching on every little transient. They also put serious marketing behind it.
So for high-end stations, the standard audio processing chain was an Audimax and Volumax, and typically the stereo generator built into your transmitter, which had to be type accepted. A few stations had custom processing chains.
To disrupt the market, we would have to do more than incremental improvement.
The Optimod 8000 had non-overshooting low-pass filters and a system approach that eliminated a whole chain of possible operator error that could happen when interconnecting compressors, limiters and stereo generators and getting wrong levels — too much noise, or clipping. The 8000 also got rid of a bunch of potential transformers in the chain. It was very clean, with only about 0.1% distortion even in Operate mode. And, with Eric Small placing a lot of 8000s in influential major-market FM stations, we were selling them by the hundreds — something like 3,000 by the time we replaced it with the 8100.
McLane: Were you aware at the time that you were doing something pretty dramatic?
Orban: Eric Small’s main gig was as a loudness consultant for major-market stations. He had a rack full of measurement equipment including one of the original CBS loudness meters. He could do objective measurements of how much louder he could get a radio station.
He verified that the 8000 really did give you about a 3 dB edge compared to the older chains — in fact, we used that as a marketing slogan, “The 3 dB Edge.” That was a big deal. If one or two stations in a market got one, everybody else had no choice but to follow along to stay competitive.
We succeeded quickly, and the operation became self-funding. And, as I mentioned last time, this was also when Orban Associates incorporated and our former contract manufacturer John Delantoni became a 50% partner with me.
McLane: If we could beam back into the mid-1970s and listen to processing on the radio dial, what would we hear?
Orban: For one thing, it would be a lot quieter, even with the 8000s.
A friend of mine used to joke that with the FM Volumax 410, cymbals sounded like trash can lids because of their high-frequency limiter design and pre-emphasized clipping. High frequencies didn’t sparkle, and they generally didn’t sound very clean. Radio would have sounded duller.
Because of clever gating, stock Audimaxes were pretty smooth — there wasn’t a lot of pumping or objectionable artifacts, although some engineers hot-rodded them to speed up the release time, and you could get pumping for the sake of loudness. If you wanted to stand out, you had to put together a custom chain in front of your peak limiter. A few engineers like Bob Kanner at RKO were doing that with their AM chains.
There was a lot of audible high-frequency loss due to the 75 microsecond preemphasis curve, and nobody had a very good solution for that yet. But processing would probably seem conservative compared to what happened later.
After the 8000, things got substantially louder. High frequencies sounded nicer because it was a very clean chain. The problem, again, is that in the major markets, people started putting stuff in front of the 8000. When you do that, there’s the danger of dueling time constants, pumping and various unmusical things happening for the sake of loudness.
This is eventually what brought about the 8100.
McLane: What was happening elsewhere in the market? I know Dorrough played a part then.
Orban: Yes, the Discriminate Audio Processor or DAP 310 was a common pre-processor for Optimods. It was a pretty benign box, smooth-sounding, and tended not to do a bunch of damage.
Also, Harris-Gates had Dave Hershberger working for them. Dave is an extremely bright guy, and he came up with a good overshoot compensator for the Harris stereo generator. I had to respond to that, which I eventually did in the 8100. Harris also came up with a three-band processor that didn’t make a big splash.
Pacific Recorders had the Multimax, a Jack Williams production. Glen Clark and Texar weren’t yet in business.
Circuit Research Labs started in the 1970s, and Ron Jones had a good set of ears. They made competitive boxes. But I had patents on the 8000, so none of them could directly play in that sandbox. The closest was Dave Hershberger’s non-overshooting low-pass filters for the Harris stereo generator. So, a lot of stuff ended up being put in front of 8000s.
We had money coming in now, so I started working on the 9000. I figured that if we had disrupted the FM processing market, AM was the next obvious step.
McLane: It came out in 1978.
Orban: The 8000 had a relatively simple processing chain; it used the FET compressor/limiter I developed at Stanford, a fairly straightforward preemphasis limiter, the clipper and the overshoot-compensated low-pass filters.
But the 9000 introduced important new concepts. There was a lot more custom processing in AM than in FM, and I was competing against custom racks like the ones used by Bob Kanner for KFRC and KHJ and Ed Buterbaugh at CKLW. I realized that multiband compression would be essential to achieve loudness and density competitive with these custom chains, and I settled on six bands.
The typical AM radio back then rolled off so that it was –3 dB at 2 kHz. That’s a lower bandwidth than a landline telephone. Without some sort of treatment, AM had speech intelligibility problems.
One of the goals of the 9000 was to implement preemphasis to equalize some of this rolloff. We measured a number of AM radios and derived a typical AM radio frequency response. Then I developed an equalizer that was complementary to the radios of the day, up 3 dB at 2 kHz and around 15 dB at 5 kHz. When you take a signal like that and put it into a clipper, you get high-frequency distortion, particularly sibilance. Esses start sounding like effs, so I realized that that wasn’t going to work.
I had the idea that, because AM radio is roughly flat only to 2 kHz, I could cancel the clipper distortion in the zero to 2 kHz frequency range by low-passing the clipper’s distortion component, then subtracting from the clipped signal. I came up with a technique of doing that using a phase-corrected 2 kHz low-pass filter.
This also was the start of serious computer-aided design at Orban. (In fact, I did a presentation for the SBE Ennes Workshop at the NAB Show this year on how computers were an intrinsic part of the Orban design process.)
So, I developed a phase-corrected 2 kHz low-pass filter using computer-aided design, and that required a 400 microsecond delay. I decided to use an analog bucket brigade delay line as the delay compensation element. These were mostly used for things like guitar effects pedals; they were widely available. I remember breadboarding the thing, measuring it and verifying that the distortion cancellation indeed worked.
One of my test tracks was Jefferson Airplane’s “White Rabbit,” which has sibilant vocals — “One pill makes you larger, and one pill makes you small” … with a big “ess” on the “small,” which had broken up in the earlier breadboards of the 9000. I ran it through this new distortion cancellation, and lo and behold, all the distortion just went away. It was jaw-dropping, very memorable. I realized I really had something here.
Also, I had to think about what to do about the distortion between 2 and 5 kHz — at 5 kHz the radios were typically down 15 to 20 dB, so the distortion didn’t matter much above 5 kHz.
I worked out a method of using a psychoacoustic masking model to measure and quantify whether the distortion was audible in the 2 kHz to 5 kHz region. That involved a set of third-octave filters and measuring the ratio between the undistorted and distortion-only signals in the third-octave bands, between 2 and 5 kHz. We called that the Smart Clipper and used it to control the amount of clipping depth that was permitted to occur.
Another innovation was a soft-switching polarity follower to ensure that when the audio was asymmetrical, the higher side modulated the transmitter in the positive direction. Previous technology for this used a hard switch with audible clicks. The 9000 used a sweepable allpass filter to change the polarity smoothly.
The clipping distortion cancelation and psychoacoustically-driven distortion control had never been done by anyone else. I figured that this, the receiver equalizer, the six-band limiter and the click-free polarity follower were enough to create a market-disruptive processor.
It was very competitive with the CRL stuff. Ron Jones’ big idea was the transmitter equalizer, which originally came from Oscar Bonello, an innovative Argentinian engineer who had a company called Solidyne Labs. He was working in a sort of South American bubble. His stuff was popular in Latin America but never got much of a foothold here in the U.S. and Canada. But he was a very smart guy, and I believe he was the first one to do the transmitter equalizer.
Ron Jones picked that up, and I picked it up as an upgrade for the 9000, which took us from the 9000A to the 9000B. Fortunately it used card-cage construction, so we easily implemented field upgrades with little disruption.
The 9000 was the final project in that era, and then I started R&D on the 8100.
McLane: I have read that the 9000 offered AM stations “a more FM-like sound and quality” and that it reduced interference. Is that a fair summary?
Orban: Yes, we built a 12 kHz low-pass filter into it, non-overshooting. I believe it was the first AM processor on the market to have a built-in filter. We were able to do all of this preemphasis without causing excessive second-adjacent interference because the 12 kHz filter allowed energy only above 3 kHz on the second adjacent, and at 3 kHz, the radios are already rolling off.
McLane: Didn’t the 9000 influence the NRSC-1 standard that followed about a decade later?
Orban: The 9000 was the first AM processor to use a formally designed receiver equalizer, and later we did a second iteration of that for the 9100. Greg Ogonowski, who was a friend but also a competitor with the Gregg Labs processors, also worked on that.
We co-authored a paper for the audio group of the Society of Automotive Engineers and presented it at one of their conventions. It suggested a standardized AM preemphasis that was complementary to the AM radios of the time, up to about 5 kHz. Our proposed preemphasis used an 18 dB/octave slope. It was a third-order shelving filter.
When it came time for the National Radio Systems Committee to propose its preemphasis, there was some disagreement between some of the receiver manufacturers and processor manufacturers. We agreed that we needed to be up 3 dB at 2 kHz, because that was complementary to the average radio frequency response. But the eventual NRSC preemphasis was only a first-order preemphasis, actually much gentler than the one in the 9000 and later the 9100.
Greg Buchwald of Motorola had proposed that as a compromise between broadcasters and the receiver manufacturers. They didn’t want any preemphasis at all, because they were fighting first-adjacent interference. They were concerned that some customers were bringing radios back to the dealerships, complaining about so-called noise, some of which was actually first-adjacent interference at night. NRSC pre-emphasis was a reasonable compromise that helped speech intelligibility but had less potential for creating first-adjacent interference.
The receiver manufacturers also wanted a 5 kHz lowpass filter, which would have completely protected the first adjacent. The second part of the NRSC standard, the 10 kHz lowpass filter, was a compromise between the receiver manufacturers and broadcasters, some of whom wanted to preserve the 15 kHz bandwidth limit in the old FCC rules. 10 kHz protected the second adjacent, but not the first.
McLane: What was the manufacturing process at that time?
Orban: My business partner John Delantoni had been the contract manufacturer for Orban/Parasound products, which were built in Silicon Valley. We continued to use some of the same people but eventually we brought the manufacturing in-house in the late ’70s.
Everything was being done in the San Francisco Bay Area. Circuit board manufacturing was done in Santa Clara and final assembly and test was done in our main Bryant Street facility in San Francisco. At the time, I believe we had around 30 employees, mostly in purchasing, manufacturing, final test and customer service.
Because of our success with the 8000 and 9000, funding was coming in to the point where we had the luxury to catch our breath and to do really good R&D and not rush products to market. One of my philosophies was “don’t serve the dish until it’s fully cooked.”
McLane: I very much appreciate your taking the time and look forward to continuing this conversation next time with the 8100.
Orban: My pleasure. Now that I’m pushing 80, I’m glad a lot of this is being memorialized, because I won’t be around forever. Time comes for all of us, so it’s good to have this oral history in place.
