Optimod 8200 Question

[OKCRadioGuy]

"gettinbyagain", You've got mail.

 

[OKCRadioGuy]

;D

 

[OKCRadioGuy]

Your P.M. has some settings for a pretty succesful CHR

 

[FFoti1]

In case it wasn't clear by the context, I was comparing 1992 Optimod technology (the 8200) to 1995 Omnia technology (their first digital model)... not to 2010 Omnia technology. Obvious the processing performance of both brands has improved tremendously since then!

Except your original post was a generalized statement, with no clarification, and that could lead to a lack of understanding by others. The entire thread, up until your post, was about a question regarding the 8200. There was no need to hijack the thread in another direction.

-Frank Foti

 

[Sgeirk]

My first impression, which has left a lasting impression, of the 8200 was "THIS is an improvement over analog processors?!?"

This impression lasted until about 2002 when I heard a digital box that finally bested analog's best efforts.

 

[rrsounds]

My first impression, which has left a lasting impression, of the 8200 was "THIS is an improvement over analog processors?!?"

This impression lasted until about 2002 when I heard a digital box that finally bested analog's best efforts.

I was not happy with the sound of the original 8200. Other people have had a completely 180 experience.

But I will say this: I never saw one crash. ;-)

Kind Regards,
David

 

[rorban]

My first impression, which has left a lasting impression, of the 8200 was "THIS is an improvement over analog processors?!?"

This impression lasted until about 2002 when I heard a digital box that finally bested analog's best efforts.

I was not happy with the sound of the original 8200. Other people have had a completely 180 experience.

But I will say this: I never saw one crash. ;-)

Kind Regards,
David

We sold about 5000 8200s and I know from customer feedback that most were happy. Many came back again and again to buy for other stations. But we were aware of the limitations of the 8200, particularly in the area of clipping distortion control. The parameters of the multiband compressor/limiters had to be tuned with extreme care because this tuning was the only thing preventing certain program material from causing objectionable clipping distortion. This is why most of the controls were hidden--changing them could cause the processing to distort in objectionable ways on certain program material.

By the late '90s, we had much more DSP power available at a cost that our customers were willing to pay, plus we had a decade more research into digital processing. The main result of this research was much better clipping distortion control (which in turn made it safe to expose more of the tuning controls to the user) and a composite limiter that did not audibly compromise the stereo image. The 8400 was the result of that R&D and in my opinion represented a big jump over the 8200.

The 8500 refined the 8400's sound, but its improvement was evolutionary. Meanwhile, for the past three years we were working on new peak limiting techniques and algorithms with the goal of improving high frequency power handling capability and distortion control while also improving transient impact, which is usually the first victim of aggressive distortion control techniques. The resulting peak limiter alone takes about twice the DSP power available in an entire 8400, so we are grateful that Freescale Semiconductor has continued to improve its chips based on the original Motorola 56000 DSP architecture. The result of this research will first appear in the 8600. With the 8600, I believe we have made a jump are large as the one we made going from the 8200 to the 8400.

I feel lucky to be involved in a fun industry where the major semiconductor manufacturers continue to give us more power at lower cost. Were now at the point where there is so much DSP power available at reasonable cost that almost anything that a processor designer can imagine can be realized at a commercially feasible price. I feel no nostalgia for the days of analog--DSP is extremely liberating.

Bob Orban

 

[cgould]

I feel lucky to be involved in a fun industry where the major semiconductor manufacturers continue to give us more power at lower cost. Were now at the point where there is so much DSP power available at reasonable cost that almost anything that a processor designer can imagine can be realized at a commercially feasible price. I feel no nostalgia for the days of analog--DSP is extremely liberating.

Bob Orban

I totally agree!!!

-C

 

[dannyscott101]

Another thread about processing with Frank, Bob & Cornelius all on board, so I'm going to hijack it in a different direction for a minute.

Though I have not heard the new boxes yet, (O.11 & 8600) it seems like the design philosophy of both is headed in a similar direction... (I know the details may be different, so keep the gloves on for a minute guys!) ...that direction being to drastically improve the quality of FM while maintaining the current "loudness level" that has become today's accepted standard... Also, to make the processing a more transparent process than ever before while maintaining that loudness level. (as per the Omnia campaign, "more revealing")

While I look forward to hearing these results from both camps, there is still a part of me that embraces the 'classic' sound of audio processing, and some of that includes things like bass-induced IM distortion.

Next weekend I have the fun task of programming some 80s music for my high school reunion. I have decided to make it a 'retro radio show'. In my quest for the perfect sound, I've been tweaking my Compellor/Prism/8100A combo for just the right nostalgic sound. I'm using just a touch of the prisms (they are modded to eliminate the gating feature so I can use just a small amount of G/R) and giving the compellor some audible 'slam'.

I want to hear the processor 'clamp down' on those attacks in the bridge of Journey's "Separate Ways", and I want to hear some of that reverb get sucked up in Joan Jett's "I Love Rock & Roll". I'm sure I will be the only one at the party who will care, but it brings me back to the old radio days.

If you have the time, check out these two threads, one started by me, about the station that was responsible for my early obsession with audio processing...

http://boards.radio-info.com/smf/index.php?topic=162360.0

http://boards.radio-info.com/smf/index.php?topic=148094.0

While today's processing is getting more refined and transparent, I think there is still a human desire to hear a hint of the sound we associate with 'radio'...

If you don't believe this, just look at the visual world. In spite of HDTV, we now have video cameras that shoot at or simulate 24P, and have grain and curve settings to simulate the look of FILM... Keep in mind, this is an intentional reduction in image quality to simulate a medium that is a century old. I think certain formats in radio deserve the same thing, if only for the subtleties that bring it to life.

and no, I'm not going to drag out an old audimax/volumax and put it on the air!

 

[FFoti1]

While I look forward to hearing these results from both camps, there is still a part of me that embraces the 'classic' sound of audio processing, and some of that includes things like bass-induced IM distortion.

I think you'll be pleasantly surprised with the abilities of Omnia.11. Reason being is you can create a 'classic' sound, as you note...but...do it minus the sonic annoyances associated with prior generations of boxes. There are some who do indeed like the 'sound' of processing...we hear you. I believe you can achieve that without the undesirable effects that are normally associated with a signature that sounds processed.

The challenge we face today is IM induced distortion is audible not from bass alone. Hence the migration to advanced IM management functions, which enable us to retain a signature which contains less distortion. Think of being able to listen to a 'classic' radio signature of yesteryear, be able to maintain the signature, but reduce the annoyances that would have been associated with that level of processing.

-Frank Foti

 

[rorban]

Another thread about processing with Frank, Bob & Cornelius all on board, so I'm going to hijack it in a different direction for a minute.

While today's processing is getting more refined and transparent, I think there is still a human desire to hear a hint of the sound we associate with 'radio'...

Since we started designing and selling transmission audio processors since 1975, our customers have made us very aware that they have a large gamut of preferences and processing goals. Because of this, we designed the 8600 so that users who like the some aspects of the older Optimod processing's sound can get it without compromise. The 8600 will run 8500 and even quasi-8200 algorithms (in the form of our so-called "ultra-low latency" presets, which we introduced a few years back). Moreover, the 8600 presets that use the new peak limiting algorithm have "flavor" controls that allow dialing in controlled amounts of distortion if that is what a user is looking for.

Regarding distortion, I prefer not to casually refer to "IM distortion" or "harmonic distortion" without carefully defining these terms. Audio engineers created these terms when they first started trying to quantify the nonlinear distortion produced by transmission channels like AM radio or optical film soundtracks. "Harmonic distortion" was measured as the r.m.s. sum of the nonlinear distortion products (which were typically harmonics) caused by single sinewave excitation, while "IM distortion" was typically measured with dual sinewave excitation and was the r.m.s. sum of certain nonlinear distortion products whose frequencies were not integer multiples of the input sinewaves.

The results of these distortion measurements were a few numbers that attempted to characterize in a simple way how nonlinear systems would behave when passing real-world program material. But because superposition does not hold in nonlinear systems (and broadcast audio processors are strongly nonlinear), using these sinewave-based measurements to predict how such systems sound with real-world program material is iffy at best. (Superposition defined: if you apply A to a system and its output is X, then remove A and apply B and get output Y, superposition holds if and only if applying A+B to the system yields nothing more or less than X+Y at its output.)

All peak limiters produce substantial amounts of nonlinear distortion; it's inherent in the job that they are required to do. If you examine the nonlinear distortion that a peak limiter produces with complex program material, it looks almost like noise on a spectrum analyzer. There are a huge number of frequencies that are not present in the original program material, all of which might considered "IM." The most useful definition of "harmonic distortion" in this context is the r.m.s. sum of frequency components in the distortion spectrum that are integer multiples of the Fourier components of the signal driving the limiter. Given these definitions, "IM distortion" and "harmonic distortion" can be either good or bad depending on the program material and, to a certain extent, on the preferences of the listener.

To me, the most interesting questions regarding limiter-produced distortion are these:

1. At any given moment, is the nonlinear distortion psychoacoustically masked by the input program material, and, if not,

2. esthetically, do the unmasked components complement the program material or clash with it?

The first question can be answered objectively (people with normal hearing can either hear the distortion or they cannot, and this can be tested with an ABX or ABC/hr test), while the second is a matter of taste and preference.

The art of peak limiting lies mostly in psychoacoustically "hiding" as much of the distortion as possible while ensuring that the remaining audibly detectable distortion is as pleasing to the ear as possible. Sometimes "IM distortion" is preferable to "harmonic distortion"--for example, much of the unpleasant sound of heavily clipped speech is caused by "harmonic distortion." The spectrum of speech vowels rolls off quickly at high frequencies, so the harmonics produced by the clipper are far removed in frequency from the input producing them and thus tend to be poorly masked psychoacoustically. In this case, if the "IM distortion" components are closer in frequency to the undistorted input and are thus better masked by it, then producing more "IM distortion" and less "harmonic distortion" will probably be more pleasing to the ear. This was the original purpose of Orban's "Hilbert Transform Clipper," which was first introduced in Optimod-TV 8182 television audio processor. This invention allowed the designer to manipulate the frequency spectrum so that certain input frequencies produced both harmonic and IM distortion while others (in the frequency range of the vowels in speech) produced IM distortion only. This peak limiter is more flattering to speech material (which predominates in television audio), while the distortion-cancelling clipper first introduced in Optimod-AM 9000 and later refined in the 8100 is more flattering to music. In fact, one of the long-standing challenges in designing the part of the processing that precedes an 8100-style distortion-cancelling clipper is to control speech distortion while minimizing the loss of punch on music that usually accompanies such attempts.

It wasn't until the 8400 that we found a solution in DSP that significantly improved this tradeoff -- an "intelligent" distortion controller/peak limiter that adapts more effectively to the program material than anything we had been able to do with analog technology. The 8600 steps up this game substantially by using a more sophisticated analysis combined with newly designed and structurally novel algorithms to do the limiting. The 8600's peak limiter is more effective at "hiding" limiter-induced distortion than are earlier Orban algorithms and when it is impossible to hide the distortion while still controlling peak levels, the limiter attempts to ensure that the resulting audible distortion degrades the program material in a way that is as "attractive" as possible. With material whose transients have not been beaten until dead by the mastering engineer, there is a large difference in transient impact between the 8500 and 8600. Even with this over-processed material, the improvement in high frequency clarity is not subtle--we measure 2 to 3 dB more energy above 10 kHz with modern mixes having substantial amounts of HF energy. The details of how we did this must remain proprietary, but we hope that "the proof is in the listening."

Bob Orban

 

[BabyDJ]

While I look forward to hearing these results from both camps, there is still a part of me that embraces the 'classic' sound of audio processing, and some of that includes things like bass-induced IM distortion.

While today's processing is getting more refined and transparent, I think there is still a human desire to hear a hint of the sound we associate with 'radio'...

If you're in NYC, check out CBS FM. They've captured this sound perfectly and it's very very very clean on top of it. I'm talking blaring horns from Chicago and even the edgy sound Spector stuff sounds great. 80's music jumps out and whatever they're doing to process the phones... they have the best sounding phones anywhere.

Lite FM plays alot of what CBS FM plays and it sounds so bad on Lite. On CBS, it jumps at you and is a real treat to listen to. I've never heard processing like this keep and retain such detail. I remember listening to Bill Lee on KTU and the processing was all over the place. On CBS, it's all right there, perfect levels ever break.

Their engineer was talking on the NY about some modified stuff in his chain and how he works with the production guy on getting the right source material in. It's great stuff! This is the kind of guy I want designing my processing.

http://boards.radio-info.com/smf/index.php?topic=171409.msg1478313#msg1478313

 

[dannyscott101]

With Orban boxes I personally also prefer the slow AGC and faster MB release combination. It tends to give more consistent sound from cut to cut without the breathing effect of the AGC. This will have more limited operating range, meaning you have to be more careful with your levels, but a good DJ behind the board can really play against the processor which can sound excellent!

However, some people prefer the more noticeable action of the faster AGC release. It tends to correct levels more but also adds some breathing to the sound. Interestingly, what may be annoying to some, others actually prefer! If it's not obviously sucking up in pauses, then it must not be working enough! They may also like the more open sound and a bit more volume you get with the slow release time in the multiband.

Like we said before, it's all subjective


Regards,
Goran Tomas

In spite of my desire to hear a bit of 'pumping' and 'slam' in my retro processing, I am using more of the multiband on the air at the REAL radio station. (Keep in mind that I'm using the Optimod-TV 8282 and Optimod 2200 combo as per my previous thread on this started back in February, but the most of the speed settings also apply to the 8200)

If you can ignore the fact that we're clearly not as loud as the competition, it really does have a desirable "big" sound. I had the mutliband release at medium-fast and the AGC release at '4', with about 10dB of multiband G/R. The programming and production guys liked the sound, but wanted it to pick up the low song intros quicker during automation hours. I reluctantly increased the multiband to fast and the AGC release to '7'. I also added about another 3dB of multiband G/R. Obviously, it sounds much more aggressive now.

I'm not actually using a whole lot of AGC before the multiband. It probably maxes at about 5dB of G/R. I didn't like the way the gating worked on the 8282 so I turned it OFF!... It freezes all the bands instead of just the AGC, and I could hear it 'correcting' when the audio would return after a pause. I don't know if the 8200 also works this way. In any case, it still does not cause too much 'suck up' effect.

In spite of the aggressive processing, I have kept the EQ a little 'warmer' sounding (fewer mids and a bit less presence) than everyone else on the dial. (at the expense of even more loudness than I'm limited to already) Now when punching to the competition, they suddenly sound loud, bright and full at first... but in my opinion the harsh presence sound quickly induces fatigue. It also seems like pushing the mid-highs tends to emphasize the extra distortion in today's hypercompressed material.

Yes, while you guys play with your shiny new boxes, I'm scrounging your dumpsters and ebay to find the next piece of "junk" to put on the air! That, and keeping my fingers crossed that it doesn't fail in the next couple of years or so... There's always the backup 8100.

 

[fm-engineer]

Bob.

Any plans on putting some audio comparison clips on the Orban site demonstrating the new peak clipper?

Another thread about processing with Frank, Bob & Cornelius all on board, so I'm going to hijack it in a different direction for a minute.

While today's processing is getting more refined and transparent, I think there is still a human desire to hear a hint of the sound we associate with 'radio'...

Even with this over-processed material, the improvement in high frequency clarity is not subtle--we measure 2 to 3 dB more energy above 10 kHz with modern mixes having substantial amounts of HF energy. The details of how we did this must remain proprietary, but we hope that "the proof is in the listening."

Bob Orban

 

[W1DAN]

Bob:

To hijack this thread a moment....

Thank you very much for explaining some of your design philosophy. As you may know, I enjoy studying analog processing technology and the use of clippers...diode and "Hilbert" style. Your 9105A shortwave processor was a very unique box that used frequency shaping, clipping and IM generation to create a punchy SW signal. Careful clipping easily separates the better analog processors from the toys as the careful use of very short duration clipping can increase the loudness of a signal than not using clipping. If you wish to share more tidbits about old obsolete products, it would be appreciated, particularly the reasons you designed the 9105A (was this funded by the VoA or BBC??) and what characteristics you learned from previous boxes and techniques.

For fun, right now I am working on fine tuning an 8182A SAP for ham use....

Now back to your regular programming!

Thanks,
Dan

 

[rorban]

Bob.

Any plans on putting some audio comparison clips on the Orban site demonstrating the new peak clipper?

This is a good idea and we will get on it in the next couple of weeks.

Bob Orban

 

[rorban]

If you wish to share more tidbits about old obsolete products, it would be appreciated, particularly the reasons you designed the 9105A (was this funded by the VoA or BBC??) and what characteristics you learned from previous boxes and techniques.

The 9105 was not funded by an international broadcasting organization; it was developed in-house using our normal R&D budget. As to its genesis, we became aware (through both dealer and customer requests) of the need for advanced audio processing to complement the newly installed, very high-powered HF transmitters (250 and 500 KW) that were coming on line. These transmitters were mostly manufactured by Marconi and Brown Boveri and used variations of PWM technology to do the modulation.

Meanwhile, we had completed the design of the 9100 and were having good success with it in the field. It seemed obvious to me to create a hybrid processor that used most of the technology we had developed for the 9100 but combining it with our Hilbert Transform Clipper technology (first developed for the 8182 television processor) to optimize the processing for speech, which was the predominant program material in high-power international shortwave broadcast.

We sent the first units out for trial with various international broadcasters and they were generally pleased with the results, although in some cases the very high average modulation produced by the 9105 challenged the new transmitters and transmitter modifications were necessary to correct this. We were given some very specific constraints, the most entertaining one being the need for a 25 Hz notch filter because high modulation at 25 Hz tended to cause one of the transmitter models to produce fireworks! (These were tube-type transmitters and used very high voltages.) We responded to the special needs of this marketplace and we eventually got large orders from the BBC and VoA.

I was invited to do an AES paper on the design of the 9105. I presented this in Hamburg, Germany, and the paper later made it into the AES Journal:

Increasing Coverage of International Shortwave Broadcast through Improved Audio Processing Techniques
A system approach to audio processing is described, which considers the audio processor, transmitter, antenna, and receiver. The processor uses several cascaded stages of processing to decrease the peak-to-average ratio of the audio signal, improve its intelligibility, provide band-limiting to CCIR, EBU, and FCC standards, and maintain acceptable subjective quality on both speech and music. The long-term average sideband power achievable with program material is approximately 65% of the power produced by a continuous sine wave at 100% modulation.
Author: Orban, Robert
JAES Volume 38 Issue 6 pp. 419-432; June 1990 Permalink

 

[W1DAN]

Bob:

Good info!

In reading the manuals of the 9105A and the 9100a, I see the very close connection with the 9100a, and the careful combination of the steep receiver EQ, midrange clipping and the Hilbert clipping. And yes, I wondered about the 25hz notch...

For entertainment, I once owned a 9105A and connected it to a 100 watt plate modulated ham transmitter called a Viking 2. The density of modulation caused the modulator to fall apart. It could not make 100% modulation as compared with an unprocessed voice signal (or even a 9100a). So I built a PDM solid state transmitter and the 9105A worked it's magic fully.

One day if I can find one again for hobby money I'll grab it. In the meantime I use a 8182A for the Hilbert IM and am feeding a Gregg 2540 for ham use. This combination sounds surprisingly good..a combination between the 9100A and the 9105A.

I appreciate your comments and contributions. Hopefully designers such as you will enjoy telling the design and marketing war stories of the old boxes.

In the meantime I have my ear out for who has an 8600 on the air in the Boston area. It would be interesting to hear your new HF limiting.

Thanks much!

Dan