Aphex Model 2020 -vs- Orban Model 8200 "Optimod"

By Donn Werrbach, VP of Engineering, Aphex Systems Ltd

Introduction
This paper discusses the merits and advantages of the new Aphex model 2020 FM Pro in relation to the Orban model 8200 Optimod.

An Historical Account
Several years ago, Orban introduced their Model 8200 digital processor to a broadcast industry embarking on a feverish "go all digital" agenda. The fever was fueled by a general ignorance in the industry about the limitations of digital audio signal processing. An incredible amount of misinformation was spread throughout the industry by well meaning but misinformed digital audio proponents. Now, after years of battleground digital experience in the broadcasting industry, the true advantages and disadvantages of the digital media are better known. It is recognized that the so-called "all digital" audio chain considerably degrades the sound quality. Consequently, broadcasters are starting to make better use of both digital and analog technology to achieve more ideal results.

The naïve belief that digital processing is inherently better than analog is finally falling apart. It has become widely acknowledged, within the broadcast community, that the nonlinear digital signal processing artifacts of the Model 8200 result in audible and undesirable "digital grunge". Nevertheless, Orban’s Model 8200 supplied, without meaningful competition for many years, a strong market demand for a true digital audio processor. As a result there are a great many radio stations running with Model 8200’s on the air. Amid the present day controversy over digital audio processing, Orban claims that owners actually love the synthetic sound quality of the Model 8200. The reality is that dissatisfaction is growing rapidly. An alternative that offers better sound quality is sought by the market place. Whether that alternative is of digital or analog technology is of less importance than quality of sound. The fact is that advanced analog signal processing still yields the best sound according to most listeners.

FM Broadcasters are ready for the next generation audio processor. One without digital grunge but with all the features and conveniences afforded by digital control: an analog audio processor that readily connects with digital audio systems while maintaining high resolution and low distortion. They are ready for the Aphex Model 2020 FM Pro.

The Aphex Model 2020 is a new benchmark for FM audio processing. It is louder, better sounding, more flexible, and easier to use. The following comparison shows the many advantages of the Model 2020 when used in digital, analog, and hybrid audio systems.

The Comparison

Loudness
Subjectively, the 2020 can be tuned louder than 8200 with more appealing results. The 8200 quickly becomes distorted and irritating when it is driven into competitive loudness settings. The 2020 gathers loudness with greater clarity and punch, avoiding the artifacts of the 8200.

Some of the 8200’s degrading factors that worsen with high loudness are: raspy multiband clipping, extreme equalization contouring, pumping, depressing of the vocals, and frequency band dissection making the spectrum seem "overhyped".

Noise
When set for dynamic processing headroom comparable to that available on the 2020, the analog-to-digital conversion noise of the 8200 is so high that it is plainly visible over the pilot tone when displayed on an oscilloscope. Programs processed by the 8200 also contain audible and annoying noise modulation. Voices carry perceptible quantizing noise that comes and goes with articulation, an artifact sometimes referred to as voice "hair". During low program levels, when the leveling gain is increased and compression is also releasing, a large amount of gritty background noise is audible. In many cases, a breathing background noise is audible behind music selections. Under similar circumstances the 2020 sounds quiet and clean.

Fatiguing Distortion
One of the common complaints by users of the 8200 is the grunge, brassiness and fatiguing quality of its sound- even at mild settings. Many hours of investigation suggests this is caused by dynamic control loop instability and inadequate sampling rate of the various digital processing algorithms of the 8200.

A host of revealing evidence of the 8200’s defects can be seen just by sweeping a sinewave through it and observing the output on an oscilloscope. You will immediately observe the multiband clipping at most frequencies. At higher frequencies, i.e., above 2KHz, you will observe severe beat frequencies that are digital artifacts of poor DSP technique. At certain frequencies and amplitudes you will observe the audio envelope to be ripped and saw-toothed, indicative of unstable limiting or compression loops, again a particular DSP limitation. These anomalies are indeed audible as you can prove just by listening. It is ludicrous to claim they are inaudibly masked in program material or that tone tests are irrelevant.

The 2020 uses clean and stable processing methods that generate no beat frequencies or other anomalies as described above.

Loudness & Distortion
The 8200 gathers loudness by the liberal use of multiband clipping. This is part of the 8200’s processing strategy and is a legacy of the prior XT analog processor. Multiband clipping causes a thick, buried kind of distortion which makes the sound "loud" but not "big". Rather than creating a large enveloping loudness, this technique creates a crunched, unflattering, one dimensional effect diminishing the apparent size of the stereo image. The main advantage (in one way of thinking) of this kind of multiband clipping is to transform widely varying program sources into a "lowest common denominator" of average sound quality. Taken to competitive limits, this method creates unlistenable sound. In contrast, the Model 2020 uses no multiband clipping but depends upon an intelligent interaction between a post-crossover multiband compressor (patented) and a separate, independent, phase coherent distortion cancelled multiband limiter (patented). With the 2020 you can achieve anything from a big enveloping loudness to a flat, thick wall of sound by proper balancing of the processing. At comparable extreme loudness settings, the Model 2020 sounds bigger, cleaner, and better.

Vocal Attenuation
Another complaint about the 8200 is the feeling that vocals are pushed back into the mix, making lyrics hard to hear. The problem gets worse when you need voice focus most: in dense mixes. This is the result of the 8200’s particular processing strategy. The problem is endemic and cannot be corrected by any of the 8200’s user adjustments. By comparison, the Aphex patented "Automatic Limiting Threshold" (patented) and "Wave Dependent Compressor" (patented) algorithms help the 2020 to balance vocals more naturally, allowing even the densest mixes to remain fully intelligible. You never get the impression something is being "pushed back" out of the mix.

Fullness Of Bass
Another complaint is the lack of any real low frequencies coming through the 8200. Bass sounds like it is restricted and forced up into a narrow band of energy. Music containing deep bass elements loses feeling. The 2020 handles bass energy in a totally different manner specifically allowing bass to fully develop dynamically at all bass frequencies. A specialized bass processor in the 2020 allows you to enhance the bass very musically, not simply through boosted equalization as with the 8200. The 2020 uses a patent pending dynamic process that fills the modulation envelope with bass energy to the greatest possible extent consistent with instantaneous overload distortion constraints.

Bass Continuity
The transition of notes from the high to low bass region is not seamless with the 8200 taking on a "dissected" characteristic. Very low bass notes seem to disappear while high bass notes take on a different tonal character. In contrast, the 2020’s bass is very fluid and flows seamlessly throughout the bass frequency range, holding a more natural or even enhanced balance of loudness as notes progress lower in the range, compensating for the dynamics restrictions of FM transmission. The model 2020 produces the best sounding bass of any audio processor yet developed.

Analog Advantage
The model 8200 suffers from typical DSP errors, i.e., truncation and rounding errors, and further multiplication of those errors as the processing continues. These errors are manifested as very characteristic digital audio distortions and are especially acute in the 8200 because of it’s fixed-point DSP architecture, minimal digital audio sample rate, and economized DSP engine. In comparison, the model 2020 processes in the analog domain using premium circuitry and is thus inherently more perfect at retaining the audio quality.

Digital I/O & Sample Rate Problems
Information revealed by Orban indicates the Model 8200 DSP sample rate runs at merely 32KHz and is spot upsampled to only 128KHz at a few critical points. This is not sufficient sampling to avoid frequency aliasing that is caused by nonlinear digital signal processing algorithms such as compression and limiting. It is also too slow a sample rate for some of the agc control loops to remain stable. You can see the instability while observing sinewaves in limiting. You'll observe a sawtooth shaped amplitude irregularity that translates as a raspy effect on the sound (more grunge). Using perfectly stable analog algorithms, the Model 2020 avoids all these grunge mechanisms. Digital I/O of the Model 2020 is handled by accepting any sample rate at the digital input, converting it to 48KHz, then converting the digital to analog for processing. The digital audio input is fully reconstructed by the comprehensive analog processing chain, maximally de-quantizing the audio signal. The Model 2020 then converts the finished analog audio back to 20-bit digital at any selected output sample rate. This makes the model 2020 the best quality sample rate converter available as well as a better digital audio processor from the viewpoint of optimum and clean signal processing.

Digital FM Exciters
Digital I/O options of the 2020 not available with the 8200 create greater flexibility and facilitate much higher audio quality in a digital audio system including interfacing with digital FM exciters. For example, the Harris "Digit" digital FM exciter can accept the full 20-bit AES/EBU bitstream at 48KS/s rate from the model 2020. On-air comparisons have shown there is a very discernible improvement in the sound of the Digit at each incremental change when switching between 32, 44.1, and 48K sample rates.

A complaint about the 8200 is the large d.c. drift present at the composite and discrete outputs. The drift is the result of the 8200’s DSP processing error and can cause alarming carrier frequency offsets. The only workaround currently available is to turn on the 8200’s 30Hz highpass input filter. This further degrades the already lacking bass feel of the 8200.

The 2020 suffers no d.c. drift problems. To absolutely guarantee d.c. stability in the digital domain, we invented a patent pending method specifically to cancel all d.c. a/d converter digital output drift. The 2020 delivers deep and full bass, with rock solid digital and analog domain d.c. stability.

Flexibility
The philosophy behind the 2020 is to allow users a great latitude in the setting up of processing to accommodate a wide range of individual requirements and preferences. The 8200 much more severely restricts the range of adjustments you may try and omits many key elements and parameters from user control altogether. For example, the 8200’s release time can only be set to one of four speeds (not continuously adjustable), and all bands have to be set equally. The 2020 allows you to set the bands individually and with continuous range so you can find the "sweet spots".

Metering
The 2020 front panel metering shows the activity of all important processing functions continuously and in real time. Metering on the 8200 is by way of an LCD panel which only approximates the activity of the audio processor and requires you to switch between screens to see all the meters. With the 2020 you get a picture of the audio processing at a single glance.

Stereo Insert
The 2020 allows you to insert additional audio processing at the ideal location: between the leveler and multiband compressor. The 8200 has no capability for external processing to be inserted.

Other 2020 Advantages

Auto Log-Out Feature
Presets Auto Save Feature
Input Auto Source Feature
Hard-wire Bypass Feature
Universal Power Input
Optimal Panel Size of 2RU

end

Trademark notification:
"FM Pro" is a trademark of Aphex Systems Ltd.
"Orban" and "Optimod" are trademarks of Harman International Industries, Inc.