F.M. Coverage Improvements with the Aphex Model 2020

By Donn Werrbach, VP of Engineering, Aphex Systems Ltd

Preface
The Model 2020 FM Pro inherited the highly acclaimed Aphex patented PPDM stereo generator originally developed in 1992 to be used in our Model 400 Digicoder, part of our Audiophile Air Chain product line. As happens frequently with users of the Audiophile Air Chain, many broadcasters are reporting significantly improved FM coverage after replacing their competing audio processor with a Model 2020. We are often asked to explain how this could be possible. It is certainly not obvious how an audio processor can affect the apparent multipath degradation suffered by radio reception.

A Historical Background
In 1993 Aphex introduced the Model 400 Digicoder, a precision stereo generator, to form the final link of the Aphex Audiophile Air Chain. Prior to the Digicoder, broadcasters using an Aphex air chain needed to adapt a third party's stereo generator that we felt was a limitation to the system. We did not want to produce a common-technology stereo generator just to have one of our own, so we embarked upon a groundbreaking program to develop new technology. This program lead to the invention of Parallel Path Digital Modulation (PPDM). Attesting to its originality, the PPDM technique has won US patent number 5,155,769 issued on October 13, 1992.

Unexpected Coverage Increases Reported
Based on lab measurements and listening tests we knew we had a very high performance stereo generator, but its actual performance in the field exceeded our expectations. From the very first unit shipped, we got reports about reduced multipath interference. We expected favorable comments about the sound quality, but we never anticipated the bonus of improved coverage area. We were skeptical about the first reports of coverage improvements but it became hard to discount them as the number of reports continued climbing. Station owners were getting calls from happy listeners because their signal became clear not only in the fringe areas but even in strong signal areas where there had been bothersome multipath distortion. From points as diverse as the Philippines, Australia, Switzerland, the United States, we kept getting the same feedback: they were mysteriously getting better coverage!

An Explanation Offered
The mystery may not be unsolvable. We know that F.M. reception problems can be traced to problems in all three elements of the broadcast medium, namely, the transmission apparatus, radio propagation effects, and the receiving apparatus. A close look at these elements does suggest an explanation of how the Digicoder and Model 2020 improve the F.M. signal coverage of some radio stations.

To begin, let's just look at the receiving apparatus and assume the transmission and propagation are perfect. There are plenty of reception difficulties traceable to the receiver and receiving antenna. Certainly in the fringe area, but even within the normal F.M. coverage areas, the signal is too weak to drive an F.M. radio into full quieting. Quieting occurs when the radio’s intermediate frequency (i.f.) amplifier is heavily saturated (or "in limiting") thereby stripping off the A.M. noise which may ride on the F.M. carrier. Full quieting is the desirable condition for good F.M. reception. If the receiver is not quieted, numerous things can go wrong with its F.M. detector. In receivers using "Foster-Sealy" or "Ratio" detectors, any A.M. noise making it through the i.f. stages will be present in the recovered audio. The more sophisticated receivers using "Pulse-counting" or "PLL" detectors can generate very loud noise when the lock threshold is modulated modulation by A.M. noise.

A.M. noise can be picked up from radiating terrestrial noise sources such as electric appliances, motors, auto ignition systems, neon signs, TV sets, computers, etc. Such noises are usually reproduced as intermittent "buzz" or "hash" but are independent from and not responsive to the F.M. audio signals. Terrestrial noise in the VHF F.M. band is relatively weak and that is a major reason the VHF band was assigned to the standard F.M. broadcasting services.

There are two other causes of A.M. noise far more damaging to F.M. reception than terrestrial A.M. interference. These are incidental A.M. and multipath reception. Incidental A.M. is a direct product of the F.M. carrier when the carrier passes through a nonlinear signal path. This is frequently the result of a mismatched receiving antenna. Multipath reception mixes two time displaced copies of the same F.M. signal together. This causes the vector sum of the carriers to produce an A.M. component in a very distorted rendition of the F.M. modulation. Incidental A.M. from path nonlinearity and multipath reception produces more distortion when a stereo subcarrier is present than for a purely mono F.M. signal. This can be explained by a couple of facts. First, the modulated ultrasonic subcarrier produces incidental A.M. components in the audio frequency range. Second, the presence of the subcarrier tends to increase the average occupied width of the F.M. sidebands increasing multipath vector interferences.

A nominal amount of incidental A.M. can be suppressed by the F.M. receiver if the incoming signal is not too weak and reasonable quieting is achieved. This is why the simple whip antenna of a portable radio can give adequate reception under good signal conditions even though it is almost certainly quite nonlinear to the F.M. carrier. Multipath distortion is another case. Multipath distortion causes such deeply modulated A.M. noise that it gets through a fully saturated i.f. amplifier to produce audio distortion even in the strongest reception areas.

In fringe area conditions, incidental A.M. will not be rejected by the F.M. receiver and the resulting audio distortion sounds typically just like multipath distortion. In these conditions we attempt to adjust the receiving antenna for best reception. Most people think they are orienting the antenna to get the strongest signal. What they ultimately are achieving, however, is usually a reduction of antenna caused incidental a.m., or the rejection of a multipath interference vector, or both.

F.M. reception problems are also magnified by stereo broadcasting because of the receiver's increased suceptance to noise in the stereo mode. A stereo F.M. receiver has a noise admitting bandwidth of 53 kilohertz versus only about 15 kilohertz for mono. Through the stereo decoder circuit, all the A.M. noise above 19 Khz demodulates back into the 15KHz audio spectrum. This amounts to about an 11dB increase of noise intensity. Second, a left or right signal modulates the main and sub carriers only 45 percent each. This means that the recovered main and subcarrier signals are noisier than a mono signal by 6.9dB. When these are dematrixed, the net resulting noise is the RMS sum of the left and right channel noise that equals 1.4 times 6.9 or 9.8dB. The total net stereo reception noise is therefore 9.8dB + 11dB = 20.8dB worse than mono reception. If the receiver is fully quieted, the stereo noise will be sufficiently low in a good receiver. However, even the slightest reception problems that produce noise or distortion will be exaggerated by almost 21dB for a stereo receiver over what's heard from a mono receiver. That is why most stereo tuners have mono switch to let you tame that nasty F.M. station.

Now, looking away from the receiving apparatus, let's address other factors causing bad reception of F.M. signals. We have generally assumed that bad reception is a function of propagation through the Ether and is therefore out of the control of the broadcaster. This is not necessarily true. A broadcast station can be transmitting an F.M. carrier with a significant amount of incidental AM. This kind of transmission will produce results exactly like multipath distortion in the fringe areas. Reducing or eliminating incidental A.M. from the transmitted signal can give a remarkable improvement to effective coverage, even to the extent of eliminating difficulties in areas that were previously believed to be plagued by multipath reception.

It is pretty widely known that incidental A.M. is caused by unoptimized transmitting antennas and transmitters. What hasn't been widely realized is that the audio processing system can cause troubles with reception! There are several ways this can occur. One way is through RF feedback into the audio processor itself. If the processor is not totally RFI immune, it can experience added noise and distortion if operated anywhere near a transmitter or exciter. The result can be both in-band and out-of-band spurious F.M. products. This can make stereo reception be especially dirty where incidental A.M. is a problem. The Aphex Digicoder is RFI immunized to an extent that totally eliminates this possibility.

Another way audio processors cause bad reception is by generating an imperfect composite signal for the transmitter to broadcast. Any stereo subcarrier noise or grunge will be recovered audibly by the receiver. Equally devastating, spurious signals that may be present at the outer margin of the multiplex spectrum can exacerbate any tendency of the transmitter or antenna to generate incidental A.M. Pilot or subcarrier phase modulation can cause stereo receivers to generate gross distortion. If the pilot receives sufficient A.M. this too will cause many receivers to generate what sounds like multipath distortion even when the reception is strong.

We generally expect any professional broadcast audio processor to be free from these kinds of deficiencies, but we may be mistaken. The legal standards for stereo transmission equipment were put in place decades ago. Even though technology has improved considerably since then, much of the equipment made today is still designed with the decades old standards as benchmarks of performance. Unfortunately, many stereo generators and audio processors in use today fall far short of perfection in terms of the stereo multiplex output.

The amount of incidental A.M. that is generated by transmission or reception depends to a great extent on how much modulation is in the stereo subcarrier at any given instant. This means that an "over-stereoed" signal will suffer worse multipath distortion than a narrower stereo signal. Processing distortion is also responsible for a lot of fringe reception problems. The more distortion that is present in the audio, the more audio trash gets buried in the subcarrier modulation. This ultimately causes nastier multipath distortion. Processing distortion is endemic in most of today's F.M. audio processors. The loudness wars have brought us a generation of audio processors trading ever increasing amounts of distortion for an incremental loudness gain. Many broadcasters suffering from loudness paranoia have traded all their audio quality for a speck more loudness but actually reduced their effective service area.

But what about digital? The introduction of digital processing has only made the situation worse since the grunge residuals due to the digital domain are especially disagreeable to the ear. Digital distortion products inevitably end up as increased multipath distortion. Setting up a digital processor for competitive loudness results a gritty sound and worsened fringe reception. In today's technology, and into the foreseeable future, analog signal processing still holds the edge over digital when it comes to producing a high quality signal.

Besides audio processing distortion, the stereo multiplex encoder can contribute significantly to fringe reception quality. A stereo generator can introduce signal problems that don't show up under laboratory conditions. For example, baseband distortion of the signal path can create distortion products in the subcarrier region. Also, phase locked pilot oscillators are prone to "wobble" or be "pulled around" by pc board influences such as audio signal ground currents. The pilot protection filters of a stereo generator may inadequately protect the pilot frequency region causing interference to pilot detection in F.M. receivers. Distortion products created by the stereo generator itself can even cause similar problems. Clean stereo reception depends on a solid pilot lock in the receiver, so any pilot instability at all makes the receiver’s multiplex decoder produce distortion or noise.

With that background information laid out, here's how I think the Model 2020 FM Pro with its PPDM stereo generator and clean analog processing is actually improving the coverage area of so many radio stations. It certainly seems clear that generating a clean and stable stereo transmission is the key to maximizing the F.M. coverage, and I submit that the transmission quality of many radio stations is being compromised by their various audio processors and stereo generators. It is not that the Model 2020 does anything magic to improve coverage, it is that most other processors and stereo generators generate a dirty enough output to degrade the quality of transmission. Let me outline why the Model 2020 can generate such a clean output compared to other processors.

First, the PPDM stereo generator creates a virtually perfect multiplex signal with absolutely stable pilot and subcarrier frequencies that remain locked together with digital "numeric precision". There are no phase locked loops to wobble. The pilot and subcarrier frequencies are accurately synthesized directly from time locked digital sine code tables that always retain an exact phase and frequency relationship. The patented PPDM technique couples precision digital modulation with the purely analog stereo signals to produce a phenomenally clean and stable multiplex output far superior to that of any other stereo generator in existence.

Second, the Model 2020 is carefully filtered to highly reject any RFI that could cause processing instabilities or distortion that would make their way into the stereo subcarrier frequency band. This also assures the lowest possible noise floor.

Third, the pilot protection (15KHz Lowpass) filters of the Model 2020 are carefully tailored to provide extremely deep rejection of unwanted signals that would interfere with the pilot recovery of receivers, even when the overshoot compensation is activating. Active overshoot compensation does not introduce a significant level of spurious energy within the stereo subcarrier frequency band. This permits high and stable average modulation further improving the signal to noise ratio of reception.

Fourth, the audio circuits in the Model 2020 are truly of audiophile quality comprising only high-grade components. From stem to stern, all processing algorithms are designed not only to gather the desired loudness and sonic performance, but also for low residual grunge. Analog signal processing is naturally cleaner than digital audio processing, and with special attention to excellent circuit design we have raised the art significantly as evidenced by the eleven United States Patents awarded for processing circuits used in the Model 1100.

Summary
If you expect the best possible F.M. coverage, then you need to start with the cleanest possible stereo multiplex signal from your air chain. By the ever mounting evidence it is apparent that the Aphex Model 2020 FM Pro is significantly better at generating a transmission-friendly output than the competing audio processors.