Suppose your local FM station announces that it's about to play a recording that you have in your collection. Wondering if you'll hear any difference, you put your record on the turntable and play it, more or less it sync with the broadcast.
As you switch back and forth between the two, the chances are good you will find that, in direct comparison with your LP, the radio version is weak at both the bottom and the top of the frequency range, has a cramped feeling because of serious lack of dynamic range, and is a little more noisy, different in timbre and perhaps a little muddy in the middle.
Is the FM broadcaster simply having a bad day? Probably not. FM radio today typically comes off second best when compared to the best home equipment, and the Compact Disc poses an even tougher challenge to broadcasters.
This comes as no news to the broadcasters themselves, and efforts are now spreading in the FM industry to remedy the situation. These efforts will intensify in coming years, as the competition from alternate program sources of very high fidelity gets hotter.
This competition is coming not only from Compact Discs, but also from new hi-fi videocassette recorders with super fidelity, from videodiscs, and even from cable television's recently upgraded stereo programs such as MTV. Too, the movies are giving us an entirely new quality of sound with Dolby Stereo, which is actually a four channel system with excellent characteristics. And home digital audio tape (DAT) cassettes should be here in a year or two.
FM, of course, has a number of important strengths that look especially good in comparison with AM. The top legal limit on frequency response on AM is 15 kHz, but the noise problem makes the practical limit much lower.
The best way to make noise tolerable on weak to moderate-strength AM signals is to limit frequency response, and most readers of this magazine's "Equipment Profiles" will know that typical AM receivers usually cut off around 3 kHz.
In the past, the most-used route to reasonable fidelity in AM listening was to concentrate on high-powered local stations, whose signals were well above the noise level, so that receiver response could be opened up to 12 kHz or so. Readers whose interest in high-quality sound goes back to the 1940s or '50s will remember TRF (tuned radio frequency) receivers, which had wider band signal amplification than the now-universal superheterodyne circuit. With TRF, frequency response could be wide and distortion low. Strong local stations came in with excellent fidelity. The main problem was an occasional 10-kHz whistle, which occurred when the carriers of strong signals only 10 kHz apart beat together; notch filters were a common solution to this.
The situation on FM is entirely different. Mono FM is a transmission system of basically marvelous fidelity. Noise is not a problem as long as the signal reaches the limiting level, which, in a sense, saturates the residual AM response of the tuner. There is neither a technical nor legal limit on frequency response up to 99 kHz or so, if a sub carrier, or SCA signal, is not used.
Even with the extra signals of SCA, the frequency response can go far above the standard 20-kHz top. The maximum legal mid-band distortion is 2.5%, but most FM stations do much better, down to fractions of 1%. Stereo FM is E. bit different. Frequency response is limited to 15 kHz to prevent the signal from "folding over" onto the LR subcarrier. To reach the no-noise level, a stereo signal must be much stronger than a mono signal-23 dB stronger is tie figure usually given, but recent tests by an EIA group studying quadraphonic =M indicated that, with real program material, the difference is more like 26 dB. The broadcasters have accepted this loss, and the resulting reduction in fringe coverage, because f is impossible to be competitive in the FM market without a stereo signal.
However, it's dynamic range that's at the center of FM's present troubles.
With first-rate design, FM transmission can have a signal-to-noise ratio of 60 dB or more, including the whole system of transmitter and receiver. This does not mean an FM operator can put a dynamic range of 60 dB on the air.
For one thing, the Federal Communications Commission (FCC) limits the modulation an FM operator can apply to his carrier, in order to reduce the chances that the signal will interfere with others on the air.
Frequency modulation theoretically produces an infinite series of side bands, new signals that go out with the carrier, at frequencies spaced at multiples of the modulating frequencies. For example, if a 10-kHz signal is applied to an FM transmitter, it will produce sidebands at 10, 20, and 30 kHz (and so on) above and below the carrier frequency. The strength of these side bands depends on how heavily the station modulates its signal.
Strong sidebands too far from the carrier frequency are likely to interfere with other FM signals. But, luckily for broadcasters, if the carrier is not too heavily modulated, the farther side bands will be weak and thus no problem to other stations. The FCC must therefore regulate modulation so that stations' signals will be strong, but not so strong that their sidebands will cause trouble.
In AM, maximum modulation is defined by the physical limits of transmitter action as a maximum of 120% above, and 100% below, nominal carrier strength. (The negative limit is lower because, below 100% negative modulation, the transmitter simply cuts off.) In FM, however, the depth of modulation depends on how far the signal pushes the carrier off its center frequency. Within broad limits, the amount of this frequency deviation is unrelated to the physical limits of the transmitter. A maximum-deviation spec is arbitrary, as far as transmitter capabilities are concerned, and the limit can be raised substantially without causing transmitter distortion from overload.
For this reason, the deviation limit is set by the FCC. To define how much deviation could be allowed without sideband interference, the FCC first defined a characteristic called "occupied bandwidth." This is the spread of frequencies (including all sidebands) that contains 99.5% of the energy in the signal. If this occupied bandwidth is no wider than 240 kHz, interference is likely to be under control. A number of studies have led to a U.S. specification: A maximum deviation of 75 kHz on each side of the carrier frequency will produce an allowable occupied bandwidth.
So far, the story sounds positive. But there is a complication: Nearly all music includes short peaks that are many times as strong as the average level of the music at any given period. If the short peaks are to be kept from going over the 75-kHz deviation limit for stereo, the average strength of the music must be kept very low. In fact, if we define 100% modulation in FM as a 75 kHz deviation, a station turning down the signal level to keep the peaks from going over the limit would typically be modulating at an average level of only 6% to 8%. This would mean a very weak signal, and a station in a metropolitan area could thus lose a large portion of its natural audience. In today's savagely competitive radio industry, losing a large part of the audience is economic suicide. Advertisers study audience surveys with manic intensity and go where the ratings numbers are highest.
How, then, can a commercial FM operation obey the rules and stay in business? The virtually universal answer has been the use of audio processing to bring the average and peak levels of the music closer together-in effect, to flatten the peaks so they have less tendency to go over the top.
The processing must also do something about pre-emphasis, the large boost of high frequencies at the transmitter to improve the signal-to-noise ratio. Pre-emphasis turns high-frequency peaks into powerful bursts that need special taming if they are not to push modulation over the 75-kHz deviation limit.
The manufacture of audio processors for radio has become a boom industry in the last 5 years, and the units have also moved far ahead technically. The typical processor today has a three-stage action: First, a slow compression circuit somewhat flattens the slower rises in music level; then, moderately fast peaks are compressed, and finally, an extremely fast circuit, usually called a "clipper," catches the very quick peaks that slip through the earlier circuits.
Applying such a device results in a tamed-down audio signal, shorn of troublesome peaks, which the operator can turn up for high overall modulation without having peaks go over the 75 kHz deviation limit. Unfortunately, another obvious result is a reduction in the dynamic range of the music, the amount of reduction depending on how the operator sets his audio processor's controls.
The operator can set those controls for just the amount of enhancement he wants. At one extreme are classical music stations in large markets. They do less processing than the pop-music stations because classical music is generally understood, at least by its listeners, to have dynamic range as an essential attribute. This is probably not true of popular music or its listeners.
Classical station WNCN in New York, for example, uses light processing, producing an average modulation level of around 30%. The management does get complaints of weak or noisy signals from some listeners, but so far the number of listeners who do get a usable signal and who appreciate the wide dynamic range has been large enough to keep the station going.
At the other extreme are stations slugging it out on the rock front, where a competitive signal is one that jumps out at the listener tuning past. The signal of a prominent rock station in New York has been measured to have an average modulation level of 70%, and a dynamic range of 2.5 dB! The music is very loud, and it maintains nearly the same loudness all the time.
Noncommercial stations have it a little better. They are not forced to compete for big, marketable audiences, although they still need to serve substantial audiences to satisfy their backers.
One example is WETA in Washington, D.C., which is said to use very little processing. The management has surveyed listeners to find whether the majority wants the increased dynamic range of a lightly processed signal or a less noisy signal (in suburban areas) with much less dynamic range. The large majority has preferred the maximum dynamic range.
Reducing the amount of processing is a good idea not only because it increases dynamic range but also because it means less coloration. Heavy processing uses program-dependent gain circuits, whose gain varies with signal level, and this adds false coloration to the music.
What can be done to lower the pressure for processing? Because "loudness wars" are an acknowledged part of operating an FM station in many cities today, help must come on the technical front, if it is to come at all.
One thing that some FM operators could do is take advantage of a loophole in the rules. The FCC doesn't say a station may never deviate beyond 75 kHz, only that its peak deviation must not exceed 75 kHz more than 10 times a minute. Still, many stations have not allowed themselves even this much overmodulation, in part because of the nonexistence of equipment to automatically restrict such peaks to the allotted 10 per minute. At least one firm, Modulation Sciences, of Brooklyn, N.Y. (of which author E.S. is technical director), plans to produce such equipment. The 10-per-minute rule makes sense because the energy in short peaks, more than a few milliseconds apart, will not "integrate" for a cumulative interference effect, and each separate peak will lack the energy to cause interference trouble.
In the long run, the 75-kHz rule might be modified. Recent studies show that, while peak deviation is directly related to occupied bandwidth in mono, this is not the case in stereo-which is to say, in virtually all FM broadcasting today.
Mathematical analyses, some made by Dr. Eric Stoll of Modulation Sciences, have shown that a stereo signal with 75 kHz of deviation does not occupy its full, 240-kHz bandwidth. For example, with a 15-kHz signal modulating a carrier to 75 kHz deviation, one study showed occupied bandwidth to be only 166 kHz.
Measurements of airborne FM signals carried out by the Environmental Protection Agency in Washington, D.C.
[1] showed similar relationships. The agency had assembled very elaborate automated equipment for measuring the strength of radio-frequency fields, to determine whether the prevailing fields posed any danger to the health of those exposed to them. As a byproduct, the group tested the signals of Washington FM stations to determine occupied bandwidths with typical program material.
===Mono FM is essentially a higher fidelity medium than stereo FM, with its 15-kHz frequency limit, lower dynamic range and weaker signal. ===
=======
Table 1--Summary of data on FM broadcast-station bandwidth (excerpted from
Reference 1).
Station | Freq. (Mhz) | Averaged (25db), Bandwidth khz
WETA-FM1 90.9 47
WGTS-FM 91.9 65
WJMD-FM 94.7 107
WPGC-FM 95.5 160
WASH-FM 97.1 124
WGAY-FM 99.5 80
WFAN-FM2 100.3 102
WWDC-FM1 101.1 126
WHFS-FM 102.3 120
WGMS-FM 103.5 94
WAVA-FM3 105.1 60
WMAL-FM1 107.3 87
Notes:
1. Stereo + SCA
2. Mono
3. Mono + SCA
All other stations stereo
=======
Table 1 shows some of the findings.
The figures, which are average bandwidths over a period of at least 10 minutes, show how far below the 240kHz allowance stereo signals typically are. (The differences reflect a number of factors in station operation, including-but not restricted to-the degree of audio processing.) It is notable that nonprofit station WETA, which uses very little processing, was extremely low on occupied bandwidth.
What these studies and tests demonstrate is that the allowable peak deviation could be increased substantially in stereo without causing unacceptable interference on the band, if the FCC would change the rules to allow this. Stations that use the least processing would benefit the most. FM operators who wanted to keep fidelity higher could reduce processing without cutting down average modulation.
Stations that still used heavy processing would not gain much from such a change. This question of the peak-deviation limit has been under discussion at the FCC, but there has been no action yet; we hope it will come soon.
Whether that change comes or not, there are a number of ways in which FM operators can raise fidelity within the present FCC rules. Replacing old transmitters with current, top-grade units would be especially effective.
Older audio processors, prime causes of below-par fidelity, could be replaced with more recent versions, a number of which-if used lightly-will greatly reduce processing distortion.
Experience with FM stations suggests that processing distortion is lowest when the sharp clipper does most of the job and broad gain-control action is eased off as much as possible. But this won't be done unless the operator wants light processing.
Another effective way to raise fidelity is to bring the studio gear-microphones, disc and tape players, and mixing consoles-up to today's best standards. All studio gear has benefited from the last decade's great advances in recording-studio quality. Experience in many FM stations shows that turntables and broadcast tape cartridge (endless-loop) machines are commonly to blame for low audio quality. The turntables and cart machines used by broadcasters a decade ago had very low quality to start with, often made worse by poor maintenance.
Better turntables and cartridge decks have appeared in the last 5 years.
Digital technology is beginning to have an effect on FM quality. A number of FM stations are, of course, occasionally using Compact Discs for programming. As suggested at the beginning of this article, the full quality of the Compact Disc--particularly its dynamic range--cannot get through the FM transmission system. But the CD's recording quality often does give the FM signal a lift that many listeners notice and welcome.
However, the digital disc puts even more pressure on FM radio stations to control the processing function. The CD's frequency response is really flat to the top of the range, so bursts of highs, when enlarged by pre-emphasis, are harder to handle than ever. The processor often has to be carefully reset to avoid serious distortion, and some of the older processors won't make it at all.
A number of FM stations are using digital technique to excellent effect in their handling of "remote" pickups music programs brought in from outside the studio. Nonprofit station WGBH in Boston, which has been particularly venturesome over a number of years in its push for higher signal quality, is among those that have used a digital processor and a VCR to record concerts for later broadcast. WGBH is also using advanced equipment on "live" feeds. A digital audio-transmission system between Symphony Hall in Boston and the station's studios has been especially successful, with listeners expressing strong, unsolicited approval of the results. The station's link from studio to transmitter will also soon be digital, using a video transmission system for the extra bandwidth needed. The existing analog link uses Dolby A noise reduction to improve the signal-to-noise ratio. . On studio broadcasts, WGBH uses very little audio processing. On live music broadcasts, they cut their processing back by another 6 dB, responding to the listener's expectations of especially high quality on such broadcasts, and conscious of the fact that these live events have a large following among listeners, close enough to the station to get a strong signal.
A number of other FM stations are making determined efforts to raise signal quality. That is one reason we feel confident, as we said at the beginning, that FM will continue to function as a source of satisfaction for the audiophile and serious music listener. It will also continue to satisfy the average listener, who now approaches FM more as background than as the concentrated listening experience that serious phonograph music so often is.
FM's somewhat reduced dynamic range is actually an advantage in a music source that listeners alternately approach and depart.
In any case, we conclude that FM can advance to a level close to the best home-play quality, close enough to make the signal thoroughly satisfying. The growing intensity of media competition in the coming decade makes it certain that many FM operators will make the effort needed to reach that high level.
Reference
1. Tell, Richard A. and John C. Nelson, "Broadcast-Signal Bandwidth Measurements Using Real-Time Data Aver aging," IEEE Transactions on Broad casting, Vol. BC-22, No. 4, December 1976.
by ERIC SMALL and ROBIN LANIER
[Eric Small, currently Vice President of Engineering for Modulation Sciences in Brooklyn, N.Y., has extensive experience with FM stations and in the development of broadcasting equipment. Robin Lanier, a freelance writer, was formerly Senior Editor of BME (Broadcast Management Engineering) magazine.]
(adapted from Audio magazine, March 1985)
Also see:
FMX: Long Distance Stereo, Clean as Mono (May 1986)
Antennas FM -- Try a Rhombic FM Antenna (Jan. 1982)
= = = =