A recent whitepaper entitled, A SYSTEMS APPROACH TO IMPROVING
SUBCARRIER PERFORMANCE by Geoffrey N. Mendenhall, P.E., Vice
President of Engineering, Broadcast Electronics Inc. in Quincy,
Illinois, served to confirm and further validate some concepts I
discovered through extensive testing and theory.
The focal point of the article was audio linearity in the modulator, followed through to the PA stage in terms of RF bandpass. In addition, attention was focused on the stereo generator's performance in terms of cleanness of its subcarrier output. The intent was to help broadcasters improve the separation between their main FM channel program and the SCA audio, to reduce or eliminate crosstalk.
For years, I focused on designing broadcast-quality stereo generators and implicitly, it seemed obvious to me that extremely low distortion in the modulator was paramount to good nulling of unmodulated channels at the receiver. Because it's a matrix balance situation, high precision and absolute phase linearity are mandatory. But many things in the chain will degrade this signal. Requirements have to be met on both the modulator as well as the stereo generator.
Finally, the paper pointed out the disadvantages of composite clippers used by many commercial broadcasters (something I always thought was utterly asinine in the first place), as they introduce momentary wideband artifacts which occupy more composite spectrum than a clean, properly limited audio + subcarrier signal.
Perfect phase response throughout all circuits from matrix to
composite amplifier/mixer.
Extremely low distortion in all composite/mixer amps.
A DC-driven balanced modulator (in frequency division multiplexers) which is self-nulling of suppressed carrier leakage (my design).
Extremely clean subcarrier source with all parasitics down at least 80-90db or better (my specs).
A spectrally-pure 19Khz pilot signal. Any harmonics here will result in audible hetrodynes in the receiver, with test signals in the 7-15KHz range. (note that many receivers introduce their own nonlinearity, so this may be difficult to measure without something along the lines of a Sequerra Model 1 Broadcast Analyzer on hand.)
Audio lowpass filters that are properly designed for group delay, or employ overshoot damper circuits to control overmodulation peaks from complex program material (my design).
An extremely linear modulator that is varactor-controlled (not
FET modulated) and compensated for symmetry (my design).
All RF stages must offer a sufficient bandpass to maintain the balance of the most important sideband information generated by an FM stereo transmission. This means it should be fairly flat over a 1.2 Mhz range (3db down points). (Solid state broadband amplifiers are immune to the issues which are rather paramount with tube finals using resonant tank circuits), but it's important to bear this in mind anyway (my design).
A low SWR and a "quiet" transmission line which doesn't radiate into the audio equipment at the transmitter (my idea).
When all these requirements are met, the broadcast signal will offer the greatest channel separation, extremely low distortion and a flat frequency response.
If you're using too narrow a bandpass filter, you could suffer from poor channel separation of your stereo signal because the sidebands are being filtered out by your RF bandwidth limitations. If you use a filter network, make sure it allows a flat bandpass over a 1.2 Mhz range, centered on your channel allocation. In extreme cases, I have seen an all-tube FM rig suffer from having only 10db of channel separation because the hi-Q tuned RF stages were TOO sharp!
And here is one you'd never suspect: the antenna. My own experiences have confirmed that antenna systems with extremely narrow band characteristics (being the case with many end-fed designs I've worked with) present significant imbalances of carrier sideband amplitudes, resulting in some non-linearity in modulation. This artifact is easily observed when modulating one channel and monitoring the unmodulated channel of a stereo transmitter connected to the antenna in question. I found that the best results are obtained with antenna systems which have at least a 200Khz bandwidth before VSWR rises more than 1.5:1 and impedance changes no more than 30-50 ohms above nominal. Some designs jump right up to 300-500 ohms with only 100Khz of deviation in carrier frequency; such antennas are the cause of many difficult-to-pinpoint modulation problems.
If there are distortion artifacts in your 38Khz subcarrier, the result could be crosstalk between audio channels when one channel is unmodulated. In addition, high treble percussion will be distorted, sounding like "aliasing".
The same could be true if your subcarrier modulator is non-linear.
A poorly designed limiter will present problems with "dynamic channel separation", this is when overshoot or the passing of brief transients such as to exceed +/- 75Khz deviation of the main carrier occurs. The result is "punch through" in the unmodulated channel when only one channel is modulating. The audio must be pre-processed to insure that maximum levels are not exceeded, while maintaining a musical quality signal. This is a complex matter, which I'll probably discuss in another paper soon.
A poorly designed low pass audio filter could cause many problems when audio frequencies are able to approach the pilot and subcarrier lower sideband spectrum. This results in aliasing distortion and often is visible at the receiver as a flickering pilot light/stereo indicator.
While these areas of adjustment are not necessarily required for a decent, listenable signal, they are necessary for the perfectionist to put out a signal that is second to none in quality. Some of these adjustments have effects that aren't audible (unless you're running SCA services), but they do add up to a technically-superior signal. If you're building a rig from scratch, then these are highly desirable characteristics to design for.
Presented to you by
The "Peg-legged" Bovine One
Please make every effort to be a responsible broadcaster. The more knowledge you have, the better your capabilities to reach this standard and very important goal.
Authored by your friendly "Peg-legged" Bovine One