Much has been written about noise problems in FM transmitters. Virtually everyone at one time has experienced the problem. All too often, so-called "experts" talk about remedies like "sheilding" and "filtering", but these words are often taken out of context and have little meaning without a full understanding of what is meant in this situation.
When you power up your transmitter and hear "hum" in your radio receiver tuned to that signal, there are several possible causes. Let's break them down here into categories:
Since most power supplies use "full-wave" rectifiers which produce switching artifacts at 2X the line frequency, the most common hum heard is 120 hertz. While there are also components of 60 hertz mixed in, the dominant frequency is always 120 hertz, with full-wave rectifier-powered equipment.
Quite often, a combination of all four causes is present in the average microbroadcasting station, if it is not built with express consideration for such factors. One by one, they can be eliminated, if you understand and follow the procedures in this document carefully.
Inexpensive transmitters are usually plagued with a poorly-filtered DC power supply. Furthermore, there is poor isolation between the load ground and the the power transformer secondary centertap ground return point. This point in particular can prove exasperating the the average microbroadcaster, because no matter how much filtering one installs, the ripple spikes remain, if the load ground is electrically too close to the transformer return ground. High current spikes from the switching of the diodes into forward conduction produce a few millivolts of noise in the ground circuit of the power supply, and this can be heard as a 120 hertz hum too.
Another leading cause of hum, and more severely so, is by RF energy getting modulated by the power line. This is caused by a free path for RF energy to get in and out of the AC line side of the transmitter's DC power supply.
Once the transmitter has been cleaned up of RF leakage, the "air chain" becomes the next source of noise. Operating in a high-RF environment, most audio equipment becomes susceptible to a form of hum, induced by RF getting into the power supply, either through the line cord, or through the audio inputs and outputs, or both.
And finally, ground loops are another less annoying, but still bothersome, cause of hum, of the 60 hertz variety. Equipment that is plugged into different outlets than other equipment feeding it, or being fed by it, can experience slight potential differences in the chassis grounds -- particularly if the different AC outlets are on opposite legs of the AC 220 V mains. What happens, for example, is that your mixer, which may be plugged into one outlet, may be floating at one potential, while your transmitter in another room and connected to a different outlet, is floating at a slightly different potential. "Loop current" flows between the grounds of these connected pieces of equipment, via the ground shield of the audio patch cord. This hum has a cause which is similar to the spike noise I described in power supply transformer center taps. That unwanted noise becomes part of the signal, and because it is a low amplitude signal, the smallest ground loop current can add significant noise to the output.
Power supply hum is either built in by design, or avoided by design. Often, simply adding more and larger capacitors may help, but ultimately will not eliminate the ground circuit "eddy currents" that result from an improper layout of the PCB holding the rectifiers, capacitors and regulators. If you are building your power supply yourself, it is revealing to think of the copper foil groundplane as a series of low ohmage resistors. Once you see the whole thing as a voltage divider circuit, it will become more obvious as to how best to isolate your load ground from your transformer return ground.
Always use a voltage regulator. No matter how well the "raw" DC is filtered, total ripple elimination does not become a reality until the circuit has the ability to set the output voltage at least 2 volts lower than the lowest dip in raw DC input to the regulator. Use a regulator that is capable of delivering double the current you expect to need. This will ensure long life and optimum ripple rejection.
Bypass your regulated outputs both with small electrolytic caps and also with ceramic disc capacitors, or similar types with low resistance at high frequencies (the electrolytics alone are ineffective at very high frequencies).
Now that we've covered basic power supply design for clean DC, let's go into making it perform well in an RF-saturated environment.