Perhaps you were one of the thousands of new hams to jump on the air after upgrading from tech to general when the code requirement was lowered. You happily put up an antenna for 20 meters, got the swr down to where it was acceptable, and went on the air. To your surprise and disappointment, you were told that your audio was no good, that you had "RF in your audio." This is also known as RF feedback.
Perhaps the most difficult part of setting up a radio station is getting a good RF ground. DC grounds are easy to make, for the length of the wire to the ground connection is not critical; however, with radio-frequency currents, it is extremely so.
Think of it this way; a wire longer than ten percent of one wavelength begins to lose its effectiveness as a radio ground. This is about twenty-four feet on the 80m band, twelve feet on the 40m band, and drops to six feet on 20m! This is why so many hams who have no trouble with RF feedback on 80m or 40m experience it on 20m or 10m.
Obviously, the best solution is to have the wire from your transmitter to the ground connections less than three feet long. In most installations, though, this isn't possible. There are other solutions.
First, use a resonant dipole with a tuned coaxial feedline. The side of the dipole connected to the shield of the coax acts as a counterpoise, an artificial ground effective only at resonant frequencies. If the feedline is tuned to a multiple of a half-wavelength (don't forget to figure in the velocity factor of the cable when making your measurements) the shield at both ends of the cable will automatically be at ground potential.
You can also install a counterpoise on the ground of your transmitter, and one on the ground connection itself. This is done by cutting a wire one-quarter wave long and attaching one end to the desired ground point, and leaving the far end free. WARNING! The far end of the wire will have a high RF potential on it! Hide it so no one can touch it by accident, and don't let it touch anything metallic!
MFJ makes an artificial RF ground, which works by tuning the ground wire so that the station end has a low RF potential. (They make numerous other useful devices as well.)
Finally, an RF ground which does not perform well when the air is saturated with radio-frequency currents is often adequate in a less RF intensive environment. In other words, moving the antenna farther away from the station often helps or eliminates the problem completely. The author of these pages has for the last half-dozen years used a ground that has a connection six feet long. When experimenting with untuned open-wire feedline (which would have allowed multiple-band operation) and antennas close to the house, problems were encountered with RF feedback, especially on 20m. When the antennas were moved away from the house, and resonant antennas were used for each band, with tuned feedlines, the problem was resolved, all without any changes to the ground system itself!