Notes on Phase Noise
Phase noise is the alligator in the pond. For many of us, it's an amorphous issue, lying beneath the surface. Sensitivity and IMD get a lot more press, and we tend to pay more attention to them.

In reality, phase noise is a major problem, especially for radios that use sythesizers. As more low power radios move away from crystal and LC oscillators, more of us in the QRP community will get acquainted with phase noise.

The ARRL Handbook has good information on phase noise (otherwise known as "reciprocal mixing"). Here are some interesting words from Chapter 14:

A poor oscillator can have significant noise sidebands extending out many tens of kilohertz on either side of its carrier. This is the same, as far as the signals in the receiver IF are concerned, as if the LO was clean and every signal entering the mixer had those noise sidebands. Not only will the wanted signal (and its noise sidebands) be received, but the noise sidebands added by the LO to signals near, but outside, the receiver's IF passband will overlap it. If the band is busy, each of many signals present will add its set of noise sidebands—and the effect is cumulative. This produces the appearance of a high background-noise level on the band. Many hams tend to accept this, blaming 'conditions.'

The ARRL tests for transmitter phase noise, but not for receiver phase noise. They are not the same, even though the transmitter and receiver in most radios share the same oscillators. There are two reasons: circuits behave differently in the presence of strong RF fields, and the receiver's selectivity can insulate the desired signal from some of the receiver's oscillators, no matter how noisy they are.

We test for phase noise at by listening to a frequency 10 kHz from the signal generator's carrier. This is the method described by Wes Hayward, W7ZOI, in Introduction to Radio Frequency Design. Sometimes, this test cannot be completed, either because the receiver is overloading, or because the receiver's selectivity is so poor that the carrier leaks into our test frequency.

Our bar chart for phase noise includes bars for "poor," "fair," "good," and "excellent." These are based on an excellent set of articles by John Grebenkemper, KI6WX, in the March and April, 1988 issues of QST.

We use an HP 8640B signal generator to generate the carrier for phase noise testing. For rigorous testing, it would be better to use a low-noise crystal oscillator. However, the HP 8640B has excellent phase noise performance and has not compromised any testing we've attempted so far. And it is much more convenient to use than a crystal oscillator.