Reciprocal Mixing

Reciprocal mixing is what happens when a local oscillator's own imperfect spectral purity lets a strong nearby interferer raise the noise floor directly under a desired, much weaker signal — a receiver impairment that has nothing to do with image frequencies or I/Q path matching, and exists purely because no real local oscillator is the single, perfectly pure tone an ideal mixer diagram assumes.

A real local oscillator is never a perfectly pure single-frequency tone — its output spreads a small amount of energy continuously around the ideal carrier frequency, a skirt of phase noise that every practical oscillator exhibits to some degree. This skirt is usually harmless when a receiver is only processing weak, well-behaved signals, since the skirt's energy at any given offset is tiny compared to the main LO tone. The problem appears specifically when a strong interferer sits nearby in frequency to a desired weak signal: the interferer doesn't just mix against the LO's main tone the way the desired signal does — it also mixes against every bit of phase-noise skirt energy sitting at the specific offset that happens to map the interferer down to exactly the same intermediate frequency the desired signal occupies.

The result is a locally elevated noise floor sitting right underneath the desired signal, created entirely by the interferer's own strength combined with the LO's imperfection, with no filtering able to separate the two after the fact. Because this degraded noise floor originates from mixing the interferer with LO phase noise rather than from any noise inherent to the desired signal's own path, it cannot be removed by any filter downstream of the mixer — the corruption happens at the exact moment of frequency conversion, baked into the IF signal before any subsequent stage ever sees it. A receiver experiencing significant reciprocal mixing shows reduced sensitivity specifically in the presence of a strong nearby interferer, even though that same receiver might perform perfectly well with the identical weak desired signal and no interferer present at all.

How close the interferer sits to the desired signal directly determines how much of the LO's phase noise actually gets involved, because phase noise is not flat across frequency offset — it is characteristically worst close to the carrier and falls off at larger offsets. A nearby interferer only a small frequency separation away from the desired signal requires correspondingly small LO offset frequencies to produce the mixing product that lands on the desired IF — exactly the close-in region where phase noise tends to be at its highest. An interferer sitting much farther away requires a correspondingly larger LO offset to reach the same IF, landing in a region of the phase-noise spectrum that has typically already fallen substantially from its close-in peak, which is why reciprocal mixing is usually a far more severe problem for closely-spaced interferers than for distant ones even at comparable interferer power levels.

Reciprocal mixing is a genuinely separate receiver impairment from image rejection and IQ imbalance, both already covered in this series, and the distinction matters because none of those earlier fixes touch this one. Image rejection exists because two different RF frequencies fold onto the same IF through the mixing process itself; IQ imbalance exists because real I/Q paths never match perfectly. Reciprocal mixing exists purely because of LO spectral impurity, and it would still degrade receiver sensitivity in a hypothetical receiver with perfect image rejection and perfectly matched I/Q paths — including the zero-IF direct-conversion architecture already covered, where there is no image problem at all by construction. A perfect receiver architecture on every other axis this series has covered can still be desensitized badly by reciprocal mixing if its LO's phase noise is poor enough.

This is precisely why LO and frequency-synthesizer phase-noise specifications are treated as a first-order receiver design parameter rather than a secondary concern, particularly for any application where strong nearby interferers are a realistic operating condition. A receiver intended for a dense, interference-heavy environment — closely packed cellular channels, a crowded ISM band, any scenario where a much stronger signal can realistically sit close in frequency to a much weaker one of interest — needs a correspondingly cleaner local oscillator than one intended for a quiet, uncontested spectrum, independent of whatever other performance specifications the rest of the receiver chain achieves.

Reciprocal mixing closes out the set of independent receiver impairments this series has covered — image rejection, IQ imbalance, and reciprocal mixing are three genuinely separate problems, each traceable to a different physical origin, each requiring its own specific fix, and each capable of degrading receiver performance even when the other two are solved perfectly. A receiver design that only tracks one or two of these three specifications is incomplete, because the remaining one can still be the actual limiting factor on real-world sensitivity no matter how well the others were addressed.