Also known as: mixer image, image response
The image frequency is the unwanted input frequency that a mixer converts to the same intermediate frequency as the wanted signal.1 Because a mixer responds to the absolute value of the difference between an input and the local oscillator, two input frequencies — one above the oscillator and one below, both offset by the IF — produce identical outputs. One is the signal you want; the other is its image, and any energy there lands on top of your signal unless it is filtered out first.
How it works
Suppose the receiver wants a signal at frequency fRF and sets its oscillator to fLO, producing an IF of fIF = |fRF − fLO|. If the wanted signal is above the oscillator (fRF = fLO + fIF), then a signal at fLO − fIF — the same distance below — also mixes down to exactly fIF. That lower frequency is the image. The two are always separated by twice the IF, and after mixing they are indistinguishable: no amount of IF filtering can separate them, because they occupy the identical IF band.
The only cure is to act before the mixer. Attenuate the image while it is still at RF — either with a preselector filter or by using a quadrature image-reject architecture that cancels it. This is why the choice of IF is a design trade-off: a high IF pushes the image far from the wanted signal, where a modest RF filter easily rejects it, but demands a higher-frequency IF stage; a low IF is easier to process but places the image close by, where only a very selective filter can help.
Variants
A related hazard is the half-IF spur: an interferer halfway between the wanted signal and the oscillator, whose second harmonic mixes with the oscillator’s second harmonic to land in the IF. Mixers also produce other spurious responses at combinations m·fLO ± n·fRF, but the primary image is the dominant one and the reason preselection exists.
Relevance to SDR
In zero-IF and low-IF SDR front ends the IF is 0 (or nearly 0), so a signal’s image is its own mirror around the tuned frequency — the faint reflected copies seen across the center of an SDR waterfall when the tuner’s IQ imbalance is imperfect. In IF-sampling and superheterodyne designs the image is a distinct band twice-the-IF away that the preselector must reject. Recognising an image — it moves in the opposite direction to real signals as you retune, and mirrors around the LO or the tuning center — is a basic SDR troubleshooting skill.
GopherTrunk decodes the IQ stream after the front end, so it inherits whatever image suppression the device achieved; a strong out-of-plan signal appearing mirrored across the tuned frequency is an image artefact of the receiver, not of GopherTrunk’s DSP.
Sources
-
Image response — Wikipedia, on the mixer image frequency, its 2×IF separation, and why it must be filtered before mixing. ↩