Field Guide · term

Also known as: image suppression, image-reject mixing

Image rejection is the suppression of the unwanted image frequency — the second input frequency a mixer folds onto the same output as the wanted signal.1 Any mixer converts two input frequencies, symmetric about the local oscillator, down to the same intermediate frequency; unless one of them is removed, a signal at the image frequency interferes with the one you want. Image rejection is how a receiver keeps the image out. Its effectiveness is quoted as the image-rejection ratio (IMRR), the dB by which the image is attenuated relative to the wanted signal.

LO wanted image equal IF each side of LO
The wanted signal and its image sit symmetrically around the local oscillator; image rejection lets the wanted one through and cancels the mirror.

How it works

There are two broad strategies.

Filtering (image-reject filter). In a superheterodyne receiver the image is separated from the wanted signal by twice the IF. A bandpass filter ahead of the mixer, tuned to the wanted band, attenuates the image before mixing. The higher the IF, the further away the image and the easier the filter — the classic reason superhets use a high first IF.

Phasing (image-reject mixers). Instead of filtering, these architectures use quadrature signals so the image cancels by interference:

  • Hartley. The signal is mixed with two local oscillator phases 90° apart, one branch is phase-shifted a further 90° (a Hilbert transform / broadband 90° network), and the two are summed. The wanted signal adds in phase while the image cancels.
  • Weaver. Replaces the troublesome broadband 90° phase-shift network with a second pair of mixers at a low frequency, achieving the same cancellation with two mixing stages and no wideband Hilbert network.

Both phasing methods are really doing complex (I/Q) signal processing: keeping the sign of the frequency offset, which a single real mixer discards, so the two sides of the oscillator stay distinguishable.

In practice

Real image rejection is finite. Filter designs are limited by filter sharpness; phasing designs are limited by how precisely the 90° phase and the branch gains are matched — the same IQ imbalance that produces mirror artefacts in a zero-IF receiver caps the IMRR of a Hartley or Weaver mixer, typically at 30–40 dB uncorrected and much higher after calibration. Modern receivers often combine a modest image-reject mixer with digital correction that estimates and cancels the residual imbalance adaptively.

Relevance to SDR

Image rejection is central to how SDR front ends and their tuners are built. The quadrature tuners in RTL-SDR dongles and other zero-/low-IF SDRs are image-reject mixers, and their residual imbalance is exactly what leaves faint mirror-image signals on the waterfall. Many SDR programs apply digital IQ-correction to push those images down, and low-IF tuner modes lean directly on image rejection to keep a neighbouring channel from folding onto the wanted one.

GopherTrunk works on the corrected IQ its source provides; it does not build mixers, but the images that survive imperfect rejection are among the artefacts its channel filtering must tolerate, and the concept explains why a strong station can appear mirrored across the tuned frequency.

Sources

  1. Image response — Wikipedia, on the mixer image and the filtering and quadrature techniques used to reject it. 

See also