Aliasing Error (NF)
Understanding Aliasing Error in Noise Figure
When engineering a superheterodyne receiver, frequency mixers are used to translate high-frequency RF signals down to a lower Intermediate Frequency (IF). However, mixers are inherently mathematically "dumb"—they do not know whether a signal is sitting above the Local Oscillator (LO) frequency or below it. They will take both bands (the desired RF band and the unwanted "Image" band) and fold them simultaneously down into the exact same IF output band. This creates a massive pitfall in RF metrology known as the Aliasing Error when measuring Noise Figure (NF).
During a Noise Figure measurement, a calibrated Noise Source blasts wideband, flat thermal white noise into the receiver covering both the desired RF band and the Image band. The mixer downconverts the noise from the RF band, but it also downconverts the noise from the Image band. These two uncorrelated noise powers sum together at the IF output, effectively doubling the measured noise power. If the engineer or the Noise Figure Analyzer (NFA) is unaware of this, the calculation will claim the receiver is exactly twice as noisy as it actually is, resulting in a 3.01 dB Aliasing Error.
Double-Sideband (DSB) vs. Single-Sideband (SSB)
This discrepancy is formalized as the difference between Double-Sideband (DSB) and Single-Sideband (SSB) Noise Figure. A DSB measurement inherently captures noise from both bands. In a real-world communications system, the antenna usually has an RF bandpass filter that completely blocks the image frequency before it reaches the mixer. Therefore, the true operational performance of the radio is its SSB Noise Figure. To measure the true SSB NF accurately in the lab, an image-reject filter must be inserted between the Noise Source and the mixer to prevent the aliasing.
NFSSB (True operating noise) = NFDSB (Raw measured noise) + 3.01 dB
Warning: If you rely on a raw DSB measurement to spec a system that uses an image-reject filter, your system will perform 3 dB worse in the field than your lab data suggests.
Comparison
| Receiver Architecture | Measurement Setup Needed | Noise Fold-Over Risk | Resulting NF Type |
|---|---|---|---|
| Direct-Conversion (Zero-IF) | Standard broadband noise source | Inherent (RF and Image are the same) | DSB (Matches reality) |
| Superhet (No RF Filter) | Standard broadband noise source | High (Image noise folds down) | DSB Measurement |
| Superhet (With RF Filter) | Noise Source + Pre-selection Filter | Prevented by filter | SSB (True performance) |
Frequently Asked Questions
Why do Direct-Conversion (Zero-IF) receivers use DSB Noise Figure?
In a Zero-IF receiver, the Local Oscillator (LO) is set exactly to the center of the RF carrier frequency. The 'upper sideband' and the 'lower sideband' of the data channel fold directly on top of each other down to DC (0 Hz). Because both sidebands contain valid, desired signal energy, capturing both sidebands of noise accurately reflects how the system operates. Thus, DSB Noise Figure is the correct metric for Zero-IF.
Can a modern Noise Figure Analyzer (NFA) automatically fix the aliasing error?
Not magically. The NFA cannot tell if the noise hitting its detector came from the RF band or the Image band. Most modern NFAs have a 'DSB to SSB' math correction feature you can toggle in the menu, which simply adds 3 dB to the measurement. However, this relies on the assumption that the mixer's gain is perfectly equal in both bands. The only mathematically rigorous way to measure SSB NF is to physically filter the noise source.
What happens if the mixer has different conversion gains for the RF and Image bands?
The 3 dB rule of thumb breaks down. If the mixer has 10 dB of gain at the RF frequency but only 7 dB of gain at the Image frequency (perhaps due to natural roll-off of the components), the aliased image noise will be weaker than the RF noise. The aliasing error might only be 1.5 dB instead of 3.0 dB. This requires complex calibration routines using a spectrum analyzer to characterize.