A: Phase noise describes random fluctuations in the phase of a signal, appearing as skirt-like spectral spreading around the carrier frequency. It degrades system performance in radar, communications, and precision timing, so accurate measurement is essential for oscillator characterization and system design. Q: How does a spectrum analyzer measure phase noise?
A: A spectrum analyzer directly displays the signal's power spectral density. By observing the carrier and its surrounding noise skirt, engineers read phase noise as dBc/Hz at various frequency offsets. This method is simple, requires no reference oscillator, and provides a fast overview. However, it is limited by the analyzer's own phase noise floor (typically −110 to −120 dBc/Hz at 10 kHz offset) and dynamic range. It works well for evaluating noisy or moderate-quality oscillators but struggles with ultra-low-phase-noise sources. Q: How does the phase detector method work?
A: The phase detector (or phase-comparator) method uses a dedicated reference oscillator phase-locked to the device under test (DUT) at a 90° quadrature condition. At quadrature, the mixer output voltage is proportional to the instantaneous phase difference between the two signals, effectively converting phase fluctuations into a low-frequency voltage noise. This voltage is measured with a low-noise baseband spectrum analyzer or FFT analyzer. Advantages include significantly lower measurement floors (−170 dBc/Hz achievable) and direct sensitivity to phase rather than amplitude. The main limitation is the need for a reference source with phase noise equal to or better than the DUT, otherwise the result conflates both sources' noise. Q: What is the cross-correlation method, and why is it superior?
A: The cross-correlation technique addresses the reference source limitation by using two independent measurement channels, each with its own reference oscillator, simultaneously measuring the same DUT. Since the DUT's phase noise is correlated across both channels while each reference's noise is uncorrelated, averaging N cross-correlations suppresses uncorrelated reference noise by 10·log₁₀(N) dB. For example, 10,000 correlations yield 40 dB of improvement. This enables measurement floors below −185 dBc/Hz, sufficient for the best quartz and sapphire oscillators. Modern dedicated phase noise analyzers (e.g., Keysight E5052B, Rohde & Schwarz FSWP) implement this method. Q: When should each method be used?
| Method | Best For | Typical Floor |
|---|---|---|
| Spectrum Analyzer | Quick screening, noisy oscillators | −120 dBc/Hz |
| Phase Detector | Single high-quality oscillator characterization | −170 dBc/Hz |
| Cross-Correlation | Ultra-low-noise oscillators, highest accuracy | < −185 dBc/Hz |
A: Choose based on required sensitivity, available reference quality, and measurement speed. For production screening, a spectrum analyzer suffices. For R&D of high-performance oscillators, cross-correlation is the gold standard.
Need precision timing solutions? Get a quote from BRIDZA