Measurements, Testing, and Calibration Additional Practical Test Questions Informational

How do I set up a test bench for characterizing the phase noise of a microwave oscillator?

Setting up a test bench for characterizing the phase noise of a microwave oscillator measures the spectral purity of the oscillator output, quantifying the noise sidebands in units of dBc/Hz at various offset frequencies from the carrier. The test bench options are: direct spectrum analyzer method (the simplest approach: connect the oscillator output to a spectrum analyzer with low phase noise (the analyzer's internal LO must have lower phase noise than the DUT); the spectrum analyzer displays the noise sidebands around the carrier; measure the noise power in a 1 Hz bandwidth at each offset frequency; limitations: the measurement is limited by the analyzer's own phase noise; for offsets less than 100 kHz: most spectrum analyzers cannot measure phase noise better than approximately -120 dBc/Hz), phase noise analyzer (dedicated instrument) (use a dedicated phase noise analyzer (Rohde & Schwarz FSWP, Keysight E5052B, or Keysight N5500A): these instruments use cross-correlation techniques (two independent references) to suppress the instrument's own noise below the DUT's phase noise; measurement floor: -180 dBc/Hz or better at 10 kHz offset; offset range: 1 Hz to 100 MHz; this is the recommended method for accurate, wide-dynamic-range phase noise measurement), and the delay line discriminator method (the oscillator signal is split: one path goes through a long delay line (coaxial cable), the other is direct; the two paths are mixed in a double-balanced mixer; the mixer output voltage is proportional to the frequency fluctuations of the oscillator; advantages: does not require a reference oscillator (self-referenced); limitations: sensitivity limited by the delay line length; cannot measure close-in phase noise (offsets less than approximately 1 kHz) without a very long delay line).

Category: Measurements, Testing, and Calibration

Updated: April 2026

Product Tie-In: VNAs, Signal Generators, Power Meters

Phase Noise Test Bench

Phase noise is one of the most critical specifications for oscillators used in: radar (determines the minimum detectable target velocity and clutter rejection), communications (affects receiver selectivity and adjacent channel interference), and test and measurement (limits the accuracy of frequency and timing measurements).

Test Setup Recommendations

Cabling: Use low-loss, phase-stable cables (semi-rigid or phase-stable flexible cables). Avoid long cables that introduce excess loss and noise
Shielding: Shield the DUT and test setup from external electromagnetic interference. Use a shielded enclosure if necessary
Power supply: Use a low-noise linear DC power supply for the DUT (switching supplies introduce spurs at the switching frequency)
Warm-up: Allow the DUT to warm up for 30-60 minutes before measurement (thermal stability affects close-in phase noise)

Phase Noise Parameters

Phase noise: L(f) = P_ssb(f)/(P_carrier × 1Hz) [dBc/Hz]
Cross-correlation improvement: 10×log₁₀(N_averages)/2
For 100 averages: 10 dB improvement in noise floor
Delay line sensitivity: S_φ = 2πf_offset×τ_delay
Integrated jitter: σ_j = (1/2πf₀)×√(2∫L(f)df)

Common Questions

Frequently Asked Questions

What equipment is needed?

For production-quality phase noise measurements: Rohde & Schwarz FSWP: phase noise analyzer and signal analyzer combined. Measurement floor: -180 dBc/Hz at 10 kHz offset (with cross-correlation). Cost: $40,000-100,000. Keysight E5052B: dedicated signal source analyzer. Excellent sensitivity and wide offset range. Cost: $50,000-80,000. Keysight N5500A Phase Noise Test Set + spectrum analyzer: Modular approach using an external spectrum analyzer. For budget measurements: a high-quality spectrum analyzer (R&S FSW, Keysight PXA) with phase noise measurement personality: approximately -120 to -140 dBc/Hz measurement floor. Cost: $25,000-60,000.

What are common pitfalls?

Common measurement errors: measuring the instrument's noise floor instead of the DUT: the analyzer's own phase noise must be lower than the DUT's at every offset frequency. Use cross-correlation to push the floor below the DUT's noise. Power supply noise: switching power supply spurs appear as discrete spurious tones in the phase noise plot. Always use a linear supply or a well-filtered switching supply. Cable vibration: microphonic effects in cables convert mechanical vibration into phase noise. Use rigid or phase-stable cables and avoid touching the setup during measurement. Temperature drift: thermal changes cause slow frequency drift that appears as elevated close-in noise (offsets less than 100 Hz). Let the DUT and setup thermally stabilize.

How do I interpret the results?

A phase noise plot shows L(f) in dBc/Hz on the Y-axis vs. offset frequency on the X-axis: close-in (1 Hz-1 kHz offset): dominated by flicker FM noise (1/f^3 slope). Affected by: oscillator design, resonator Q, and active device flicker noise. 1 kHz-100 kHz offset: typically shows 1/f^2 slope (white FM noise or random-walk phase noise). This region is critical for: radar clutter rejection and communication channel selectivity. 100 kHz-10 MHz offset: approaches the broadband noise floor (white phase noise, flat slope). Determined by: the amplifier noise figure and the oscillator output power. Spurs: discrete tones above the noise floor indicate: power supply switching, reference leakage, or mechanical vibration.

Keep Reading