Measurements, Testing, and Calibration Noise and Specialized Measurements Informational

How do I measure the phase linearity of a filter across its passband?

Phase linearity of a filter is measured by analyzing the S21 phase response across the filter passband using a VNA. A perfectly linear-phase filter has a straight-line phase vs frequency relationship (constant group delay). Any deviation from straightness indicates phase distortion that will cause signal distortion. Procedure: (1) Calibrate the VNA with a full two-port calibration at the filter reference planes. (2) Connect the filter. Measure S21 (magnitude and phase) across the passband and transition bands. (3) Analyze the phase: method 1: plot the S21 phase vs frequency. A linear-phase filter produces a straight line with slope = -2×pi×tau_d (the constant group delay). Deviations from the straight line are the phase nonlinearity. Method 2: compute the group delay: tau_g(f) = -(1/360) × d(phase)/df. A linear-phase filter has constant group delay. The group delay deviation (GDD) from the average is the measure of phase nonlinearity: GDD(f) = tau_g(f) - tau_g_average. Maximum GDD across the passband is the specification. Method 3: fit a straight line (linear regression) to the phase data. The residual (difference between measured phase and the best-fit line) at each frequency is the phase deviation. Peak-to-peak phase deviation across the passband is the specification. (4) Specify: phase linearity is reported as maximum group delay deviation (in ns) or maximum phase deviation from linear (in degrees). Specifications: satellite transponder filter: GDD < 10-50 ns across 36 MHz BW. Microwave backhaul filter: GDD < 1-5 ns across 28-56 MHz BW. Digital radio filter: GDD < 5-20 ns across 10-25 MHz BW.
Category: Measurements, Testing, and Calibration
Updated: April 2026
Product Tie-In: Noise Sources, Analyzers, Calibration Standards

Filter Phase Linearity Measurement

Phase linearity is one of the most important filter specifications for digital communication systems because phase distortion causes intersymbol interference (ISI), which directly limits the achievable data rate and modulation order.

ParameterSOLT CalTRL CaleCal
AccuracyGoodExcellentGood-very good
Standards Needed4 (S,O,L,T)3 (T,R,L)1 (module)
BandwidthBroadbandBand-limitedBroadband
Setup Time5-10 min10-20 min1-2 min
Best ForCoaxial, generalOn-wafer, waveguideProduction, speed
  • Performance verification: confirm specifications against the application requirements before finalizing the design
  • Environmental factors: temperature range, humidity, and vibration affect long-term reliability and parameter drift
  • Cost vs. performance: evaluate whether the application demands premium components or standard commercial grades
Common Questions

Frequently Asked Questions

How much group delay variation is acceptable?

Depends on the signal bandwidth and modulation order: for narrow-bandwidth signals (< 1 MHz): group delay variations of 100-1000 ns are tolerable (the signal bandwidth is narrow enough that all frequency components experience nearly the same delay). For 20 MHz LTE (up to 64-QAM): GDD < 50-100 ns is acceptable (the adaptive equalizer in the receiver handles this). For 100 MHz 5G NR (256-QAM): GDD < 10-20 ns. For 400 MHz 5G NR (256-QAM): GDD < 5 ns. For analog video: GDD < 50 ns across 6 MHz (to avoid group delay distortion in the picture). For radar (pulse compression): GDD < 1/(10 × BW) for acceptable sidelobe degradation.

Why does group delay peak at the filter band edges?

Group delay peaks at the filter band edges because the phase changes most rapidly in the transition band (the filter is transitioning from passband to stopband). The steep phase slope corresponds to high group delay. For a Chebyshev filter: the group delay peak at the band edge is approximately: tau_peak ≈ N / (pi × BW) × Q_factor, where N is the filter order and Q_factor accounts for the passband ripple (higher ripple = higher Q = higher group delay peak). An N=5 Chebyshev filter with 0.1 dB ripple and 50 MHz bandwidth: tau_peak ≈ 5 / (pi × 50e6) × 1.1 ≈ 35 ns. An N=5 Butterworth: tau_peak ≈ 5 / (pi × 50e6) × 1.0 ≈ 32 ns (similar but slightly lower). An N=5 Bessel: tau_peak ≈ 5 / (pi × 50e6) × 0.3 ≈ 10 ns (much lower, by design).

Can I compensate phase nonlinearity digitally?

Yes, digital equalization can compensate for filter phase nonlinearity if: (1) The signal is sampled with sufficient resolution (ADC bits and sample rate). (2) The filter phase distortion is characterized (known or estimated from training symbols). (3) The noise level is sufficient to support the equalized SNR. In modern communications: the receiver digital equalizer (OFDM inherently handles this with per subcarrier equalization) automatically corrects for filter phase distortion. No analog equalizer is needed. However: the phase distortion still creates a noise penalty. The equalizer amplifies noise at frequencies where the filter gain is low (band edges), reducing the effective SNR. Severe phase distortion can make some subcarriers unusable, reducing the achievable data rate.

Need expert RF components?

Request a Quote

RF Essentials supplies precision components for noise-critical, high-linearity, and impedance-matched systems.

Get in Touch