What is the effect of impedance mismatch on the error vector magnitude of a modulated signal?

Impedance mismatch affects the EVM (Error Vector Magnitude) of a modulated signal by introducing amplitude and phase distortion that varies across the signal bandwidth: (1) Mechanism: a modulated signal has a finite bandwidth (e.g., 20 MHz for LTE, 100 MHz for 5G NR). The impedance mismatch creates frequency-dependent amplitude and phase ripple across this bandwidth. Different frequency components of the modulated signal experience different gain and phase shift, distorting the signal constellation. (2) EVM contribution from mismatch: the EVM due to amplitude ripple: EVM_amp ≈ Ripple_rms / sqrt(2) (as a fraction, not in dB). For a 0.5 dB peak-to-peak ripple across the channel: EVM_amp ≈ 0.029 (2.9%). For 1.0 dB ripple: EVM_amp ≈ 0.058 (5.8%). The EVM due to group delay variation: EVM_gd ≈ 2 × pi × B × delta_tau_rms. Where B = signal bandwidth and delta_tau_rms = RMS group delay variation. For 100 MHz signal bandwidth and 0.5 ns group delay variation: EVM_gd ≈ 2×pi×100e6×0.5e-9 = 0.314 (31.4%). This shows that group delay variation is often more damaging than amplitude ripple for wideband signals. (3) Practical impact: for 64-QAM (requires EVM < 8%): the mismatch-induced EVM should be < 3-4% (leaving margin for other EVM sources: PA nonlinearity, phase noise, ADC quantization). This requires VSWR < 1.5 at all interfaces and group delay variation < 0.3 ns across the channel bandwidth. For 256-QAM (requires EVM < 3.5%): even tighter matching is needed: VSWR < 1.3 and group delay variation < 0.1 ns. (4) Mitigation: minimize VSWR at all interfaces in the signal chain, keep connections short to push the ripple period outside the channel bandwidth, use digital equalization (an adaptive equalizer in the receiver can compensate for known or slowly varying mismatch effects), and add attenuator pads between stages (after the LNA) to absorb reflections.