Implementing Direct RF Sampling at Sub-Nyquist Rate for Error Vector Magnitude Measurements

We investigate an under-sampling method for error vector magnitude (EVM) measurements. The approach is to simulate the demodulation of the time-domain waveform with both conventional oversampling and sub-Nyquist sampling methods. The dominant increase in EVM from sub-Nyquist sampling compared to oversampling is due to noise aliased from higher-order Nyquist zones. Simulations quantify this behavior by comparing the additive white noise response of under- and over-sampling receivers. We also examined the jitter effects in the simulation and found that additive noise dominates the degradation in EVM changes for the root mean square (RMS) jitter no more than 1 ps. As the Nyquist rate becomes more and more difficult to reach in the high-frequency communication, the sub-Nyquist sampling method provides an alternative by use of a sophisticated yet agile digital processing algorithm.