Identifying the Structure of Nonlinear Perturbations in Mixers using Multisine Signals

Telecommunication transceivers (e.g., radio, Global System for Mobile communications [GSM] Bluetooth, wireless local area network [WLAN] etc.) use mixers to translate the frequencies of signals carrying information. Mathematically, this ideal nonlinear frequency translation behavior corresponds to a multiplication. Real mixers also present nonideal nonlinear behavior that result in the creation of higher-order harmonics, additional intermodulation products, and even linear spurious response terms. For high performance designs, a microwave designer needs to know how close a real mixer is t an ideal frequency translator. If it is possible to understand where the nonideal nonlinearities are located in the mixer, the design can be optimized to reduce unwanted perturbation levels. To answer these concerns, we formally define an ideal mixer as a block-oriented model that contains a perfect multiplier but that is surrounded by two linear time-invariant systems. This model must then be extended to take into account the nonlinearities. There are two possible locations in the signal path that might introduce nonlinearity - the input path or the output path of the mixer. To identify the position of the nonlinearity, a custom designed multisine signal can be used to excite the signal path of the mixer. This special random-phase multisine signal with a random harmonic grid contains several carefully selected spectral lines that are left unexited. These holes in the spectrum of the signal act as detectors for the presence and the location of nonlinearities that deviate from the ideal. First we test the proposed detection algorithm in a simulation to show that it can work in an ideally controlled environment. Afterwards, we perform measurements on a real mixer excited with a special odd randomphase multisine with a two-port large signal network analyzer (LSNA) and detect the position of the unwanted nonlinearities in the device model.