I was at Agilent workshop last year when they presented nonlinear/ large S par NA. Nonlinear NA is for the moment in development phase, it is not a piece of equipment but a number of instruments, including a standard NA, harmonic tuners etc. You can measure, as stated above, a complete nonlinear behaviour of your device. Results shown are quite promising for accurate prediction e.g. of PA performance concerning IMD and other nonlinear parameters. It is quite a procedure to use, a simple measurement is a whole day exersise including very complicated callibration. Even a mesured data interpretation is not straight forward, Agilent can do it for you as a service.
flyhigh