Welcome to our site! EDAboard.com is an international Electronic Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Depends on the method used. If you are generating a distorting signal and adding it linearly to the main signal before transmit amplifier, you are adding delay so that the distorting signal vector is 180 degrees out of phase over a wide frequency range with the amplifier generated distorting signal. If it were narrowband modulation, you could get away with just a phase shifter to perform the cancelation. With wideband modulation, a phase shifter would not be able to cancel out all of the distortion frequency components simultaneously, so you generate the phase shift with a time delay.
Well, If I were doing it I would build a circuit with two connectors for attaching a coaxial delay line to, and in the lab I would input my modulation to the predistorter, pass the resulting signal thru the transmitter, and look at 2tone 3rd order products on a spectrum analyzer. I would then tweek the delay line length until they were nulled over the biggest portion of their bandwidth.
In practical circuits, you would probably have both amplitude, phase, and delay line lengths to fool around width until you got it right.
If you want to calculate such a thing, you would need a non-linear simulation program that could simulate both the predistorter and transmit amplifier non-linearities accurately, and you would do the same as above.
If you are up on FFT processing in something like Mathcad or Matlab, you could simulate it all to a first order there, modeling the predistorter like a nonlinear diode I-V curve, etc. Would not be very accurate though.