depending upon your device model, you should consider agilent ads. orcad spice can do the job, if you set it up properly. if you have bsim4 model, then you probably need to use cadence virtuoso.
ORCAD should work well for this design. I suspect that you may not have modeled the transformers properly. They have coupling coefficients and polarities of coupling.
To check for this, run a simulation of just a transformer using an ideal sine wave source and a resistive load in the time domain. See if you get the right waveform shape and polarity (phase) with respect to the primary.
You should post the .dat file from orcad so we can examine it and give you some more suggestions.
Are you trying to do an AC analysis? If so this will not work as it is a small signal analysis based on the transistor parameters at their bias point (no current in this case).
You need to do a time domain transient analysis with full level sine wave drive. You should set the time duration to get the steady state waveforms.
You will have to go to the manufacturers stated and look at their catalog for the part numbers. The catalogs will have formulas for calculating inductances.
Once you get the inductance of the windings of the transformers, you put them into pspice as two inductors. The dots on the schematics are for the first net node number of the transformer. You then need to have a mutual inductance parameter of the coils which includes a K= number. For your simulation try 0.995.
If you want to do a preliminary check, run the simulation at a single frequency and arbitrarily select the inductance of each winding as having a reactance 4x the resistance at that point and use 0.999 for the coupling.
The transformers in the output line and the power filter have equal turns and inductances on each winding.
The transformer between the transistors and the output shifts the impedance. I think that the data sheet you posted gave the turns ratio. The impedance ratio is the square of this number. From this you can get the impedance on the transistor side when you know the load side is 50 ohms.
Once I desingned an 430 Mhz 2W power amplifier. The first two stages were easy to simulate, the results were more or less similar to the real world. The last stage very different. As always maybe is the model but how do you know that the model is good. Many times the Manufacturer says it is not responsible for the models correlation with real life. I think you can use the simulation to compare with the real life but without spending much time in it. High power Rf amplifiers I think that are quite difficult to simulate. S paremeters are not valid anymore. Maybe I am wrong and I will be willing to hear about people that simulate high power RF amplifier.
simulation for High power stage always useless.The most worthful method is load-pull,however it expensive and complex.nonlinear model I think is not acurrate very much current now.The input and output impedance at specified power and frequency is very usefule for design.Although tune and tune is always the final solve for high power amplifier.
The schematic at the start of this topic is for the HF band. This is low enough that the models for the FETs are very good.
These amplifiers are always operated into 50 +j0 ohm loads. This is because they are always followed by a LC network that is either manually or automatically adjusted for this condition.
The automatically adjusted ones are very old technology. The AN/ART-13 used by the US military in 1943 had servo motors to vary the coils and capacitors. This was using an 813 valve/tube and the tuning was done to put the right load on the plates.
Automatic tuners have found their way into amateur equipment in the last 15 years or so.
Load pull charts can be deceptive. Some of them are based on the idea that distortion caused by supply voltage limiting is bad and distortion caused by the elliptic load line caused by reactive loads is acceptable.