AD8351 output voltage swing

Apologies for the long delay in responding to this question. My original response below got stuck in the system.

There is a plot in the datasheet that is will answer some of your questions. Take a look at Figure 16 on page 9 of the datasheet. It's a plot of OP1dB in dBm so we need to convert that into Vpp. For starters the 4.75Vpp output swing in the spec table converts to 12.7dBm which the compression number that you also see in the spec tables (the number varies with frequency but is in the 13dBm range). To convert from pp to dBm, you muliply the pp voltage by 0.5x0.707, to get rms, then convert that to power (V^2/R) using 150 or 1000 ohms, the calculate 10log10(Power(watts)/1mW). 

There are a few things that are apparent from this plot. Firstly a 5V supply gives you a higher output signal. So if you did take the supply up to 5.5, you would get a little more output signal. 

The load resistance has a big impact on output voltage swing and output power. When you increase the load resistance, you get a lot more voltage swing but less RF output power. So the second trace down (approx 9.5dBm for a 5V supply and a load resistance) corresponds to a peak to peak voltage swing of 8.44Vpp differential and 4.22Vpp on each output. 

So if you want more voltage swing and have the ability to change your output load resistor, then a lighter load (bigger value of Rload) would be a good way to go. 

If you do operate in this mode (i.e. RL=1k), then the value of VOCM  is going to matter. To achieve this output swing, you will probably need VOCM at 2.5V because each differential output arm is swinging over 4V pp. 

If you go back to loading the part with 150 ohms, each differential arm will be swinging 2.375Vpp (i.e. 4.75Vppdiff/2). In this case, there is more scope to move VOCM around because your output swing is much further away from the rails (each output will swing 2.5V +/-1.3125). 

One other thing, the 4.75Vpp swing is the output voltage at 1dB compression. This means that if you drive the input harder, you will get a bit more output power before the output saturates. Usually, P1dB to Psat is 1-2 dB. If we conservatively say it's an extra 1dB, that would take the swing to 5.3Vpp differential. But now you are operating at saturation, so signal distortion will be significant.