Val,
Since the Gaussian filter is not a zero-ISI filter, both the filter length and data itself will affect the deviation. You would have to make sure the lengths match, and then do some sort of sync word synchronization to ensure you were measuring the same exact bit sequence in the software. And even then there might be slight differences in the implementation that lead to freq deltas, but I suspect that would eliminate most of the difference.
If you need ultimate control over the symbols and deviation, you could generate your own I/Q waveform and transmit it with the VSG’s.
Otherwise, what you are seeing seems normal.
