There might be a case for arguing that the ideal primary beam pattern is an omnidirectional one, i.e. your instrument sees the whole sky. In fact I'd encourage the astronomers, the software people and the hardware people to get into their respective camps on this matter and have a big fight about it.

Before that happens though, let's just introduce the sort of primary beam you get in an ideal world. It's the simplest direction dependent effect: the gain of your instrument drops as you move away from the phase centre, eventually it drops to nothing and stays there until the horizon. It doesn't change with time and it only changes with frequency in a way you don't mind.

Ladies and gentlemen: the WSRT cos-cubed beam.

What we'll do now is take the KAT-7 simulation that has appeared previously with the sky model derived from SUMSS. We'll quickly repeat the initial 'Simulations 101' setup for comparison. Then we'll repeat this with the analytic cos-cubed model for the primary beam and compare the results.