I've seen that, despite some phased array antennas are rectangular, many of them are not. They are octagonal, decagonal and even more.
Why are they built in this way?
Electrical Engineering Stack Exchange is a question and answer site for electronics and electrical engineering professionals, students, and enthusiasts. It only takes a minute to sign up.Sign up to join this community
There are several reasons for those shapes.
First, the cut off/rounded corners make it easier to fit into the nose of an aircraft. This is true of the phased arrays, both active and passive, that are found on the B-1B, F-16 (newer versions), F-22, F-35, etc.
Most square or rectangular arrays are found in ground or ship born applications, where packaging constraints aren't quite as severe. Typical of such an array is the Ground/Air Task Oriented Radar (G/ATOR) for the United States Marine Corps. Picture below is from Defence Blog:
Second, if the array is an AESA (Active Electronically Steered Array), there's a whole bunch of expensive electronics (LNAs, filters, phase shifters, attenuators) behind each antenna element. For cost reasons, you like to keep the number of elements to the minimum needed to provide the aperture gain needed.
Most arrays use amplitude weighting in both the X and Y axis of the antenna in receive in order to control the sidelobe level. This amplitude weighting means that the end elements along both axis, and especially the elements in what would be the square corner of the array are heavily attenuated and so contribute little to the overall performance of the array. So in the case of a receive-only antenna those elements are removed in order to save cost.
For an antenna that's used to transmit (and receive), the name of the game is usually to put as much power out as possible. That is, to maximize the Effective Isotropic Radiated Power, or EIRP. In that application, edge or corner elements are usually left in place to take advantage of the transmit output power of those elements, even if they do not contribute much to receive operation. That is, assuming they're not removed in order to fit the array into the volume constraints.
Added Antenna Below
Here's an array antenna that is similar to some that I have worked with in the past. It is shaped like a parallelogram, and the elements are located parallel to the tilted side.
An antenna shaped like this has certain characteristics related to sidelobes that make it attractive for some applications. Note that the individual radiating elements, denoted by the "X" in the diagram, are not located on a nice Cartesian grid. This makes the computation of the element's phase shifts a bit more difficult, but not overly so.