Phase shifts in antenna arrays

Question

Let's consider a simple antenna array. For instance, this three-elements equispaced dipole array:

Each dipole is connected to a signal voltage source. It's well known that it's possible to choose the phase delay between the input voltage sources of the elements in order to decide the direction of the maximum radiated power (beam scanning for phased arrays). It's known also that, if all the elements have the same voltage signal, without phase delays between them, the peak direction is orthogonal to the array axis.

I understand this in theory. But in practice, let's consider the previous picture, in which all voltage sources are supposed to be identical (let's call their signal $V_i$ like in the picture), without phase delays. Each dipole has two arms, which I have indicated with A and B.

Which is the connection (between $V_i$ and the dipole arms A and B), with respect to which we say there are phase delays or not?

Does 0 phase delay means all + terminals of all sources connected to A arms (so, not adjacent arms of adjacent dipoles)? Or does it means + terminals connected to A for an element and to B for the adjacent element?

The voltage source is alsways $V_i$, but according to how we connect its poles, we may introduce a 180° phase shift, and so a 0.5T delay (with T period of the input signal).

Answer 1

In your diagram, the + terminals would all be connected to the A arms. Assuming a half wave spacing of the dipole feedpoints, the EM fields can cancel parallel to the antenna elements, then add up to a stronger waveform propagating in the perpendicular direction.

Assuming all the voltage sources are isolated, the EM field will simply keep "adding up" right to left; and the EM coupling between dipoles might generate a much higher voltage differential across the entire array than within any individual dipole.