Dana designed a phase shifter that input a single hf freq, and output 6 copies of this, each with a programmable amplitude and phase shift. This allowed us to generate the circular polarizations needed for the ionospheric studies.
The technique split an HF signal
into 6 copies (one for each dipole). Each of these signals was
input to the lo inputs of 2 mixers (with a 90 deg phase
shift). The IF input of the mixers was driven by dacs outputing DC
levels. The output of the two mixers was then combined and output
to a dipole. By changing the amplitude of the dac outputs, the
amplitude and phase of the signal to a dipole element could be
set.
This worked fine until we tried to do AM modulation, and dual
frequency experiments. Since the lo input of a mixer is
non-linear, what we got out was not what we put in. (more info)
Mike came up with a technique to get around
this problem.