I think it's better to measure visibility directly.

>> We also considered the loss introduced by non-optimal visibility is the square of (1-visibility).

This seems not correct. This is how I calculated additional loss in entry 1587.

1. Measure voltage of LO, BAB (HWP 0deg), DC offset, visibility (HWP 0deg) 

2. Rotate HWP and measure maximum and minimum of visibility

3. Solve the following equation in terms of V_BAB

(V_max-V_min)/(V_max+V_min-2*V_DC)/(2*sqrt((V_LO-V_DC)*(V_BAB-V_DC))/(V_LO+V_BAB-2*V_DC)) == visibility (HWP 0deg)

4. Additional loss should be 1-V_BAB/V_BAB (HWP 0deg)