Do I need cross strips to compute the PPAC?
Cross-strips always help any calibration, self-calibration, or pcgrid generation.
However, it does not mean they are always absolutely necessary: it all depends on the error shape in camera space.
For DMC, PPAC calibration can help better fitting of GPS/INS, especially with very tight std_devs.
In any case, I would advise using PPAC calibration only if you have good distribution of control points.
If you do, run PPAC/FL calibration with all controls, then key in the calibrated values for PPAC/FL into camera file, and re-run AT with most of control points as check points. If you see improvement, use it.