Reply to thread

Question is if you can live with the differential phase approach. E.g. both all-pass chains have a total phase shift of 0 to 540 degree, but the frequency dependence is different so that you achieve a difference of 90 over the designed frequency range.

Example:

[ATTACH=full]198476[/ATTACH]

A pure 90 degree phase shift (Hilbert transform) can be practically only implemented with a digital (FIR) filter and involves a relative large delay.