We have another issue regarding phase control. Please see attached slides for more detail of it .
Look forward to hearing from you, thanks.
Can I ask you to use 1kHz test signal AND 5kHz. Also, what sample rate are you using? This information will tell us the precision/performance of the existing Hilbert Transform closer to DC and FS/2
Yes, It's the same problem. The sample rate is 48kHz.
As you have discovered, the Hilbert Transform provides a 90-degree phase shift between its two outputs, yet with no guaranteed phase relationship between its input and either output. All Hilbert implementations (and all filters!) by nature suffer a delay from input to output(s), affecting their phase relationship from input to output(s). A FIR-based Hilbert's delay would be constant over frequency ("linear phase") -- so one could add a compensating delay of half the FIR's length to signals which need have a fixed phase relationship to the Hilbert-ransformed signals. Unfortunately an FIR Hilbert that works down to bass frequencies won't fit in a ADAU1701, so the provided Hilbert transform is a IRR (recursive) implementation. Thus its delay varies with frequency -- see ADAU1446 Hilbert transform response for a rough measurement of the IIR Hilbert's delay vs. frequency. Since the delay isn't constant, you cannot compensate for this Hilbert's phase shift with a fixed compensating delay. You can, however, use another Hilbert transform as a compensating delay.
Perhaps what you're looking for is an absolute phase relationship between for example a subwoofer output and associated full-range satellite outputs. Placing a Hilbert in the satellites' audio path will synchronize their phase to that of the subwoofer channel.
Varying phase shifts of mechanical drivers will also affect the audible phase shifts. Also, the ear is not especially sensitive to nonlinear phase anyway.
Retrieving data ...