The HF filter changes the original data within the audio bandwidth, so it will certainly 'fail' on a null test. But that doesn't mean it fails perceptively. To give an example - I tested the (note not the same filter as Dave) IIR bandwidth limiting filter with one of my decimation filters for the ADC, which eliminates everything above 20k, but is completely flat with zero phase shift up to 20k. With the decimation filter in place, you could hear the effects of the IIR filter easily, as usual. But with the decimation filter working, you could hear no difference at all. This proved that the phase and amplitude shifts of the IIR filter (which are significant in the audio bandwidth) are not perceptively important. This proved that a -1dB or so at 20k with attendant phase shifts are inaudible (at least to my old ears) and that the SQ changes of bandwidth limiting filters are due to reducing out of band noise above 50k and hence reducing noise floor modulation - which is at unmeasurable levels either way.
