Originally Posted by ab initio
These plots look like the soundcard has a chebyshev filter analog section prior to the ADC with a corner frequency somewhere before 20kHz. I think the ripples and frequency dive prior to 20kHz reflects the limitations of the soundcard hardware/implementation, and not the software, etc., as was being discussed earlier in the thread.
I think it is more likely to be a software problem. The frequency response shown on the graphs would be unacceptably poor performance for the Titanium HD; even my Realtek ALC887 onboard codec does not have ripples like that. On the other hand, Windows sound card drivers do tend to be buggy, and it is not always easy to avoid (low quality) sample rate conversion by the drivers or Windows. Another possibility is that the sample rate is not natively supported by the hardware (maybe the ADC is limited to 48 and 96 kHz), and resampling was needed for that reason.
A very steep analog lowpass filter is not needed with a decent delta-sigma ADC chip, such as the one used on the Titanium HD (which is a high quality TI model if I recall correctly), and the ADC chip itself is not likely to have such a poor digital filter either. Also, there is not much point measuring at 44.1 kHz when even onboard HDA codecs support 96 and 192 kHz.
In any case, for the purpose of amplifier frequency response measurements, the frequency response of the sound card can be measured separately, and the measured response of the amplifier can be compensated with an inverse filter.
Edited by stv014 - 6/18/13 at 1:35am