It could work but you take on a big bunch of presumptions about the two channels of the sound card inputs which include:

a) Are the gains of the two channels the same within the desired accuracy specifications?

b) Are the filters in the two channels exact enough that they attenuate the same on both channels?

c) Is the signal phase delay through both channels the same and constant with frequency?

d) Is the signal envelope delay through the two channels the same?

I suspect that any such system, for it to be workable, would have to include a set of calibration feedback from the software that handles the sound card outputs back to the front end circuitry so that it can adjust the gain, phase, and references on each channel so that the result appearing inside can undergo the mathematical manipulations that you suggest.

I also suggest that is is not suitable to just add some AC to the equivalent DC level of the input signal. Such DC level simply gets immediately lost (canceled out) by the AC coupling of the input at the sound card. You instead have to make the amplitude of the AC waveform correspond in some way to the equivalent DC level of the input signal. Another way would be to encode the equivalent DC level as a phase difference of the AC signal fed into one channel versus that fed into the other channel. In either case the factors I mentioned above are of huge concern of making anything close to reasonably accurate.

In the end the cost of circuitry to make something satisfactory quickly outstrips the cost of a simple USB A/D converter of USB oscilloscope product. I can say that I can see enough problems with the sound card concept that I would not be willing to spend even one hour trying to experiment with making some prototype to try out.

Michael Karas