Displacement of frequency by a sound card.

[Shoo]

I have made the following construction (see pict.) and has met with strange effect.
I place earphone against the microphone.
You think f1 is equal f2?
NO!
I have combined these signals, and at sampling rate is lower 48000, I observe slow (30 sec) drift of a phase. (see schematic). The frequencies are near, but nevertheless are various.
How such could happen?

[Embedded]

There are so many things that will effect your circuit.

1) Driver Latency (Direct Sound can have a delay of around 0.5 seconds)
2) No speaker can truly represent a square wave, it always gets rounded and split into several frequencies(do an FFT).
3) Your mic will have a specific frequency response so may not even pick up all of the frequencies.
4) Both the speaker and the mic will also have some intrinsic phase delay.
5) Adding the signals together will just make a mess of the signals as they will be delayed and filtered by your design.

So it's no surprise you are getting odd signals.

What is it you are trying to do?

This works:

No drift clip.png (124.04 KiB) Viewed 10286 times

[Shoo]

No, in a square wave all harmonics are always multiple of a base frequency. Can not be arbitrary.
And unceasing drift of a phase it not delays, it is beatings of differing frequencies.
In your schematic the same drift is present, irrespective of entered delay.
(I think, is a hardware effect.)