Thanks so much, Attila, for your attention to this. Rather than bury readers in text, I'll just answer your questions and use your clever text graphing. But if you or anybody is interested, I could attach a few relevant pages from a US Navy document about taking free-field measurements in a laboratory tank.
The generated signal need only be five or ten milliseconds. Averaging will be useful, but we could always average multiple tests if needed. We could keep it simple for now and say one pulse. The pulse should be a sine wave.
I'll put realistic example times in milliseconds in parentheses as examples below. We should be able to set those times based on test conditions.
|freq1............................................................................................................................................................................................|freq2...
|sine pulse 1 (10ms duration) | wait (1ms) | acquisition (3ms) | log RMS voltage (two channels) | wait for sound decay (250ms) |sine pulse 2...
Again, we're trying to create a sound (pulsed constant wave), wait for that sound to arrive at the DUT and reference (same spacing), measure that direct sound before reflections arrive, then wait for that sound to decay in amplitude while logging the measurement and then incrementing the the test frequency. The acquisition could be very short, or we could trigger an acquisition on wavegen and only look at the appropriate few milliseconds or wave cycles by gating a portion of that acquisition.
phase information would be nice, but not at all essential.