From: Winfield Hill
Subject: Re: Detecting original ultrasonic signal vs. reflected?
Date: 24 Sep 2002 02:42:25 -0700
Organization: Rowland Institute
X-Newsreader: Direct Read News 2.97
John Rahn wrote...
> Is it at all possible, to tell the difference, of a reflected
> ultrasonic signal to an original signal.
> I am starting a project that sends a signal from a transmitter to
> separate receivers and want to try to figure out how to detect/filter
> out reflected signals, because I will be timing the received signals,
> and a reflected signal would cause an inaccurate reading.
Generally reflected signals arrive after direct signals, because
the path is longer. :-)
But you asked, and yes, in the case of very short pulses in low-Q
wide-bandwidth high-frequency systems, they do have very different
signatures. Direct signals can be short and sweet, wellformed and
ideal. By contrast reflected signals are generally stretched by
the geometry and expanse of the reflecting surfaces. Furthermore,
scattering, refraction and attenuation contribute to an overall
deterioration of the stretched and delayed portions of a reflected
waveform, creating regions of reinforcement and cancellation, both
the signal phase and amplitude vary wildly throughout the pulse.
Use wide-bandwidth drivers and receivers, sample your signal 5 or
10x per cycle. Cross-correlate the digitized signal with a stored
version of the direct signal to reveal the dramatically different
signatures for the two cases. The cross-correlation peak is also
a very good way to determine an actual direct-path time delay.
Simple examination for the presence of any reverberation energy
after the beginning of the pulse and after it should have ended is
a simplified method, but examining the entire cross-correlation
waveform is an excellent way to evaluate the reverberation energy,
thereby improving overall recognition robustness.