🚀 go-pugleaf

In article <1992Apr14.094205@andrea-doria.ifr.ethz.ch> vestli@andrea-doria.ifr.ethz.ch (Sjur Jonas Vestli) writes:
>|> 
>|> Using v=330 m/s, even just 1 MHz timing resolution will give you the
>location
>|> accurate to 0.3 mm.  
>|> 
>
>The accuracy of such a measuring system will also be limited by the
>wavelength of the ultrasound (US). Let us assume that the
>we need one full wave for reliable detection and that we are operating
>at 250 kHz. Tperiod=4 us, v=330 M/s, Lambda
>(wavelenght)=Tperiod*v=0.00132 m (1 mm).
>You might be able to detect within less than one wavelenght,
>but usually in US there is plenty of noise around and the detection
>circuitry introduces errors of this kind.
>
> Sjur

And why do you say that measurement resolution is limited by wavelength
and noise?  We're talking delay estimation here, not optics.  Radar
people know how to design waveforms with good correlation properties.
I'm not a radar person, but I don't see why you can't get arbitrarily
good resolution using chirps.

As for noise, just average your measurements.  If this 3D tracker is
following hand motions, 10-20 measurements/sec should be sufficient.
Use a rep rate of 1000-2000 pulses/sec, and average 100 pulses for each
reading.
-- 
   Dave Kemp   dpkemp@afterlife.ncsc.mil              "Pave the Bay!"
   #include <std/disclaimer>