Michael --

I think something you should try is bandpass filtering, which is a
combination of high-pass and low-pass filtering.  Make some recordings of
people walking past your sensors and do an FFT (using Winquake or the
like), and get an idea in what frequencies the majority of the energy
falls.  Then filter out everything else.  Also, geophones have a resonant
frequency.  You should be sure to properly dampen them with a resistor, or
undampen them if the resonant frequency falls in your bandwidth of
interest.  That way the geophone is acting like a bandpass filter already.

Be sure to record the waveforms of cattle too so you can figure out how to
ignore them.

I have an old US Border Patrol geophone that has a resonant frequency of
about 10Hz as I remember.  You've probably seen these molded with a spike.
Those should be exactly what you want since they were made for this purpose
initially.  They might be available surplus these days if the US Border
Patrol has given up using them.

Karl Cunningham
La Mesa, California


--On Wednesday, April 09, 2003 10:02 AM -0700 Michael King
 wrote:

> Here is the overall design.
> 
> The sensor consists of (2) geophones.  They are places aprox. 30-40 feet
> apart on a known trail that illegals use. 
> 
> The prototypes we are using have (2) LT1677 opamp circuits that amplify
> the geophone outputs and feed the signal into a basic stamp 1.  The level
> of signal isn't converted to digital, the signal when large enough (about
> 1.67 volts) will drive the input pin high and thus we have a logic switch.
> 
> In order to limit false hits the software in the basic stamp listens for
> a hit on a geophone.  When there is a hit it basically says if the second
> geopghone is hit within the next 2 seconds cancel the entire hit becuase
> it is most likely rain or some other seismic disturbance.  If the gephone
> is hit after that to seconds the it is a real hit and send a signal back
> to us through our repeater.  The data coming to us tells us what "port"
> hit first and what sensor it is.  By knowing what port was hit first we
> can tell what direction the contact is going.  
> 
> During initial powerup we callibrate the instrument by "walking the line"
> or walking by both geophones.  By doing this and counting the number of
> footfalls on an average man we can the have a rudimentary counting of the
> number of people walking by.  By dividing the number of detected footfall
> by the number of detected footfalls in the calibration we have the number
> of intruders.  It all works but the noisy circuit is killing me.
> 
> This system works but our circuit is noisy and when I make the software
> just send hits on either geophone as they come in, sometimes, at random,
> one or both of the "ports" will show hits for minutes on end.  If I touch
> the circuit board with my finger i can kill the noise until some kind of
> seismic activity start the noise again.
> 
> 
> 
> 
> Michael S. King
> Technical Director
> American Border Patrol


__________________________________________________________

Public Seismic Network Mailing List (PSN-L)