PSN-L Email List Message

Subject: Re: Microseism filter
From: ChrisAtUpw@.......
Date: Thu, 24 Feb 2005 22:13:44 EST


 
In a message dated 24/02/2005, dyouden@............. writes:

Tell me  more about your filter. I live on the East coast, and am 
troubled by the  level of microseisms. I have messed a bit with switched 
capacitor circuits  in the past, but for different purposes. Could you 
send a circuit diagram  directly to me? Or point me to the chip you used 
so I can puzzle out my  own filter.

Dave...

Dave Nelson wrote:
>  >   I have developed a simple filter based on switched capacitor 
>  technology that virtualy eliminates the microseism background. The 
>  filter has no critical componants or  capacitors and is tuned by a  
> clock frequency. The Q is 0.5 which gives roughly an octave bandwidth  
> centered at 0.015 hz or 6.6 seconds . Since the center frequncy is  
> clock dependant it can be shifted arbitarily but 6.6 seconds seems to  
> work very well. The filter is placed directly ahead of the A/D at the  
> highest voltage swing  level in the system to avoid adding  noise.  
> Just two wires -- in and out (plus power of  course).
>  
> The results are very good, the harmonic like  signal from microseism is 
> gone and the effective signal to noise  ratio of eathquakes is 
> significantly better. The spectrum has a deep  null at 6.6 seconds 
> instead of a  dominant peak .
>   
> I know this kind of spectrum shaping can be easily done in software  
> but that is usually beyond the scope of an amateur or computer dum-  
> dum like me. The whole thing is  two IC's and a 15 hz clock  generator. 
> ( 100 times the center frequency)
>  
> I  would be glad to correspond with anyone interested in or commenting 
>  on what I am doing.
> Dave Nelson 



Hi Dave,
 
    An alternative approach is to use a twin Tee  bandstop filter. You need 
to sharpen the frequency response up quite a bit over  a simple twin Tee 
circuit, which has a Q of about 0.2 ! You can make the Q  variable by feeding the 
output into a unity gain opamp connected to a  potentiometer to earth. You use a 
second unity gain opamp connected to the pot  slider to provide the normal 
'earth' connection of the twin Tee. Since you have  to vary three resistors to 
change the frequency, it is easiest to use  switched resistors. An alternative 
is to use potentiometers and set up the  values with an Ohmmeter to suit your 
measured dominant period. You get a  maximum reduction of a bit over 50 dB. 
They can also be useful if there is  a local constant frequency vibration source.
    See also filter books for 'band stop' filter  circuits and Linear 
LTC1043. Switched filters usually need a good low pass  analogue filter on the output 
to reduce chopper noise and this can also  significantly limit your low 
signal range. I prefer to use analogue filters  - very low noise amplifiers do not 
mix too easily with digital  circuits.  
    You can expect to see different  dominant periods depending on the 
nearest ocean or sea and to some extent  on the changing weather conditions. The 
microseism amplitudes are strongly  dependant on the ocean weather. 
    It is relatively easy to filter out microseisms  digitally with analysis 
programmes such as WinQuake. If you live right on the  coast, filtering can be 
desirable, but otherwise they can provide a monitor for  the proper operation 
of the sensors and may be interesting in themselves.
 
    Roberts published a circuit for extending the  response of geophones to 
1/10 the natural frequency.
 
    Regards,
 
    Chris Chapman





In a message dated 24/02/2005, dyouden@............. writes:
<= FONT=20 style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size= =3D2>Tell me=20 more about your filter. I live on the East coast, and am
troubled by t= he=20 level of microseisms. I have messed a bit with switched
capacitor circ= uits=20 in the past, but for different purposes. Could you
send a circuit diag= ram=20 directly to me? Or point me to the chip you used
so I can puzzle out m= y=20 own filter.

Dave...

Dave Nelson wrote:
>  >&nb= sp;=20 I have developed a simple filter based on switched capacitor
>=20 technology that virtualy eliminates the microseism background. The
>= ;=20 filter has no critical componants or  capacitors and is tuned by a=20
> clock frequency. The Q is 0.5 which gives roughly an octave bandw= idth=20
> centered at 0.015 hz or 6.6 seconds . Since the center frequncy i= s=20
> clock dependant it can be shifted arbitarily but 6.6 seconds seem= s to=20
> work very well. The filter is placed directly ahead of the A/D at= the=20
> highest voltage swing  level in the system to avoid adding=20 noise. 
> Just two wires -- in and out (plus power of=20 course).

> The results are very good, the harmonic li= ke=20 signal from microseism is
> gone and the effective signal to noise=20 ratio of eathquakes is
> significantly better. The spectrum has a d= eep=20 null at 6.6 seconds
> instead of a  dominant peak .
>&nb= sp;=20
> I know this kind of spectrum shaping can be easily done in softwa= re=20
> but that is usually beyond the scope of an amateur or computer du= m-=20
> dum like me. The whole thing is  two IC's and a 15 hz clock=20 generator.
> ( 100 times the center frequency)

&g= t; I=20 would be glad to correspond with anyone interested in or commenting
&g= t;=20 on what I am doing.
> Dave Nelson 
Hi Dave,
 
    An alternative approach is to use a twin Tee=20 bandstop filter. You need to sharpen the frequency response up quite a bit o= ver=20 a simple twin Tee circuit, which has a Q of about 0.2 ! You can make the Q=20 variable by feeding the output into a unity gain opamp connected to a=20 potentiometer to earth. You use a second unity gain opamp connected to the p= ot=20 slider to provide the normal 'earth' connection of the twin Tee. Since you h= ave=20 to vary three resistors to change the frequency, it is easiest to use=20 switched resistors. An alternative is to use potentiometers and set up the=20 values with an Ohmmeter to suit your measured dominant period. You get=20= a=20 maximum reduction of a bit over 50 dB. They can also be useful if there= is=20 a local constant frequency vibration source.
    See also filter books for 'band stop' filter=20 circuits and Linear LTC1043. Switched filters usually need a good low pass=20 analogue filter on the output to reduce chopper noise and this can also=20 significantly limit your low signal range. I prefer to use analogue fil= ters=20 - very low noise amplifiers do not mix too easily with digital=20 circuits.  
    You can expect to see different=20 dominant periods depending on the nearest ocean or sea and to some exte= nt=20 on the changing weather conditions. The microseism amplitudes are strongly=20 dependant on the ocean weather. 
    It is relatively easy to filter out microseisms= =20 digitally with analysis programmes such as WinQuake. If you live right on th= e=20 coast, filtering can be desirable, but otherwise they can provide a monitor=20= for=20 the proper operation of the sensors and may be interesting in themselves.
 
    Roberts published a circuit for extending the=20 response of geophones to 1/10 the natural frequency.
 
    Regards,
 
    Chris Chapman

[ Top ] [ Back ] [ Home Page ]