Larry Cochrane lcochrane@..............
Sent: Tue, 7 Feb 2012 6:12
Subject: Robert's circuit and new WinQuake / WinSDR releases
Hi Everyone,
In WinQuake the inverse filter can be found in the Time-Domain Filter=20
dialog box and in WinSDR the Channel Settings -> Integrate / Filter tab=20
dialog box.
A few days ago there were some emails on the list regarding what numbers=20
to use for the pendulum frequency and high-pass filter. The pendulum=20
frequency is the undamped period of your sensor. If you have a geophone=20
this would normally be 4.5 Hz (WinSDR, .222 in WinQuake) or maybe 1 if you=
=20
gave a 1 Hz geophone or AS1 type of sensor. This number sets where the filt=
er=20
starts to boost the lower frequencies to compensate for the drop off of=20
sensitivity below the period of the sensor. The high-pass filter sets how m=
uch=20
you want to extend the period of the sensor. It should be set to some numbe=
r=20
elow the period of the sensor. Depending on the sensor and electronics, mai=
nly=20
he first op-amp, you might be able to extend the period by one decade.
Before adding the software filter to WinSDR and WinQuake I played around wi=
th=20
he Robert's circuit. My results regarding using this filter, both in hardwa=
re=20
and software, can be found here: http://www.seismicnet.com/epf/=20
Hi Larry,
With regard to the 4.5 Hz comperiod compensation circuit, you have two =
series=20
high gain frequency compensating amplifiers, so you need at least a second =
order=20
high pass filter, or better, on the output as well as a 4th order LP filter=
to limit=20
HF signals into the ADC. You have re-designed an amplifier board!
The feedback resistor R3 on the first LT1007 opamp needs a parallel cap=
acitor to=20
reduce noise over ~10 Hz.=20
C10 and C33 at 22 muF look too large for a 4.5 Hz geophone and the freq=
uency=20
compensating amplifiers seem to have too high a DC gain (x50^2), which will=
likely=20
result in increased 1/f noise. You need an achieved gain of x10^2 =3D 100 a=
t 0.45 Hz.=20
I use x 20^2. Since the gain of the direct channel is 1 and the gain of the=
=20
amplified channel is also ~1 with a ~90 deg phase at 4.5 Hz, won't this giv=
e a peak=20
in the output voltage? I am a bit puzzled as to the functional requirement=
of C9 ?=20
You might want to make the feedback resistor on U2C either variable or swit=
ch selected=20
to vairy the overall gain ?=20
We have used a slightly different circuit.=20
Regards,=20
Chris Chapman
<=
/FONT>
=
Larry Cochrane lcochrane@webtro=
nics.com
Sent: Tue, 7 Feb 2012 6:12
Subject: Robert's circuit and new WinQuake / WinSDR releases
Hi Everyone,
In WinQuake the inverse filter can be found in the Time-Domain Filter
dialog box an=
d in WinSDR the Channel Settings -> Integrate / Filter tab <=
/PRE>dialog box.
A few days ago there were some emails on the list regarding what numbers
to use for=
the pendulum frequency and high-pass filter. The pendulum frequency is the undampe=
d period of your sensor. If you have a geophone
this would normally be 4.5 Hz (WinS=
DR, .222 in WinQuake) or maybe 1 if you
gave a 1 Hz geophone or AS1 type of sensor.=
This number sets where the filter
starts to boost the lower frequencies to compens=
ate for the drop off of
sensitivity below the period of the sensor. The high-pass f=
ilter sets how much
you want to extend the period of the sensor. It should be set t=
o some number=20
below the period of the sensor. Depending on the sensor and electronics, ma=
inly=20
the first op-amp, you might be able to extend the period by one decade.
Before adding the software filter to WinSDR and WinQuake I played around wi=
th=20
the Robert's circuit. My results regarding using this filter, both in hardw=
are
and =
software, can be found here: http://www.seismicnet.com/epf/
Hi Larry,
=
With regard to the 4.5 Hz comperiod compensation circuit, you have two ser=
ies
high gain fre=
quency compensating amplifiers, so you need at least a second order =
high pass filter, or bet=
ter, on the output as well as a 4th order LP filter to limit
<=
PRE style=3D"FONT-SIZE: 9pt">HF signal=
s into the ADC. You have re-designed an amplifier board!
The feedback resistor R3 on the =
first LT1007 opamp needs a parallel capacitor to
reduce noise over ~10 Hz.
C10 and C33 at 22 muF look too large for a 4.5 Hz geophone =
and the frequency
compensating amplifiers seem to have too high a DC g=
ain (x50^2), which will likely
result in increas=
ed 1/f noise. You need an achieved gain of x10^2 =3D 100 at 0.45 Hz.
I use x 20^2. Since the gain of the direct channel is =
1 and the gain of the
amplified channel is also ~1 with a ~90 deg phase at 4.5 Hz, won't this give a peak
in the output voltage? I am a bit puzzled as to the functional requirement of C9 ?
You might w=
ant to make the feedback resistor on U2C either variable or switch selected=
to vairy the ove=
rall gain ?
W=
e have used a slightly different circuit.
Regards,
Chris Chapman