PSN-L Email List Message
Subject: Re: My first Sensor
From: ChrisAtUpw@.......
Date: Thu, 21 Sep 2006 23:22:02 EDT
In a message dated 21/09/2006, tchannel@.............. writes:
2 Have you set up the damping to near 0.7 critical? No damp made yet, but
hope to use magnets I know the importance of this but was hoping to prove the
amp and filter first.
Hi Ted,
Without any damping the arm should swing freely for maybe 20 mins. You
time one complete oscillation to ge the period
I have put designs for NdFeB magnetic dampers and high output sensors at
_http://jclahr.com./science/psn/chapman/lehman/index.html_
(http://jclahr.com./science/psn/chapman/lehman/index.html) Both relay type coils and flat
rectangular high linearity coils are considered. The dampers use a horizontally
mounted plate. Copper is preferable, but soft Al will work OK.
What construction and suspension systems are you using?
It is vastly preferable to mount the whole apparatus on one base frame
which can then be adjusted and levelled. See the photo. You need only make the
arm about 2 ft long
3 2.5 Hz lowpass is on the low side.
The amp/filter is still on the breadboard so I can change the values, Do
you think I should use 10Hz instead of 2.5Hz?
I would be inclined to use at least 5 Hz. You can download a free filter
design program and application notes from
_http://focus.ti.com/docs/toolsw/folders/print/filterpro.html_
(http://focus.ti.com/docs/toolsw/folders/print/filterpro.html)
I usually use a 5 pole Bessel design, with the first pole on the
feedback resistor of the low noise input opamp. For a 10 Hz bandwidth, I set the
filter rolloff at 7 Hz.
You can also use a 10 Hz Butterworth filter, but they give about a 10%
overshoot and have extended delays at the roll off frequency.
Environmental noise, which you don't usually want to record, tends to
get serious above about 15 Hz, maybe less in an urban environment, so the
filter needs to reduce the amplitude of these signals quite drastically.
Sorry I don't know how many poles, It goes from a op amp "the filter" into
another op amp "the amp" with a gain of 100 I think.
What low noise input opamp are you using?
What circuit diagram? You need to build in a variable gain from maybe
200 up to several thousand.
I suggest that you check on the frequency and gain specification of
Larry's amplifiers at _http://psn.quake.net_ (http://psn.quake.net)
and the output voltage is min. 1.4v and max 7.5 volts. I adjust it to 2vs
into an AD Converter DATAQ 194 which requires 0 to 10v.
Have you modified your Dataq, or something? The standard DI-194 has
+/-10 V input, but it has only 10 bits resolution. Used in the way you describe,
it will only have a range of about 300 counts!! This is seriously inadequate.
Other experimenters PLEASE NOTE that Dataq are still selling off their
DI-154 +/-10V 12 bit ADCs for $39 at the moment. See _www.dataq.com_
(http://www.dataq.com) This is about the minimum resolution which is useful for
seismic recording, which requires a considerable dynamic range. They are supported
by Amaseis. 16 bit ADCs are preferable.
I am puzzled as to why your are doing this?
You need +/-12 V stabilised supply rails for your opamps!!
4 Sorry I don't understand this part. I am in Idaho hundreds of miles from
the ocean, could I still see these?
You get a microseism background from the ocean between 4 and 8 sec, so you
may want to filter above and below this. Can you take an unfiltered trace of
background and call up an FFT plot of amplitude versus frequency? The ocean
background peak should show up clearly. The period is a bit different for the
west and east coasts. In the middle of the USA, you may see two peaks.
The microseismic background can be observed anywhere on Earth and has an
amplitude of 0.5 to 15 microns, more typically about 1 to 2 microns. The
Pacific coast ones may have shorter periods of 4 to 6 seconds, whereas the
Atlantic ones are often 6 to 8 seconds. You will likely see both in Idaho.
Regards,
Chris Chapman
In a message dated 21/09/2006, tchannel@.............. writes:
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>
2 Have you set up the damping to near 0.7 critical? No dam=
p=20
made yet, but hope to use magnets I know the importance of this but=20=
was=20
hoping to prove the amp and filter first.
Hi Ted,
Without any damping the arm should swing freely=
for=20
maybe 20 mins. You time one complete oscillation to ge the period
I have put designs for NdFeB magnetic dampers a=
nd=20
high output sensors at
http://jcl=
ahr.com./science/psn/chapman/lehman/index.html Both=20
relay type coils and flat rectangular high linearity coils are considered. T=
he=20
dampers use a horizontally mounted plate. Copper is preferable, but soft Al=20=
will=20
work OK.
What construction and suspension systems are yo=
u=20
using?
It is vastly preferable to mount the who=
le=20
apparatus on one base frame which can then be adjusted and levelled. See the=
=20
photo. You need only make the arm about 2 ft long
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>
3 2.5 Hz lowpass is on the low side.
The amp/filter is still on the breadboard so I can change the=20
values, Do you think I should use 10Hz instead of=20
2.5Hz?
I usually use a 5 pole Bessel design, with the=20
first pole on the feedback resistor of the low noise input opamp. For a 10 H=
z=20
bandwidth, I set the filter rolloff at 7 Hz.
You can also use a 10 Hz Butterworth filter, bu=
t=20
they give about a 10% overshoot and have extended delays at the roll off=20
frequency.
Environmental noise, which you don't usually wa=
nt=20
to record, tends to get serious above about 15 Hz, maybe less in an urban=20
environment, so the filter needs to reduce the amplitude of these signals qu=
ite=20
drastically.
Sorry I don't know how many poles, It goes from a op amp "the=
=20
filter" into another op amp "the amp" with a gain of 100 I think.
What low noise input opamp are you using?
What circuit diagram? You need to build in a=20
variable gain from maybe 200 up to several thousand.
I suggest that you check on the frequency=20=
and=20
gain specification of Larry's amplifiers at
http://psn.quake.net
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>
and the output voltage is min. 1.4v and max 7.5 volts. I adjust=20=
it=20
to 2vs into an AD Converter DATAQ 194 which requires 0 to=20
10v.
Have you modified your Dataq, or something? The=
=20
standard DI-194 has +/-10 V input, but it has only 10 bits resolution. Used=20=
in=20
the way you describe, it will only have a range of about 300 counts!! This i=
s=20
seriously inadequate.
Other experimenters PLEASE NOTE that Dataq are=20
still selling off their
DI-154 +/-10V 12 bit ADCs for $39 at the=20
moment. See
www.dataq.com This is a=
bout=20
the
minimum resolution which is useful for seismic recording, which=20
requires a considerable dynamic range. They are supported by Amaseis. 1=
6=20
bit ADCs are preferable.
I am puzzled as to why your are doing this?
You need +/-12 V stabilised supply rails for=
=20
your opamps!!
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>
4 Sorry I don't understand this part. I am in Idaho hundre=
ds=20
of miles from the ocean, could I still see these?
You get a microseism background from the ocean between 4 and 8=20=
sec,=20
so you may want to filter above and below this. Can you take an unfiltered=
=20
trace of background and call up an FFT plot of amplitude versus frequency?=
The=20
ocean background peak should show up clearly. The period is a bit differen=
t=20
for the west and east coasts. In the middle of the USA, you may see two=20
peaks.
The microseismic background can be observed=20
anywhere on Earth and has an amplitude of 0.5 to 15 microns, more=20
typically about 1 to 2 microns. The Pacific coast ones may have shorter peri=
ods=20
of 4 to 6 seconds, whereas the Atlantic ones are often 6 to 8 seconds. You w=
ill=20
likely see both in Idaho.
Regards,
Chris Chapman
[ Top ]
[ Back ]
[ Home Page ]