PSN-L Email List Message
Subject: Re: Designing a new vertical sensor
From: ChrisAtUpw@.......
Date: Sat, 4 Apr 2009 22:06:24 EDT
In a message dated 04/04/2009, gmvoeth@........... writes:
I believe that increasing the mass will increase the signal to noise ratio
but not sure.
I used an old 500Lb. seismometer once
Hi Geoff,
You can get away fine with 500gm to 1kg mass. Anything more just adds
cost and construction problems. You only needed massive weights to drive a
mechanical amplifier + a pen on chart paper. This went out with photographic and
later with direct electronic recording.
I believe that using a balanced sensor will make the electronics easier to
build by providing a proper input for a op amp diff amp.
You should NOT be limited by electronic noise. It is background seismic
noise which is the problem. A single ended stage is quite satisfactory. If
you are troubled by electronic noise, redesign your amplifier, or copy a known
good circuit.
You can wind a bifilar coil, but differential input opamps do not have
the very low noise that you can get from single opamps. You can also buy
bifilar insulated equipment wire. The INA118 is one of the best of the
differential amplifier types.
A balanced sensor is two coils/magnets wired in additive fashion with the
center point grounded
making a single three wire sensor. But both must be matched as close as
possible to being
identical. Both sensors can use the same mass or be two separate single
devices on the same
foundation (platform).
Do it if it makes you happy, but it is not necessary
Cable should be 100% shielded and possibly the preamp located at the sensor
itself with maybe 50Ft of cable at most.
Short, buried cable runs are to be preferred. Use cable with a woven
braid shield. You also need to consider lightning protection for long cable runs.
The shield should probably be cut just before reaching the sensor but not
sure. If everything is insulated from EARTH you might surround even the sensor
in the electrical ground. Grounding is important to eliminate static and RFI
noise as well as the 60hz power hum that might originate
from power lines.
Earth the sensor frame through the coaxial shield braid. Earth the cable
at the amplifier input.
Use magnetic damping separate from the sensor and you will get better output
from the sensor.
This will lower the noise.
IF you could place the mass in a vacuum there would be no buoyancy effects
for changes in air density on the mass.
You can either place the sensor in a sealed container or compensate the
arm with a sealed balance wolume. A vacuum is not essential, but it can
eliminate any convection noise.
The greater the A/D Resolution the lower the overall gain needs to be in the
electronics.
Preferably use an ADC with 16 bit +/-1/2 LSB resolution. Some of the
Sigma-Delta converters give this, but with old fast ADCs you may need to use
signal averaging to remove the converter noise, sometines three bits.
Filter Low Pass the output so that the Aliasing Freq occurs at 0Db or less
of electrical gain if possible.
Set the loss at the Aliasing frequency to be at least 20 dB, preferably
30 to 40 dB
Match all your components as closely as possible in the preamp to avoid
baseline drifting and common mode problems (same signal on the two main signal
leads).
Use 1% Metal Film resistors only, never carbon. You may want to buy 5%
capacitors and then pick out near matching pairs for filters. Use polyester
coupling capacitors, NOT Alumium or Tantalum Electrolytics.
Double regulate the power supply with at least three + volts between the
output and the source.
So if you got 5 volts out you need at very least 8 volts in.
Using 15V and then 12V regulators in series will reduce noise and drift.
Anything at all that varies could affect the input
into your A/D converter. And all you want is pure signal.
The best of all would possibly be a three axis signal that is mathematically
combined to create a single magnitude. But then you are not concerning
yourself with the three dimensions but only phases and first time of arrival which
is what interests myself the most.
You usually want to know the amplitudes and phases of the three
components separately. This gives you the directional information.
Some people believe that the sensor magnet is fixed to the physical ground
is the best way to go but not sure here since I have had excellent results
with short period devices the other way around.
DO NOT put magnets on the arm, unless they are totally shielded and even
this is inadvisable. DO NOT use ferromagnetic components on the arm unless
absloutely necessary. A Brass mass is preferable to Lead. It is easier to
machine, physically stronger and you can drill and tap holes in it. Lead is so
soft that it can relax under a clamp, but you can soft solder it to brass
fixture.
If you could string 100 devices together 50 on each side of Signal ground
And locate them on the same platform just think of the sensitivity that might
produce. But who has that kind of monies
This would be an array. They are principally used in direction finding,
CTBT Arays, Volcanoes and for reducing local seismic noise.
I think you are not interested in the electronics but not sure about that.
Need the tools. Expensive.
(designing/Making Your Own or building premade kits) But you can do exactly
what you want this way. I do not recommend this unless you have great amount
time on your hands.
Either buy an amplifier from Larry, build a similar type from John's
website, or get interested in electronics. Some systems like the above require a
period extending amplifier.
Take a look at Dewayne Hill's vertical at
_http://jclahr.com/science/psn/hill/index.html_ (http://jclahr.com/science/psn/hill/index.html)
Then add quad magnetic block + plate damping. DON'T bother to try out
oil damping!
Remove the knife edge suspension and substitute two flex wires in
tension. This reduces noise.
Clamp both ends of the suspension spring. This reduces noise.
Add a quad magnet block + rectangular sensor coil to give a greatly
increased signal output, probably > 20x.
Fit a 1/f^2 to linear amplifier and you can sense quakes from 5 Hz to 20
seconds very nicely.
This avoids the problems associated with the high thermal coefficient of
steel springs / providing Ni-SpanC springs.
Aim for a mass of about 1 lb and cut the spring length to suit.
Regards,
Chris Chapman
In a message dated 04/04/2009, gmvoeth@........... writes:
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>I=20
believe that increasing the mass will increase the signal to noise ratio b=
ut=20
not sure.
I used an old 500Lb. seismometer once
Hi Geoff,
You can get away fine with 500gm to 1kg mass.=20
Anything more just adds cost and construction problems. You only needed=
=20
massive weights to drive a mechanical amplifier + a pen on chart paper. This=
=20
went out with photographic and later with direct electronic recording.
I believe that using a balanced sensor will make the electronics ea=
sier=20
to build by providing a proper input for a op amp diff amp.
You should NOT be limited by electronic noise.=20=
It=20
is background seismic noise which is the problem. A single ended stage is qu=
ite=20
satisfactory. If you are troubled by electronic noise, redesign your amplifi=
er,=20
or copy a known good circuit.
You can wind a bifilar coil, but differential i=
nput=20
opamps do not have the very low noise that you can get from single opamps. Y=
ou=20
can also buy bifilar insulated equipment wire. The INA118 is one of the=
=20
best of the differential amplifier types.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>A=20
balanced sensor is two coils/magnets wired in additive fashion with the ce=
nter=20
point grounded
making a single three wire sensor. But both must be=
=20
matched as close as possible to being
identical. Both sensors can use t=
he=20
same mass or be two separate single devices on the same
foundation=20
(platform).
Do it if it makes you happy, but it is not=20
necessary
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Cable=20
should be 100% shielded and possibly the preamp located at the sensor itse=
lf=20
with maybe 50Ft of cable at most.
Short, buried cable runs are to be preferred. U=
se=20
cable with a woven braid shield. You also need to consider lightning=20
protection for long cable runs.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>The=20
shield should probably be cut just before reaching the sensor but not sure=
.. If=20
everything is insulated from EARTH you might surround even the sensor in t=
he=20
electrical ground. Grounding is important to eliminate static and RFI nois=
e as=20
well as the 60hz power hum that might originate
from power=20
lines.
Earth the sensor frame through the coaxial shie=
ld=20
braid. Earth the cable at the amplifier input.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Use=20
magnetic damping separate from the sensor and you will get better output f=
rom=20
the sensor.
This will lower the noise.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>IF you=20
could place the mass in a vacuum there would be no buoyancy effects for=20
changes in air density on the mass.
You can either place the sensor in a sealed=20
container or compensate the arm with a sealed balance wolume. A vacuum is no=
t=20
essential, but it can eliminate any convection noise.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>The=20
greater the A/D Resolution the lower the overall gain needs to be in the=20
electronics.
Preferably use an ADC with 16 bit +/-1/2 LSB=20
resolution. Some of the Sigma-Delta converters give this, but with old=20=
fast=20
ADCs you may need to use signal averaging to remove the converter noise,=20
sometines three bits.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Filter=20
Low Pass the output so that the Aliasing Freq occurs at 0Db or less of=20
electrical gain if possible.
Set the loss at the Aliasing frequency to be at=
=20
least 20 dB, preferably 30 to 40 dB
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Match=20
all your components as closely as possible in the preamp to avoid baseline=
=20
drifting and common mode problems (same signal on the two main signal=20
leads).
Use 1% Metal Film resistors only, never carbon.=
You=20
may want to buy 5% capacitors and then pick out near matching pairs for filt=
ers.=20
Use polyester coupling capacitors, NOT Alumium or Tantalum=20
Electrolytics.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Double=20
regulate the power supply with at least three + volts between the output a=
nd=20
the source.
So if you got 5 volts out you need at very least 8 volts=20
in.
Using 15V and then 12V regulators in series wil=
l=20
reduce noise and drift.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Anything=20
at all that varies could affect the input
into your A/D converter. And=20=
all=20
you want is pure signal.
The best of all would possibly be a three=20=
axis=20
signal that is mathematically combined to create a single magnitude. But t=
hen=20
you are not concerning yourself with the three dimensions but only phases=20=
and=20
first time of arrival which is what interests myself the=20
most.
You usually want to know the amplitudes and pha=
ses=20
of the three components separately. This gives you the directional=20
information.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Some=20
people believe that the sensor magnet is fixed to the physical ground is t=
he=20
best way to go but not sure here since I have had excellent results with s=
hort=20
period devices the other way around.
DO NOT put magnets on the arm, unless they are=20
totally shielded and even this is inadvisable. DO NOT use ferromagnetic=20
components on the arm unless absloutely necessary. A Brass mass is preferabl=
e to=20
Lead. It is easier to machine, physically stronger and you can drill an=
d=20
tap holes in it. Lead is so soft that it can relax under a clamp, but you ca=
n=20
soft solder it to brass fixture.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>If you=20
could string 100 devices together 50 on each side of Signal ground And loc=
ate=20
them on the same platform just think of the sensitivity that might produce=
..=20
But who has that kind of monies
This would be an array. They are principally us=
ed=20
in direction finding, CTBT Arays, Volcanoes and for reducing local seis=
mic=20
noise.
<=
FONT=20
style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>I think=20
you are not interested in the electronics but not sure about that. Need th=
e=20
tools. Expensive.
(designing/Making Your Own or building premade kits)=20=
But=20
you can do exactly what you want this way. I do not recommend this unless=20=
you=20
have great amount time on your hands.
Either buy an amplifier from Larry, build a sim=
ilar=20
type from John's website, or get interested in electronics. Some systems lik=
e=20
the above require a period extending amplifier.
=
DIV>
Then add quad magnetic block + plate damping. D=
ON'T=20
bother to try out oil damping!
Remove the knife edge suspension and substitute=
two=20
flex wires in tension. This reduces noise.
Clamp both ends of the suspension spring. This=20
reduces noise.
Add a quad magnet block + rectangular sensor co=
il=20
to give a greatly increased signal output, probably > 20x.
Fit a 1/f^2 to linear amplifier and you=20
can sense quakes from 5 Hz to 20 seconds very nicely.
This avoids the problems associated with the hi=
gh=20
thermal coefficient of steel springs / providing Ni-SpanC springs.
Aim for a mass of about 1 lb and cut the spring=
=20
length to suit.
Regards,
Chris=20
Chapman
[ Top ]
[ Back ]
[ Home Page ]