Hi Chris,
The seismometer is not a Lehman but a single pendulum with a natural
period of 1 second. It is supported by two pairs of leaf suspensions arranged
in a gimble form and give the pendulum two degrees of freedom but no rotation
about it's axis. The LED and projection optics are mounted on the lower end
and projects a 4mm spot of light modulated at 10KHz onto the quadrant photo
cell. The quadrant detector produces the X & Y components of motion which in
turn, via phase-sensitive detectors and appropriate feedback to orthogonal
force transducers, hold the pendulum in a fixed position. A two axis broadband in
fact but with a period of 60 seconds. The two axis output signals are of
course the restraining force in volts/meter/second. Unlike the vertical
broadband, temperature correction is unnecessary and tilt errors are compensated by
giving the integrator a large dynamic range.
The capacitance does not directly generate noise but modifies the source
impedance which in turn change the ratio of the noise voltage to noise current
contributions to the total noise. The photo current of 0.7 micro amps is large
compared with the FET gate leakage current and is the dominate noise source.
Narrower bandwidths of lock-in systems will improve the S/N ratio but may
compromise the transient response, it's a question of optimisation.
The important characteristics of the LED are uniformity of the light spot
and linearity. The LED control feedback is dynamic i.e. both amplitude and wave
form are controlled, but there is a limit to how much the loop can correct
non-linearity before the onset of instability.
The theoretical analysis of Force Balance Feedback is covered in Thomas
Forbriger and Erhard Wielandt papers, the former is in German but the more
practical aspects are well documented by Sean-Thomas Morrissey and Allen Coleman.
I should perhaps point out that large quadrant silicon photo detectors are
a little expensive to say the least and I would not have use this approach if
I had not been given a handful.
Regards Martin
Hi Chris,
The seismometer is not a Lehman but a single=20
pendulum with a natural period of 1 second. It is supported by two pairs of=20=
leaf=20
suspensions arranged in a gimble form and give the pendulum two degrees of=20
freedom but no rotation about it's axis. The LED and projection optics=20=
are=20
mounted on the lower end and projects a 4mm spot of light modulated at 10KHz=
=20
onto the quadrant photo cell. The quadrant detector produces the X & Y=20
components of motion which in turn, via phase-sensitive detectors and=20
appropriate feedback to orthogonal force transducers, hold the pendulum in a=
=20
fixed position. A two axis broadband in fact but with a period of 60=20
seconds. The two axis output signals are of course the restraining forc=
e in=20
volts/meter/second. Unlike the vertical broadband, temperature correction is=
=20
unnecessary and tilt errors are compensated by giving the integrator a large=
=20
dynamic range.
The capacitance does not directly generate noise but modifies the=20
source impedance which in turn change the ratio of the noise voltage to nois=
e=20
current contributions to the total noise. The photo current of 0.7 micr=
o=20
amps is large compared with the FET gate leakage current and is the dominate=
=20
noise source.
Narrower bandwidths of lock-in systems will improve the S/N ratio but m=
ay=20
compromise the transient response, it's a question of optimisation.
The important characteristics of the LED are uniformity of the light sp=
ot=20
and linearity. The LED control feedback is dynamic i.e. both amplitude and w=
ave=20
form are controlled, but there is a limit to how much the loop can=20
correct non-linearity before the onset of instability.
The theoretical analysis of Force Balance Feedback is covered in Thomas=
=20
Forbriger and Erhard Wielandt papers, the former is in German but the more=20
practical aspects are well documented by Sean-Thomas Morrissey and Allen=20
Coleman.
I should perhaps point out that large quadrant silicon photo=20
detectors are a little expensive to say the least and I would not have use t=
his=20
approach if I had not been given a handful.
Regards Martin