Brett,
Assuming my rough estimate for the calibration constant of 10^8 cts/ra=
d, your max count value of 1.3e6 corresponds to an equivalent pendulum angu=
lar deflection of 13 mrad. Assuming (only a rough) estimate for the dynam=
ic range of 70 dB (minimum measurable PG plate displacement of 60 nm), my g=
uess is that this corresponds also to something close to the largest angle =
that could be measured without clipping.
A minimum measurable displacement of 60 nm corresponds to a smallest a=
cceleration that could be measured of 6 e -7 m/s^2, independent of drive fr=
equency (assuming no electronics drift), for excitation frequencies below 0=
..5 Hz. This amounts to -120 dB per one-7th decade noise equivalent instrum=
ent power (flat below =BD Hz). Your feedback instrument is no doubt signif=
icantly better than this for frequencies down to somewhere near the telesei=
smic Rayleigh wave frequency of 1/20th Hz. On the other hand, it would be =
interesting to see how the two might compare in the range from 1/20th Hz do=
wn to eigenmode frequencies (less than 1 mHz).
Eric and I need a good helicord-type data processing/display software=
package before such comparisons with long-time records should become strai=
ghtforward. Larry has already made this happen with WinSDR/WinQuake for th=
e VolksMeter (which also uses the AD7745).
The thinking that my instrument and yours might compare more favorably =
at the lowest detectable frequencies relates to the following. As previou=
sly mentioned (posting about instrument sensitivity)-- the ultimate sensiti=
vity of every seismometer depends overwhelmingly on one thing-the reciproca=
l of the product of [ (i)the square of the characteristic frequency of the =
mechanical system and (ii) the smallest displacement of the inertial mass o=
f that system that can be measured]. One can only do so much with electron=
ics to overcome the influence of this quadratic dependence on period of the=
oscillator, independent of the electronics. And when the derivative of m=
ass displacement is performed (to produce a velocity detector for accelerat=
ions having frequency above the instrument's characteristic-typical of all =
the feedback instruments that I know about)-there is an unavoidable falloff=
in sensitivity as the frequency goes toward zero. Your instrument (if I u=
nderstand how it works) cannot avoid going toward zero as f goes toward zer=
o. But our tiltmeter response does not fall-off as f goes toward zero (as =
a property of its displacement sensor). The only limit to its so-called d.c=
.. response is the stability of the electronics.
What too few appreciate is also the following: When the period of an i=
nstrument gets very long, it is not only very sensitive to external acceler=
ation. It is also increasingly very, very sensitive to the internal struct=
ural changes that are an unavoidable characteristic of the mechanical compo=
nents, primarily the spring in a vertical seismometer. To change their thi=
nking, anybody who questions the physics of this statement needs only to lo=
ok at the herculean challenges of trying to detect gravitational waves. Th=
e researchers at LIGO have time and again bumped up against this curse-wor=
thy feature of the real world. It was a cause for me meeting some of the =
LIGO personnel at the 2004 IRIS broadband conference. They began to increa=
singly interact with the seismology world through the discovery that they w=
ere being impacted by similar material property challenges. Such limitatio=
ns become almost (maybe 'almost' is not applicable) insurmountable when tr=
ying to measure displacements at the level of 10^(-17) m.
Randall
Brett,
=A0=A0=A0=A0 Assuming my rough estimate for the c=
alibration constant of 10^8 cts/rad, your max count value of 1.3e6 correspo=
nds to an equivalent pendulum angular deflection of 13 mrad.=A0 =A0Assuming=
(only a rough) estimate for the dynamic range of 70 dB (minimum measurable=
PG plate displacement of 60 nm), my guess is that this corresponds also to=
something close to the largest angle that could be measured without clippi=
ng.
=A0=A0=A0=A0=A0A minimum measurable=
displacement of 60 nm corresponds to a smallest acceleration that could be=
measured of 6 e -7 m/s^2, independent of drive frequency (assuming no elec=
tronics drift), for excitation frequencies below 0.5 Hz.=A0 This amounts to=
-120 dB per one-7th decade noise equivalent instrument power (f=
lat below =BD Hz).=A0 Your feedback instrument is no doubt significantly be=
tter than this for frequencies down to somewhere near the teleseismic Rayle=
igh wave frequency of 1/20th Hz.=A0 On the other hand, it would =
be interesting to see how the two might compare in the range from 1/20=
th Hz down to eigenmode frequencies (less than 1 mHz).=A0
=A0=A0=A0=A0=A0=A0Eric and I need a good helicord-=
type data processing/display software package before such comparisons with =
long-time records should become straightforward.=A0 Larry has already made =
this happen with WinSDR/WinQuake for the VolksMeter (which also uses the AD=
7745).=A0
=A0=A0=A0=A0The thinking that=
my instrument and yours might compare more favorably at the lowest detecta=
ble frequencies relates to the following. =A0=A0As previously mentioned (po=
sting about instrument sensitivity)-- the ultimate sensitivity of every sei=
smometer depends overwhelmingly on one thing—the reciprocal of the pr=
oduct of [ (i)the square of the characteristic frequency of the mechanical =
system and (ii) the smallest displacement of the inertial mass of that syst=
em that can be measured].=A0 One can only do so much with electronics to ov=
ercome the influence of this quadratic dependence on period of the oscillat=
or, independent of the electronics.=A0 =A0And when the derivative of mass d=
isplacement is performed (to produce a velocity detector for accelerations =
having frequency above the instrument’s characteristic—typical =
of all the feedback instruments that I know about)—there is an unavoi=
dable falloff in sensitivity as the frequency goes toward zero.=A0 Your ins=
trument (if I understand how it works) cannot avoid going toward zero as f =
goes toward zero.=A0 But our tiltmeter response does not fall-off as f goes=
toward zero (as a property of its displacement sensor). The only limit to =
its so-called d.c. response is the stability of the electronics.
=A0=A0=A0=A0What too few appreciate is also the f=
ollowing:=A0 When the period of an instrument gets very long, it is not onl=
y very sensitive to external acceleration.=A0 It is also increasingly very,=
very sensitive to the internal structural changes that are an unavoidable =
characteristic of the mechanical components, primarily the spring in a vert=
ical seismometer.=A0 To change their thinking, anybody who questions the ph=
ysics of this statement needs only to look at the herculean challenges of t=
rying to detect gravitational waves.=A0 The researchers at LIGO =A0have tim=
e and again bumped up against this curse-worthy feature of the real world. =
=A0=A0It was a cause for me meeting some of the LIGO personnel at the 2004 =
IRIS broadband conference.=A0 They began to increasingly interact with the =
seismology world through the discovery that they were being impacted by sim=
ilar material property challenges.=A0 Such limitations become almost (maybe=
‘almost’ is =A0not applicable) insurmountable when trying to m=
easure displacements at the level of 10^(-17) m.=A0
=A0=A0 Randall<=
/o:p>
=