Randall,

Thank you for pointing me to that paper.  Theirs=20
is a very clever design.  Their measurements have=20
already gotten Dave and me thinking about some=20
measurements we would like to make on our verticals.

Let me try to summarize what I think I read.
It still suffers from the fundamental problem of all horizontals, tilt=
 noise.
"Of course, although the new FP [Folded Pendulum]=20
configuration presented here solves many=20
technical and sensitivity problems, the problem=20
that the FP configuration couples horizontal=20
forces and frame tilts is still an open problem,=20
only partly solved, for example, if they are=20
separate in band or if the FP is coupled with a=20
tiltmeter of comparable sensivity."
It's not clear how much of what they have done=20
can be applied to verticals, which has been our main interest.

As I understand it from the plots, the Optical=20
Lever portion of their position sensor has RMS=20
noise density at 50 mHz, of 8E-9 m/sqrt-Hz while=20
the interferometer's noise contribution at 50 mHz=20
is about 800x smaller, i.e. it is insignificant.

For the position sensor, 8E-9 m/sqrt-Hz at that=20
frequency implies velocity noise of 2.5E-9 m/s=20
per sqrt-Hz and acceleration noise of 7.9E-10=20
m/sec^2 per sqrt-Hz.  Squaring to get=20
acceleration noise "power" gives 6.23 E-19=20
m^2/sec^3 or -182 dB vs 1m^2/sec^3 at 0.05 Hz,=20
somewhere around what we might be getting with=20
our position sensor.  That's what we now want to go measure on ours.

They show the entire instrument recording ground=20
noise at 50 mHz of 1e-7 m/sqrt-Hz, implying an=20
acceleration noise power spectral density there=20
of  -160 dB vs 1m^2/sec^3.   What I don't see is=20
any attempt to separate true ground noise from=20
noise originating in the instrument.  To do that=20
they would need to do correlation measurements=20
with one or two reference instruments to separate=20
out the ground noise.  Maybe they assume that the=20
Folded Pendulum mechanism is noise free.  Clearly=20
Brownian noise is a non-issue, but as you have=20
pointed out, mechanisms, have many ways of=20
generating noise when they are so sensitive.

One other issue I don't understand is how they=20
plan to deconvolve data recorded by an instrument=20
having such an extremely high-Q response.  I=20
would think that normalizing data, having what=20
must be an enormous signal at 0.1 Hz (in the=20
configuration they used for the noise plots),=20
would be difficult to do with any accuracy.  I=20
probably don't understand.  Usually, if they=20
weren't using feedback, I would expect to see=20
them using a Q in the neighborhood of 0.7, unless=20
they are only concerned with motion at=20
frequencies far below 10 seconds period.  What am I missing?

Regards,
Brett

PS
For those who haven't run into the term,=20
deconvolving is the process of running the=20
recorded data through a digital filter which is=20
the inverse of an instrument's response.  This=20
makes the data approximately represent "true"=20
ground motion, unaffected by the=20
instrument.  When you want to do a detailed=20
comparison of data recorded by two different=20
instruments, to get useful results you should=20
first deconvolve the data from each of them.


At 09:24 AM 10/8/2012, you wrote:
>With a lot of money and a lot of good scientists=20
>working on a project with a lot of incentive to succeed=ADit is not=
 surprising
>that outstanding results can happen.
>      Take a look at the high-tech monolithic=20
> folded pendulum =91seismometer=92 that has been developed by
>an Italian team trying to be the first to see=20
>gravitational waves.   A recent paper (June=20
>2012) describing their instrument is online:
>
>=93Low Frequency - High Sensitivity Horizontal Inertial
>Sensor based on Folded Pendulum=94,
>Fausto Acernese, Rosario De Rosa, Gerardo Giordano, Rocco
>Romano, Silvia Vilasi, Fabrizio Barone
>
>http://iopscience.iop.org=
/1742-6596/363/1/012001
>
>Their figure 6 illustrates the outstanding=20
>broadband performance of the instrument, when using the interferometric=
 sensor.
>
>I was also pleased to see their Figure 4, which=20
>shows that the ultimate limiting performance of=20
>their instrument (operating in vacuum) will be determined by internal=
 friction.
>How can I conclude this?  Because the quality=20
>factor is proportional to the square of its=20
>(tunable) eigenfrequency.   My career has been devoted to studies of=
 nonlinear
>damping of mechanical oscillators, and this=20
>feature (quadratic dependence of Q on frequency)=20
>was first noted by Gunar Streckeisen in his=20
>studies of a LaCoste type spring vertical=20
>seismograph.  For a =91synopsis=92 providing some=20
>details, look at the following page
>http://www.iri=
s.edu/stations/seisWorkshop04/PDF/tahoeI1.pdf
>
>We have recently engaged here on the list-serve=20
>in spirited discussions of mechanical versus=20
>electronic considerations, relative to differing=20
>views on what limits the ultimate performance of=20
>a seismometer.  You will find in this paper an=20
>expressed opinion on the matter by these=20
>scientists, by the statement that reads:
>
>=93    the force feed-back configuration=20
>(accelerometer) was an obliged choice, with the consequent
>limitations in band and sensitivity due to the=20
>control electronic noise=94.  (talking about a=20
>previous folded pendulum configuration that used feedback).
>
>And then later, you find the following=20
>comment:     =93=85=85.(no force feed-back control) has the great advantage
>that no limitations to the band and sensitivity=20
>are introduced by the control electronics, so
>that the quality of the instrument depends=20
>mainly on a careful and optimized mechanical
>design=94.
>
>Randall
>
>
>
>
>
>
>


__________________________________________________________

Public Seismic Network Mailing List (PSNLIST)

To leave this list email PSNLIST-REQUEST@.............. with 
the body of the message (first line only): unsubscribe
See http://www.seismicnet.com/maillist.html for more information.