Hi Mauro,

It's good to hear your thoughts on these creatures.

At 05:07 AM 8/18/2011, you wrote:

>Hi all,
>
>i am not considering myself a true expert.
>I am basicly an amataeur. I have made a lot of experiments (from 2008)
>with force balance systems for both BB sensor and accelerometers.
>I have built two BB sensors prototypes and several FB cells.
>What we found is that all is useful and all contribute a little in the
>instrument improvement. Amateurs's instruments at the same time can and
>cannot perform like a commercial (better to say professional) instrument.
>Can, because you may little by little experiment and improve one detail
>at a time, step by step.
>Cannot, because this require time and money an amateour often don't want
>to invest in a super super instrument.
>Cannot also because the best materials are not supplied (or not easy to
>find) to private people.

When we were looking at the STS-1 replacement, it=20
became obvious that we were having to address=20
exactly the same problems as they were, though=20
their solutions could, to some degree, cost more=20
and be more elaborate to construct.  Our=20
solutions had to be fairly cheap and easier to=20
build, but the problems were identical.

>What we found in our experiment is that the most important factor for
>noise cancelling is the thermal stability of the sensor installation.
>In the STS2 manual is written that you barely can go down to 30 seconds
>in a "non protected" installation (in terms of themperature).
>If you want to get the 120 seconds performance of the STS2 you have to
>follow exactly their guidelines in installation, insulation, cable
>deployent, basement design, etc.

For the typical instrument, having a velocity=20
response with two zeros at zero and two poles at=20
period TL seconds, and having a spring with=20
temperature coefficient (of modulus of=20
elasticity) of K ppm/degC; for slow changes, and=20
neglecting other temperature effects (which are=20
generally smaller), the output response to the=20
rate of temperature change is given by,

dVel / dTemp/dt =3D -6.90E-5 * K * TL^2 um/sec per DegC/hour
For the record, 6.90E-5 =3D g / (4 Pi^2 * 3600)

For our instrument with TL =3D 50 sec and assuming=20
K =3D -240ppm/degC we would expect to see=20
approximately 41.4 Um/s per degC/hour--quite a=20
large effect.  Note that this increases as the=20
square of TL and is one of the principal reasons=20
why we limit our low-end response to 50=20
seconds.  Because of that, our stainless steel=20
spring-temperature drift effect is actually less=20
than that of the new M2166, STS-1 replacement=20
which has TL of 360 seconds and uses an exotic=20
alloy spring.  To keep our instrument drift less=20
than one count, or 10 nm/s, would require that=20
the temperature change at the spring be held to=20
less than 4 microdegrees C per minute.

Lots of insulation and added thermal mass is good.

>We surprisingly found confirmed this (at the beginning we was a little
>skeptical on this) but we found that (for example) only few degrees of
>themperature variation on one BB leg (the adjustable leg over a BB
>sensor usually stands for levelling) can dramatically impact on noise
>performance.
>
>I found VERY (and i repeat VERY) interesting Randall's consideration on
>materials regarding the spring.
>It is true that a spring has the only function to keep the mass in the
>"rest" position. You can consider the spring like a mechanical reference
>(or bias?) for the electronic FB loop but this is not completely true.
>For example any themperature variation impacting on the springs will
>intruduce a noise in the bias.

Though it is still acting as a mechanical bias,=20
just not a perfect one.  The loop has to pick up the difference.

>In the case of the STS2 (or ours BB prototypes) the hinge and the
>springs contribute to keep the mass center of gravity at 54.7 degree
>respect to the fulcrum because they are triaxial homogeneous seismometer
>in this case both the picometric variations of the hinge flexibility and
>spring stiffness contribute to the noise generation.
>
>At 120 seconds a velocity of 1 nanometer per second correspond to a
>displacement of about 10 nanometers; the INVAR alloy, in majority used
>in BB technology, has a thermal coefficient of 1 ppm/=B0C
>In a bar of such material of 0.1 m wich represent the averaged dimension
>of one of the component on an amateur design this coefficient
>means a displacement of 100 nanometers per 1 (one) Celsius degrees.=F9
>And i am considering linear expansion, not volumetric expansion which is
>much more.
>So the contribution of the thermal noise is a factor greather 10 times
>the displacement you are going to measure at 120 seconds with an high
>performance alloy.
>
>What is wondering me at momemnt is what kind of mechanical noise can
>generate (due to internal frictions on materials) at higher frequency
>this 100 nanometers strain.
>Yes, strain because this 0.1 m bar is likely screwed and tighted with
>other components.
>This is one reason that lead BB manufacturers to bake in ovens all
>mechanical parts and not only springs during the manufacturing process.

Experience would suggest that the majority of the=20
pops occuring after assembly come from the=20
spring, not the joints.  If the instrument is=20
baked, then taken apart and reassembled there may=20
be relatively few pops.  However, if the spring=20
happened be bending in the opposite direction=20
from when it was baked, the pops will be frequent and severe.

>Another great noise contribution is the convective flows of air inside
>the instrument case. STS-2 is air-tight and there is vacuum (don't know
>how strong) inside in order to avoid air flows to deteriorate or disturb
>the mass position.

We always try to keep the temperature warmer at=20
the top, to minimize convection, though it isn't always so easy.

You may want to look at  some tests on Gary=20
Lindgren's coil-spring vertical that preceeded=20
his building the leaf spring design.
http://sites.google.com/site/seismicsensorinfo/checking-resonances

Regards,
Brett


__________________________________________________________

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.