In a message dated 05/04/03, dyouden@......... writes:

> I have been thinking about a seismometer that has no natural frequency. 
> 

Hi Dave,

> Imagine, if you will, a stack of, say 1" diameter aluminum tubes about 10 
> inches long. There are 9 tubes, two on the bottom, three in the next layer 
> 

       I would rather not thank you! Why make it extremely complicated, when 
you can make it dead simple?

       Take a circular piece of plate glass and drill a hole in the middle 
for the air supply. Support the outside edges on a circular metal structure 
with three levelling screws. Use another, but smaller diameter sheet of plate 
glass or of Al etc. more or less centrally placed. This will need to have the 
bottom surface modified, possibly by sticking on foil or by etching, to give 
either a circular or a clover leaf shaped air bearing. Fit the top of the 
plate with two pairs of Maxwell coil + magnet force actuators, at right 
angles and a shadow type capacitative distance transducer and you might be 
'in business'! See http://physics.mercer.edu/petepag/tutorial.html 
       You might need to lap polish the two glass plates together with fine 
carborundum and rouge, rather like you do when lapping optical flats, to get 
a near perfect fit. It is usually inviting trouble to put unscreened magnets 
on a seismic mass, but you could probably get away with small opposed magnet 
pairs. You might need an additional pair at some point on the periphery of 
the disk, to prevent rotation. 

       In operation, the top disk is floated using the air supply at the 
centre of the base plate. There is no wiring or tube connections needed for 
the armature disk. The distance transducer measures the deflections in the X 
and Y directions and feeds back PID signals to the coil pairs to keep the 
plate centralised. You need maybe 1 to 2 lb mass total to keep down the 
inherent noise.
 
> The surface of the aluminum disks which face the glass is recessed about 
> .001" leaving a smooth, lapped rim about 0.100" wide touching the glass. In 
> the centre of the recess is a small hole, say about .062" which is cross 
> drilled to a small air fitting, I think the common size is 10-32 (Look in 
> an aquarium shop for this.) This hose is also connected to an aquarium air 
> pump. When the pump is turned on - Voila! The free mass will rise slightly 
> on an air film which will form between the 
> plates, and the mass will float freely, with zero static coefficient of 
> 

       Can you give us references for the design of air bearings, please? I 
know that they are designed as overdamped pneumatic LCR circuits, but I don't 
have any formulae, etc. They will need very careful design. While air 
bearings have zero static friction, they may have small dynamic forces. 
       We used to use a glass plate surrounded by spring wires and heavy 
metal 'pucks' to demonstrate the laws of mechanics. Each puck had a gas tight 
chamber which was filled with dry ice. This slowly sublimated and provided 
the gas flow for the bearing underneath. However, there was always a very 
slight 'dither' on them. I do not know if this was due to varying gas 
pressure, to inadequate sublimation control, to poor bearing design, to 
variations in the surface of the glass plate, or to turbulent flow.   

> There are enhancements possible, including an air receiver and filter to 
> reduce air pressure pulsations and dirt (although the pressure pulsations 
> are far above our frequency of interest, and the mass can't follow them 
> anyway) and a temperature control system. (Always include a temperature 
> 

       You will need a very stable and clean air supply. The air bearings 
will need to be run with laminar flow, or turbulence noise and random forces 
will be generated. If the rig could be totally enclosed, the air could be 
filtered and circulated. The pumping pressure fluctuations will need a 
considerable amount of smoothing. The mass will tend to follow pressure 
fluctuations, whatever the frequency - it depends directly on the gas 
pressure to elevate it! I would be concerned about possible transitions from 
laminar to turbulent gas flow in the bearings. You might need to use a 
multiple tube peristaltic pump.

       One reason why professional systems do not often use temperature 
control, is that they are usually buried / installed below ground level. Once 
you get below about 1 m depth in most rock formations, the daily temperature 
change is reduced to milli degree levels. It is very difficult indeed = 
expensive to get anywhere near this sort of stability using active 
thermostatic control.        

> I hereby declare the season open.

       It would be an interesting experiment to try. You might initially use 
simple differential optical sensors. A pair of VTD34 7.4 sq mm photodiodes 
and an OPA2134 opamp can give about 20 nm resolution at 10 Hz. If this works, 
you can invest in capacitative transducers which give better than 1 nm 
resolution.

       Regards,

       Chris Chapman
In a message dated 05/04/=
03, dyouden@......... writes:



I have been thinking about=20=
a seismometer that has no natural frequency. Perhaps (probably) this idea is=
 not new, but never-the-less, here it is:



Hi Dave,



Imagine, if you will, a sta=
ck of, say 1" diameter aluminum tubes about 10 inches long. There are 9 tube=
s, two on the bottom, three in the next layer up, and four in the top layer.=
........ 



       I would rather not thank you! Why m=
ake it extremely complicated, when you can make it dead simple?



       Take a circular piece of plate glas=
s and drill a hole in the middle for the air supply. Support the outside edg=
es on a circular metal structure with three levelling screws. Use another, b=
ut smaller diameter sheet of plate glass or of Al etc. more or less centrall=
y placed. This will need to have the bottom surface modified, possibly by st=
icking on foil or by etching, to give either a circular or a clover leaf sha=
ped air bearing. Fit the top of the plate with two pairs of Maxwell coil + m=
agnet force actuators, at right angles and a shadow type capacitative distan=
ce transducer and you might be 'in business'! See http://physics.mercer.edu/petepag/tutoria=
l.html=20

       You might need to lap polish the tw=
o glass plates together with fine carborundum and rouge, rather like you do=20=
when lapping optical flats, to get a near perfect fit. It is usually invitin=
g trouble to put unscreened magnets on a seismic mass, but you could probabl=
y get away with small opposed magnet pairs. You might need an additional pai=
r at some point on the periphery of the disk, to prevent rotation.=20



       In operation, the top disk is float=
ed using the air supply at the centre of the base plate. There is no wiring=20=
or tube connections needed for the armature disk. The distance transducer me=
asures the deflections in the X and Y directions and feeds back PID signals=20=
to the coil pairs to keep the plate centralised. You need maybe 1 to 2 lb ma=
ss total to keep down the inherent noise.

=20

The surface of the aluminum=
 disks which face the glass is recessed about .001" leaving a smooth, lapped=
 rim about 0.100" wide touching the glass. In the centre of the recess is a=20=
small hole, say about .062" which is cross drilled to a small air fitting, I=
 think the common size is 10-32 (Look in an aquarium shop for this.) This ho=
se is also connected to an aquarium air pump. When the pump is turned on - V=
oila! The free mass will rise slightly on an air film which will form betwee=
n the=20

plates, and the mass will float freely, with zero static coefficient of=20=
friction.



       Can you give us references for the=20=
design of air bearings, please? I know that they are designed as overdamped=20=
pneumatic LCR circuits, but I don't have any formulae, etc. They will need v=
ery careful design. While air bearings have zero static friction, they may h=
ave small dynamic forces.=20

       We used to use a glass plate surrou=
nded by spring wires and heavy metal 'pucks' to demonstrate the laws of mech=
anics. Each puck had a gas tight chamber which was filled with dry ice. This=
 slowly sublimated and provided the gas flow for the bearing underneath. How=
ever, there was always a very slight 'dither' on them. I do not know if this=
 was due to varying gas pressure, to inadequate sublimation control, to poor=
 bearing design, to variations in the surface of the glass plate, or to turb=
ulent flow.   



There are enhancements poss=
ible, including an air receiver and filter to reduce air pressure pulsations=
 and dirt (although the pressure pulsations are far above our frequency of i=
nterest, and the mass can't follow them anyway) and a temperature control sy=
stem. (Always include a temperature control system. Always.)



       You will need a very stable and cle=
an air supply. The air bearings will need to be run with laminar flow, or tu=
rbulence noise and random forces will be generated. If the rig could be tota=
lly enclosed, the air could be filtered and circulated. The pumping pressure=
 fluctuations will need a considerable amount of smoothing. The mass will te=
nd to follow pressure fluctuations, whatever the frequency - it depends dire=
ctly on the gas pressure to elevate it! I would be concerned about possible=20=
transitions from laminar to turbulent gas flow in the bearings. You might ne=
ed to use a multiple tube peristaltic pump.



       One reason why professional systems=
 do not often use temperature control, is that they are usually buried / ins=
talled below ground level. Once you get below about 1 m depth in most rock f=
ormations, the daily temperature change is reduced to milli degree levels. I=
t is very difficult indeed =3D expensive to get anywhere near this sort of s=
tability using active thermostatic control.      &n=
bsp; 



I hereby declare the season=
 open.




       It would be an interesting experime=
nt to try. You might initially use simple differential optical sensors. A pa=
ir of VTD34 7.4 sq mm photodiodes and an OPA2134 opamp can give about 20 nm=20=
resolution at 10 Hz. If this works, you can invest in capacitative transduce=
rs which give better than 1 nm resolution.



       Regards,



       Chris Chapman