In a message dated 02/09/2005, gcouger@.......... writes:

Chris,

I am late to the discussion so I have missed things. Are  you using the 
principle of two cans with large areas connect together  by a tube with a very 
small diameter bore and an air bubble introduced  into the bore will amplify the 
change in levels in 
one can related to the  other and the change in the position of the bubble 
depends on the are of the  can to the area of the tube connecting them.

Fantastic amplification is  possible with very simple tools. The surface of 
the bubble blocks nearly 100%  of the light shined on it so photocells or 
rotating photo paper can easily  track the earths movements.

Gordon
Hi Gordon,
 
    I am aware that it is possible to optically sense  the ends of an air 
bubble in a tube, but I am rather doubtful if this would be  practical in the 
middle of a seismometer. Air bubbles are inherently very  compressible. Maybe a 
heavily dyed oil 'bubble' or some mercury would work? You  can choose the 
diameter of the main tube to give roughly the correct  damping. Without knowing the 
dimension of what you are calling 'a  small diameter tube',  I can't really 
comment.
 
    Ian was suggesting making a horizontal seismometer  using a 15 cm dia 
tube, half filled with water and weighing the ends. Half  filled sensor tubes are 
used in some large tiltmeters. I don't  think that Ian had found space on the 
back of his envelope to do detailed  analysis of issues like the end 
fittings, the type of sensor, the limits of  accuracy and drift, or how such a system 
behaves dynamically, damping, wave  formation, etc. I am not sure how you 
could support a 10m tube at the ends,  without the centre section oscillating like 
fun. 
 
    I hope that Ian can produce a full scale working  model and measure it's 
performance. 
 
I did a  back-of-the-envelope calculation today for a 10 m long pipe, 15cm in 
diameter,  half filled with liquid, sitting on 1 support at each end.  When 
the pipe  is tilted by 1 micro radian, the difference in loading between the 
posts is  the equivalent of around 3 or 4 grams.
Could this be approached by  monitoring the loading on the supports?
Ian Smith


_ChrisAtUpw@........ (mailto:ChrisAtUpw@........  wrote:
> In a  message dated 01/09/2005, ian@........... writes:
>
>   Hi,
>     one assumption I made was that the  mean signal, generated by the
>     mean mass is  subtracted and you are then only measuring the
>      changes.  If some piezo sensor can withstand a load generated by  44
>     Kg (2 supports), then 2 of them might give the  required signal.  By
>     electrically differencing  the signals from the 2 piezo sensors, the
>     remainder  is the changing mass + noise.  Just a thought...
>
> Hi  Ian,
>
>     I don't really see why you need such a  large pipe, when something
> nearer 1" would do fine. If you put on  larger diameter end pots, you
> increase the period.
>   Trying to match up two sensors allowing for temperature  coefficients
> and drifts just doesn't work to better than about  0.1%.
>     When it is dead easy to measure water levels  to microns, why bother
> with weight sensors? The inexpensive piezo  sensors drift in output
> voltage with temperature.
>   There are plenty of problems in making precision measurements.  'Doing
> things the hard way' is just plain dumb.


Regards,
 
    Chris Chapman





In a message dated 02/09/2005, gcouger@.......... writes:
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000=20
  size=3D2>Chris,

I am late to the discussion so I have missed things=
.. Are=20
  you using the principle of two cans with large areas connect together=20
  by a tube with a very small diameter bore and an air bubble introduce=
d=20
  into the bore will amplify the change in levels in 
one can related to=20=
the=20
  other and the change in the position of the bubble depends on the are of t=
he=20
  can to the area of the tube connecting them.

Fantastic amplificatio=
n is=20
  possible with very simple tools. The surface of the bubble blocks nearly 1=
00%=20
  of the light shined on it so photocells or rotating photo paper can easily=
=20
  track the earths movements.

Gordon
Hi Gordon,
 
    I am aware that it is possible to optically sen=
se=20
the ends of an air bubble in a tube, but I am rather doubtful if this would=20=
be=20
practical in the middle of a seismometer. Air bubbles are inherently very=20
compressible. Maybe a heavily dyed oil 'bubble' or some mercury would work?=20=
You=20
can choose the diameter of the main tube to give roughly the correct=20
damping. Without knowing the dimension of what you are calling 'a=20
small diameter tube',  I can't really comment.
 
    Ian was suggesting making a horizontal seismome=
ter=20
using a 15 cm dia tube, half filled with water and weighing the ends. H=
alf=20
filled sensor tubes are used in some large tiltmeters. I don'=
t=20
think that Ian had found space on the back of his envelope to do detail=
ed=20
analysis of issues like the end fittings, the type of sensor, the limits of=20
accuracy and drift, or how such a system behaves dynamically, damping, wave=20
formation, etc. I am not sure how you could support a 10m tube at the ends,=20
without the centre section oscillating like fun. 
 
    I hope that Ian can produce a full scale workin=
g=20
model and measure it's performance. 

<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>I did a=20
  back-of-the-envelope calculation today for a 10 m long pipe, 15cm in diame=
ter,=20
  half filled with liquid, sitting on 1 support at each end.  When the=20=
pipe=20
  is tilted by 1 micro radian, the difference in loading between the posts i=
s=20
  the equivalent of around 3 or 4 grams.
Could this be approached by=20
  monitoring the loading on the supports?
Ian Smith
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>ChrisAtUpw@....... wrote:
> I=
n a=20
  message dated 01/09/2005, ian@........... writes:
>
> =20
     Hi,
>     one assumption I made was that=
 the=20
  mean signal, generated by the
>     mean mass is=20
  subtracted and you are then only measuring the
>    =20
  changes.  If some piezo sensor can withstand a load generated by=20
  44
>     Kg (2 supports), then 2 of them might give t=
he=20
  required signal.  By
>     electrically differen=
cing=20
  the signals from the 2 piezo sensors, the
>     remai=
nder=20
  is the changing mass + noise.  Just a thought...
>
> Hi=20
  Ian,
>
>     I don't really see why you need su=
ch a=20
  large pipe, when something
> nearer 1" would do fine. If you put on=20
  larger diameter end pots, you
> increase the period.
> =20
     Trying to match up two sensors allowing for temperature=20
  coefficients
> and drifts just doesn't work to better than about=20
  0.1%.
>     When it is dead easy to measure water lev=
els=20
  to microns, why bother
> with weight sensors? The inexpensive piezo=20
  sensors drift in output
> voltage with temperature.
> =20
     There are plenty of problems in making precision measurements=
..=20
  'Doing
> things the hard way' is just plain dumb.


    Regards,
 
    Chris Chapman