In a message dated 26/05/01, blottobear@.......... writes:

Hi Chris,
> You are almost there-forget the photocells and use a CCD linear array 
> and a laser beam reflected off the vertical pendulum. The CCD array 
> gives you a clean, robust output that is a direct analogue of the POSITION 
> of the seismic mass. Hence it is not frequency-response limited. It is 
> truly an absolute position sensor, rather than the classic velocity 
> sensor with all its' attendant problems. I am cobbling up a system 
> using the guts from an old HP Laser-Jet printer.  Another similar 
> method is to monitor the reflected laser beam with a simple optical 
> interferometer and count the resulting fringe patterns. This is done in the 
> USGS gravity observatory instrument here in Boulder. Now you can 
> measure displacements down to to wavelength of whatever color of 
> light your laser is! 

Dear Dave,

       What limitations are the USGS gravity observatory getting in terms of 
labda? For direct measurements with instruments like seismographs, you 
probably do need to get better resolution than can be easily provided by the 
wavelength of light. Measuring to small fractions of the wavelength is 
possible but difficult.
  
      I hadn't considered using a CCD strip because you have the addressing 
to do as well as the sensing, there is a minimum size of cell and a maximum 
length of strip that you can get. Using an optical lever increases the 
sensitivity but reduces the range. The mirrors can introduce vibration and 
thermal noise. I don't remember CCD arrays being particularly cheap. 

       On a practical point, I don't think that you can mode lock the small 
cheap solid state lasers can you? They certainly don't focus very well 
compared to a He/Ne laser and the edges of the spot seem to move about. There 
will be a response rate / illumination level dependency for the CCD's, but it 
should not worry us. Aren't interferometers a bit expensive to construct, in 
requiring optical components finished to a great accuracy?

> The main design problem is to get a CCD array with LOTS of pixels. This 
will >directly determine the dynamic range of the instrument in Db (decibels 
related to >some given mechanical displacement.) So to have 30db dynamic 
range, you would >need to have 1000 pixels in the CCD array. This is not very 
good, even for a crude >seismo. So to get 40db. dynamic range, you need 
10,000 pixels. This is about the >limit for affordable CCD chips.  

       What are the physical sizes of the CCD pixels and how many can you get 
in a single strip, please?

      If you choose a method which relies on the wavelength of the light, 
like optical interference fringes, you will inevitably be limited by this. If 
you choose to use photon counting, the wavelength limitations do not apply, 
although other limits do. With a linear tungsten bulb and some simple optics, 
a couple of razor blades for the slit and two good Si photocells, you get an 
output which depends on the statistics of photon counting and not on the 
wavelength of the light. The method is remarkably good for our sort of tasks. 
The response frequency is limited by the output low pass filter and by the 
count statistics / position movement relationship. 
       
      Regards,  

      Chris Chapman      

In a message dated 26/05/01, blottobear@.......... writes:



Hi Chris,

You ar
e almost there-forget the photocells and use a CCD linear array 

and a laser beam reflected off the vertical pendulum. The CCD array 

gives you a clean, robust output that is a direct analogue of the POSITION 

of the seismic mass. Hence it is not frequency-response limited. It is 

truly an absolute position sensor, rather than the classic velocity 

sensor with all its' attendant problems. I am cobbling up a system 

using the guts from an old HP Laser-Jet printer.  Another similar 

method is to monitor the reflected laser beam with a simple optical 

interferometer and count the resulting fringe patterns. This is done in the 

USGS gravity observatory instrument here in Boulder. Now you can 

measure displacements down to to wavelength of whatever color of 

light your laser is! 




Dear Dave,



       What limitations are the USGS gravity observatory getting in terms of 

labda? For direct measurements with instruments like seismographs, you 

probably do need to get better resolution than can be easily provided by the 

wavelength of light. Measuring to small fractions of the wavelength is 

possible but difficult.

  

      I hadn't considered using a CCD strip because you have the addressing 

to do as well as the sensing, there is a minimum size of cell and a maximum 

length of strip that you can get. Using an optical lever increases the 

sensitivity but reduces the range. The mirrors can introduce vibration and 

thermal noise. I don't remember CCD arrays being particularly cheap. 



       On a practical point, I don't think that you can mode lock the small 

cheap solid state lasers can you? They certainly don't focus very well 

compared to a He/Ne laser and the edges of the spot seem to move about. There 

will be a response rate / illumination level dependency for the CCD's, but it 

should not worry us. Aren't interferometers a bit expensive to construct, in 

requiring optical components finished to a great accuracy?



> The main design problem is to get a CCD array with LOTS of pixels. This 

will >directly determine the dynamic range of the instrument in Db (decibels 

related to >some given mechanical displacement.) So to have 30db dynamic 

range, you would >need to have 1000 pixels in the CCD array. This is not very 

good, even for a crude >seismo. So to get 40db. dynamic range, you need 

10,000 pixels. This is about the >limit for affordable CCD chips.  



       What are the physical sizes of the CCD pixels and how many can you get 

in a single strip, please?



      If you choose a method which relies on the wavelength of the light, 

like optical interference fringes, you will inevitably be limited by this. If 

you choose to use photon counting, the wavelength limitations do not apply, 

although other limits do. With a linear tungsten bulb and some simple optics, 

a couple of razor blades for the slit and two good Si photocells, you get an 

output which depends on the statistics of photon counting and not on the 

wavelength of the light. The method is remarkably good for our sort of tasks. 

The response frequency is limited by the output low pass filter and by the 

count statistics / position movement relationship. 

       

      Regards,  



      Chris Chapman