PSN-L Email List Message

Subject: Re: Vertical Seismometer with Feedback, Transducer Question
From: ChrisAtUpw@.......
Date: Mon, 15 Sep 2008 18:15:00 EDT


In a message dated 2008/09/15, brett3nt@............. writes:

> You may want to consider 0.5kg as a maximum.  Even slightly lower might be 
> easier to work with in a feedback design.  Also, positioning the forcing 
> coil at a greater distance from the pivot than the mass, has the effect of 
> reducing the effective mass by the ratio of their distances.

Hi Brett, 

       It is perfectly practicable to use rectangular coils and quad NdFeB 
magnet blocks for a force feedback actuator. Modern speakers seem to use Ferrite 
magnets, which are both relatively weak and very temperature sensitive. The 
pole gaps have very little clearance. This may be a limiting factor when the 
feedback coil is moving in an arc of a circle. The arm needs to be very rigid 
and accurately positioned. Sean used to machine a larger gap and use many more 
turns on his coil. This was relatively easy to do when using a cylindrical 
Alnico magnet. It my be very difficult to do using a ferrite system glued solid 
with epoxy.

> It is likely that a 16-bit D/A will be the principal limit.  From what I 
> have been told, if you make its bit sensitivity high enough to see microseisms, 
> the digital clipping level will end up being fairly low.  Others who are 
> doing that may want to comment on what they are getting.

       Strangely enough, I can just manage with a 12 bit ADC. The problem 
with the 'bare' 16 bit ADC chips is that there is usually >3 bits of digital 
noise on them, so their dynamic range is seriously restricted. This problem can e
asily be avoided with the fast ADCs by taking say 32 readings in rapid sequence 
and adding them together to give one low noise reading. You can actually get 
18 bit resolution from a 16 bit ADC this way at low sample rates.
       Larry is now marketing ADCs with reduced digital noise. Dynamic range 
should not be a problem in practice with a +/- 32768 count range and 1 bit 
noise. 
       The factors that are likely to limit you are environmental, ocean and 
circuit noise, in that order.

>  It's only when you go to a 24 bit digitizer that you can have high 
> sensitivity to weak signals and still be able to display large mid-distance quakes 
> without clipping.  

       The resolution that you can actually get with a "24 bit" ADC depends 
quite strongly on the sample rate. It may be as low as 18 bits. They tend to be 
quite expensive and you need a 50% increase in your digital storage capacity 
over a 16 bit system. Some have a very low input voltage range. This generates 
a requirement for very low noise electronics = expensive.

The instrument itself will probably not be the limit when using 16 bits. The 
highest 
> coil currents will be needed at the highest frequencies, so if you don't 
> push for too high an upper corner frequency, you should be also a little better 
> off.

       This is why you need to restrict the resistance of the feedback coil 
to <100 Ohms maximum and also keep down the coil inductance. Restricting the 
maximum frequency to 10 Hz is likely to be quite adequate for an amateur 
seismologist in most locations and eases both the electronics and the digital storage 
requirements.

> To understand the clipping issues for your location, determine what your 
> maximum acceleration is likely to be, often expressed in % of g, then use F 
> = m A to determine the peak force and calculate what that corresponds to in 
> terms of coil current.  I suspect that's what you have already 
> done.  Balancing sensitivity vs clipping level seems to be a fundamental 
> problem, which often results in the use of more than one type of instrument 
> in seismically active areas.

       Which is a problem generated by the assumed requirement to be capable 
of measuring seismic signals below the environmental background levels 
anywhere on Earth - irrespective of the actual noise at your location. Part of the 
"specmanship" which goes quite a way to explaining the cost of commercial 
seismometers.

       Regards,

       Chris Chapman   
In a me=
ssage dated 2008/09/15, brett3nt@............. writes:

You may want to consider 0.5kg=20= as a maximum.  Even slightly lower might be
easier to work with in a feedback design.  Also, positioning the forcin= g
coil at a greater distance from the pivot than the mass, has the effect of <= BR> reducing the effective mass by the ratio of their distances.

Hi Brett,

       It is perfectly practicable to use rect= angular coils and quad NdFeB magnet blocks for a force feedback actuator. Mo= dern speakers seem to use Ferrite magnets, which are both relatively weak an= d very temperature sensitive. The pole gaps have very little clearance. This= may be a limiting factor when the feedback coil is moving in an arc of a ci= rcle. The arm needs to be very rigid and accurately positioned. Sean used to= machine a larger gap and use many more turns on his coil. This was relative= ly easy to do when using a cylindrical Alnico magnet. It my be very difficul= t to do using a ferrite system glued solid with epoxy.

It is likely that a 16-bit D/A=20= will be the principal limit.  From what I have been told, if you make i= ts bit sensitivity high enough to see microseisms, the digital clipping leve= l will end up being fairly low.  Others who are doing that may want to=20= comment on what they are getting.


       Strangely enough, I can just manage wit= h a 12 bit ADC. The problem with the 'bare' 16 bit ADC chips is that there i= s usually >3 bits of digital noise on them, so their dynamic range is ser= iously restricted. This problem can easily be avoided with the fast ADCs by=20= taking say 32 readings in rapid sequence and adding them together to give on= e low noise reading. You can actually get 18 bit resolution from a 16 bit AD= C this way at low sample rates.
       Larry is now marketing ADCs with reduce= d digital noise. Dynamic range should not be a problem in practice with a +/= - 32768 count range and 1 bit noise.
       The factors that are likely to limit yo= u are environmental, ocean and circuit noise, in that order.

It's only when you go to a 24=20= bit digitizer that you can have high sensitivity to weak signals and still b= e able to display large mid-distance quakes without clipping.  <= FONT COLOR=3D"#000000" BACK=3D"#ffffff" style=3D"BACKGROUND-COLOR: #ffffff"= SIZE=3D2 PTSIZE=3D10 FAMILY=3D"SANSSERIF" FACE=3D"Arial" LANG=3D"0">

       The resolution that you can actually g= et with a "24 bit" ADC depends quite strongly on the sample rate. It may be=20= as low as 18 bits. They tend to be quite expensive and you need a 50% increa= se in your digital storage capacity over a 16 bit system. Some have a very l= ow input voltage range. This generates a requirement for very low noise elec= tronics =3D expensive.

The instrument itself will probably not be the limit when using 16 bits. The= highest

coil currents will be needed a= t the highest frequencies, so if you don't push for too high an upper corner= frequency, you should be also a little better off.


       This is why you need to restrict the r= esistance of the feedback coil to <100 Ohms maximum and also keep down th= e coil inductance. Restricting the maximum frequency to 10 Hz is likely to b= e quite adequate for an amateur seismologist in most locations and eases bot= h the electronics and the digital storage requirements.

To understand the clipping issu= es for your location, determine what your
maximum acceleration is likely to be, often expressed in % of g, then use F=20=
=3D m A to determine the peak force and calculate what that corresponds to i= n
terms of coil current.  I suspect that's what you have already
done.  Balancing sensitivity vs clipping level seems to be a fundamenta= l
problem, which often results in the use of more than one type of instrument=20=
in seismically active areas.


       Which is a problem generated by the ass= umed requirement to be capable of measuring seismic signals below the enviro= nmental background levels anywhere on Earth - irrespective of the actual noi= se at your location. Part of the "specmanship" which goes quite a way to exp= laining the cost of commercial seismometers.

       Regards,

       Chris Chapman

[ Top ] [ Back ] [ Home Page ]