On Wed, 18 Sep 2002 22:20:06 EDT ChrisAtUpw@....... writes:
      The capacitor plates have an inverse separation capacity
relationship. Depends whether you want to compare frequency or period.
Using moving cores in inductors might give a more linear relationship and
allow larger movements of the pendulum. 
That is why I decided on moveable cores, although with force-feedback the
actual displacement should be minimal and the inverse relationship would
only tend to increase sensitivity anyway. So capacitive tuning would
probably work as well. The main reason I wanted to try cores was that I
thought using the same coil for detection and feedback was neat...



      You could cut a ferrite rod in half and use the halves to tune two
coils. You might have to provide screening to prevent the oscillators
from locking. You might put a flat ferrite plate on the pendulum in
between two C ferrite transformer cores. You can buy coil formers with
two equal sections for some core sizes. 
I suspect there may be a lot of experimentation here, including
separating two iron core sections with a plate of MU, which will be the
first experiment if I find too much coupling at the outset.

      A CD4046 phase lock loop IC comes complete with two types of
precision phase comparitors. This would allow continuous sampling. 

I considered that type of route, but I was concerned about general
free-running oscillator stability and felt that one way or the other I
would need to average out their rate through longer sampling periods. I
would either have to filter the CD4046 output or simply go the counter
route, and the junk box ruled the day. Since I am in CT, long period is
what I look for, and by simply choosing the bits on the counters I can
have sample rates from 8 Hz on up. I think I settled on about 30 Hz which
was then integrated in the pen recorder with nice smooth results.

      Another alternative which avoids oscillator lock problems is to use
the two coils in a variable reluctance switching oscillator and look for
tiny changes in the mark / space ratio.

Actually, I have pondered taking that a step further to see whether the
PWM effect would actually provide the centering force at the same time if
the oscillator current was sufficient...that would be the ultimate.
Anyway, it's a lot of fun trying to reinvent the wheel. I suppose I'll
eventually try a beam of light on photographic paper wrapped around a
drum and come full circle.






On Wed, 18 Sep 2002 22:20:06 EDT ChrisAtUpw@....... writes:

  
      The capacitor plates have an=
=20
  inverse separation capacity relationship. Depends whether you want to =
compare=20
  frequency or period. Using moving cores in inductors might give a more =
linear=20
  relationship and allow larger movements of the pendulum.=20
  

That is why I decided on =
moveable cores,=20
  although with force-feedback the actual displacement should be minimal =
and the=20
  inverse relationship would only tend to increase sensitivity anyway. So=20
  capacitive tuning would probably work as well. The main reason I wanted =
to try=20
  cores was that I thought using the same coil for detection and feedback =
was=20
  neat...
  



  
      You could cut a =
ferrite=20
  rod in half and use the halves to tune two coils. You might have to =
provide=20
  screening to prevent the oscillators from locking. You might put a =
flat=20
  ferrite plate on the pendulum in between two C ferrite transformer cores.=
 You=20
  can buy coil formers with two equal sections for some core=20
  sizes. 


  
I suspect there may be a lot of experimentation =
here,=20
  including separating two iron core sections with a plate of MU, =
which=20
  will be the first experiment if I find too much coupling at the=20
  outset.

  

      A CD4046 phase lock loop IC =
comes=20
  complete with two types of precision phase comparitors. This would allow=
=20
  continuous sampling. 
  
 
  
I considered that type of route, but I was concerned about =
general=20
  free-running oscillator stability and felt that one way or the other I =
would=20
  need to average out their rate through longer sampling periods. I would =
either=20
  have to filter the CD4046 output or simply go the counter route, and the =
junk=20
  box ruled the day. Since I am in CT, long period is what I look for,=20
  and by simply choosing the bits on the counters I can have sample =
rates=20
  from 8 Hz on up. I think I settled on about 30 Hz which was then=20
  integrated in the pen recorder with nice smooth=20
  results.

      Another =
alternative=20
  which avoids oscillator lock problems is to use the two coils in a =
variable=20
  reluctance switching oscillator and look for tiny changes in the mark / =
space=20
  ratio.
  
 
  
Actually, I have pondered taking that a step further to see=
=20
  whether the PWM effect would actually provide the centering force at the =
same=20
  time if the oscillator current was sufficient...that would be the =
ultimate.=20
  Anyway, it's a lot of fun trying to reinvent the wheel. I suppose I'll=20
  eventually try a beam of light on photographic paper wrapped around a =
drum and=20
  come full circle.
  
 
  
Tom