In a message dated 22/08/2009, barry_lotz@............. writes:
I was=20
trying to decide the best displacement sensor style. What was the proble=
m with=20
inductive style sensors?
Hi Barry,
Ordinary amateur coil + magnet sensors tend=
to be=20
more limited by noise. You can't use them to correct position drift with=
=20
temperature. The use of NdFeB quad magnet blocks will give a greatly incre=
ased=20
output compared to Alnico U magnets.
I agree=20
LVDT's can have clearance issues.
This depends on whether you use commercial se=
nsors=20
or design your own. You can buy commercial sensors with a wide clearance.=
=20
Schaevitz produce them amongst others. Commercial LVDTs seem to have quite=
noisy=20
electronic detector circuits, but good designs are available
VRTD's -=20
I like, but someone had a problem with it but I can't remember what the=
reason=20
was.
I don't know of any problems / can't remember=
=20
hearing of any. You do need the moving plate to considerably overlap the=
field=20
coil poles.
I use it=20
on a small version of STM style sensor. I do get an occasional low frequ=
ency=20
oscillation which I can't nail down (~0.01 hz). I tentatively attribute=
to=20
maybe my triple feedback is slightly off. I would use a capacitive senso=
r but=20
they seem so large to get a nominal capacitance value. Especially when=
one is=20
trying to design a small unit.
A capacitance of ~10 pF is commonly used. You=
can=20
get much lower overall noise with a capacitative system - so long as you=
use=20
sine wave excitation. The common problem with the variable plate sepa=
ration=20
design, is the large change possible in the air damping as the plate=
gap=20
varies. The allowable movement is small and this is better used in a=
=20
feedback design. Plates with drilled array of holes may be used. A sh=
aded=20
pole design with the shadow plate moving parallel to the other plates=
may=20
be used. These can be used in open circuit or feedback designs and can mea=
sure=20
movements from nano metres to metres. See also LCDTs.
Could this be an undamped torsional oscillati=
on?=20
Alternatively, are you sure that it is correctly damped? What sort of capa=
citors=20
did you use? Sounds more like a phase shift problem?
Regards,
Chris Chapman