David,
Thanks for prompt advice. Like you suggested I want to develop the model.
Are there any tutorials or available models that can be modified for this
problem that I can get some guidance? I appreciate if you can help me
finding some working models to go about. Thanks.
Vijay
-----Original Message-----
From: David Meeker [SMTP:dmeeker@xxxxxxxx]
Sent: Wednesday, May 28, 2003 10:33 AM
To: femm@xxxxxxxxxxxxxxx
Subject: Re: [femm] Speed Sensor
Earanky, Vijay (GEAE, Unison-JAX) wrote:
> David,
>
> I want to simulate a speed sensor using FEMM. I am new to this
program and
> its capability. I will describe the problem briefly here. I
have a permanent
> magnet and attached to it is a core (1010 steel). There are
5000 turns of
> wire on it. At the end of the core/coil assembly, there is a
gear wheel of
> ferrous material. As the wheel rotates it creates flux in the
coil and
> hence an emf generated of ac voltage. Now I want to develop a
model to
> see the relationship between no of turns, core material, the
air gap
> (between gear tooth and core) and magnet strength. The idea is
to get
> a linear curve between the speed and the induced voltage. Can
the FEMM
> model this and if so can someone show a direction. Thanks in
advance.
>
> Vijay
FEMM can model this sort of problem (as long as 2D is a good
approximation of your geometry), but you'll get a lot more
mileage as
far as getting a feel for number of turns, core material, and
air gap,
by first making a relatively simple analytical model of your
device--the
effects of all of these parameters would be clearly in evidence
in such
a model.
One interesting thing a simple model would show that if eddy
currents
(and possibly mechanical vibrations that would create "significant"
variations in the nominal air gap or interacting pole area) are
neglected, the induced voltage waveform ought to to have the same
profile regardless of speed (just compressed over a shorter
time), and
the voltage amplitude ought to scale linearly with speed. This
means
that if you are seeing big nonlinearities in the output
amplitude, I'd
guess that you should be either looking at motion-induced eddy
current
effects in your assembly (FEMM can be used as /part/ of such an
analysis, but not trivially--again, it would be best to start by
including eddy current effects in the simple model as parasitic
"shorted
turns" around various parts of the teeth, core, and magnet), or
artifacts of the way that you are measuring and/or filtering the
output
voltage. Then again, this could be the kind of situation where
a little
Horowitz and Hill would go a long way, e.g. make a circuit that
looks
for zero crossings in the induced voltage and ignore second-order
effects in your sensor.
Dave.