From: bwhite@oucsace.cs.ohiou.edu (William E. White )
Subject: Polhemus sensor info request
Date: 3 May 91 14:44:48 GMT
Organization: O.U., Harvard on the Hocking, Berkeley in the Boondocks!



According to my understanding of how the Polhemus sensor works, three
coils are used to generate a magnetic field, which are received by three
coils in the receiving unit.  Distance is calculated from the overall
signal strength; orientation from the relative induction in the three
different receiver coils.  Given that this is true, ...

1) Wouldn't it be possible to simply broadcast with one coils, and receive
with three (or vice versa)?  The orientation can still be determined since
the "image" of the transmitter coil will show up in all three receiver
coils, with strength varying depending on the angular difference; and the
distance can be calculated from the sum (or other function) of the three
receiver coils.

2) I assume (wrongfully, perhaps) that the induced current in a coil is
inversely proportional to the square of the distance; ie

                I2 ~= I1 / d^2

I have no idea what the function is for difference in angles.

My questions are:

A) How correct or incorrect is my first assumption?  Mathematically I can
see no reason for the extra two transmit coils except for increased
accuracy and for redundancy.

B) How does the Polhemus detect a 180 degree rotation around one of its
coil axes?  Does it actually check for the phase difference between the
transmit and receive signals?

C) Can anyone supply the correct formulae for the induced current in a coil
given a current in another coil; both magnitude and phase would be 
appreciated.  (Actually, would it be current or voltage?)

D) Why is the Polhemus so expensive?  Frankly I don't see any reason why all
of this couldn't be done with the coils, some phase-locked loops, a bunch of
op-amps (maybe log instead of linear to take care of the squared term in
the current), and/or a good microprocessor and some EPROM?  Sure it wouldn't
be a $20 project, but it wouldn't be $1000 either.  Plus, with sufficient
computational power, one could put transmit coils wherever one wanted (say,
one on each finger, and then calculate finger positions in reference to the
base coil on the hand).

E) Any references for more info?  I'd like to try and build a glove interface
using only Polhemus-type sensors (or mostly Polhemus-type).  After all, if you
stick the transmit coils over the third segment of each finger (like a ring),
you'd have a good idea where each finger was.  From that, and given the
human anatomy, one could determine the position of all finger segments.

Thanks...


|  Bill White   +1-614-594-3434 |  bwhite@oucsace.cs.ohiou.edu (normal)  |
|  Standard login hours:        |  bwhite@view.cs.ohiou.edu (alternate)  |
|   1-3,5-7PM EST M-F (oucsace) |  more info via `finger bwhite@view...' |



[MODERATOR'S NOTE:  Please reply publicly.  This is important info.  Your
expertise and insights are welcome by all! -- Bob Jacobson]