Alt-BEAM Archive

Message #03316



To: beam beam@corp.sgi.com
From: Sean Rigter rigter@cafe.net
Date: Mon, 17 May 1999 16:06:21 -0700
Subject: [alt-beam] Re:NEW IDEA


Oops, forgot to send it to the list. Well here it goes:

Hello Sean,

It sounds like a classic case of motor noise/voltage drop feedback. I
expect a 2 Nv microcore to be oscillating with one (saturation) process
Perhaps you have some LEDs connected that confirm this. The passive
components 0.1uf and 1M affect the frequency of oscillation and if the
microcore uses a 74HC14, I expect motor reversals about 10 times per
second. This is too fast for the motor to follow but if the forward and
reverse on times are not equal then there may be a slow rotation in one
direction. This is similar to a BiCore head which when not locked on,
appears to be smoothly rotating in one direction despite the fact that
it is receiving both forward and reverse signals because the oscillation
is too rapid.

Without confirmation of some details, I'll stick my neck out and take a
guess. I think what happens is that current pulses applied to the motor
in the direction it is already moving meet less "resistance" and stay on
longer than pulses which are trying to "brake" and reverse the direction
of the motor. The higher level of "motor noise" while applying motor
"braking" pulses feeds back into the core and causes a shorter pulse in
that direction. The net result is that once in motion the MicroCore
reinforces the motion in that direction.
When a collision with an obstacle occurs, the motor is stalled and now
the the direction with the least resistance will win out and the motion
reverses away from the obstacle. Finally, when trapped in a corner, it
will reverse back and forth, successively colliding with each wall until
it is free to continue to move in one direction.

All that feedback without any sensors! That's a great example of simple
emergent behaviour.

Perhaps you can give us a little feedback on the details of behaviour,
power supply type, any bypass capacitors, microcore layout, output stage
type (h-bridge, 245, etc), and motor type in order to understand what
possible feedback paths make sense to explain this seemingly very useful
behaviour.

regards

wilf



Sean wrote:
>
> Has anyone on this list run a Microcore with .1 uf caps and 1meg resistors? The
results are pretty interesting, if you use only two neurons, the motor
will run in
one direction until it gets some resistance, then switch directions, or
keep moving
back and forth until it's free.

------------------------------------------------------------------------

eGroups.com home: http://www.egroups.com/group/alt-beam
http://www.egroups.com - Simplifying group communications



Home