Alt-BEAM Archive

Message #02992



To: beam@corp.sgi.com (mailing list)
From: Benjamin Edward Hitchcock beh01@uow.edu.au
Date: Thu, 6 May 1999 07:44:29 +1000 (EST)
Subject: [alt-beam] Re: LC circuits


I think what you are talking about is a microcore - style circuit, that
uses LC tank circuits to give the timing. Is this right?

The reason they aren't used is because of:
a) simplicity. You need a resistor, a capacitor, and an inverter to make
a traditional neuron, whereas to persuade a tank circuit to oscillate once
and once only is no mean feat. The things resonate (By their very nature)
so once you tell it to fire, it will 'ring' for a short time. The
microcore circuit uses lots of neurons connected in series, and each
neuron is expected to fire once (and only once) every trip around the
microcore. If you used inductors, they would resonate, and cause you big
headaches.

b) variable pulse-width. Apart from the notable 'capaciflector' circuit
that did the rounds a few months ago, most circuits rely on changing the
resistance (or at least current flow) to change the duty cycle of the
pulses around the microcore. Put another way, the way to make a pulse
longer is to make the resistance to ground of the neuron input higher.
It's easy to make a microcore this way - several types of sensors change
their resistance based on light levels or heat or touch etc. Changing the
resistance of an LRC circuit only changes its damping factor, not it's
frequency. So to wire this circuit up with a LRC circuit, and a resistive
sensor, would take a fair bit of electronics. You can't just stick the
sensor in parallel the way you can with a 'traditional' microcore.

c) size. Inductors are big. When you can get surface mount resistors and
capacitors the size of half a grain of rice, there's no need to use
anything else. Especially since inductors are usually quite large. If
you decided to use small inductors, then the inductance of the hookup
wires affects things - so the circuit that worked fine on the breadboard
doesn't work at all when free-formed.


I hope this answers your questions. If you do feel that I'm talking
through my hat, feel free to try using inductors. I would be interested
to
see, maybe, a microcore metal detector that changed course to follow a
steel rod on the ground. That would work.

Ben

----- Forwarded message from Alexander H Farley -----

This may seem like a stupid question (it may just be so obvious
that it does not immediately occur to me why not), but why aren't LC
(inductor-capacitor) circuits used in any of the higher level circuits?
Can't LC's be used for central pattern generation? Or is there a reason
why inductors do not mesh well with the desired outputs? Are the outputs
too regular (if they are why not have some type of LRC circuit)?
Just wondering.
Alex Farley

------------------------------------------------------------------------
Too much effort to find the stock info you want each day? StockMaster
lets you enter a company name and quickly shows you a daily quote,
chart, and news all on one page. Free! http://clickhere.egroups.com/click/71

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


----- End of forwarded message from Alexander H Farley -----

--
Nothing is foolproof to a sufficiently talented fool.

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

Home