Alt-BEAM Archive

Message #03496



To: Ian Bernstein ian@beam-online.com
From: Bruce Robinson Bruce_Robinson@bc.sympatico.ca
Date: Fri, 21 May 1999 21:42:59 -0700
Subject: [alt-beam] Re: Microcore problem


Ian Bernstein wrote:
>
> Hi, I just was messing around with the microcore (I've never really
> used the microcore) ...
>
> After going around in state 2 once it goes back to state 1. Sometimes
> it will stay in state 2 for 5-10 seconds but then it just goes back to
> state 1. I don't think there are any lose connections or anything so...
> Whats up?

Hey, Ian, I just had virtually the same problem ... and fixed it
somehow. I breadboarded my first microcore last night (74HC14) using all
six neurons. It worked just fine -- came up in the saturated state as
advertised.

I tried killing the processes by grounding the input to one of the
inverters and that worked as expected, no processes after that. Then I
tried killing the processes by holding one input high. Again, as
expected, it held the first neuron "on" and "absorbed" the processes as
they came around. Removing the jumper released the one process and it
just kept going around as it should.

So today's adventure was to cobble up a PNC. My circuit:
- one 74HC14 inverter.
- input connected to ground with a 2.2 uf cap.
- input connected to Vcc with a 1 meg resistor.
- output connected to one process via a 1N4248 diode.

The THEORY is, the inverter input starts low and builds slowly as the
cap charges. This keeps the output high, which is the same as shorting a
neuron input to Vcc. When the cap is charged sufficiently, the inverter
output goes low, which triggers the first process. After that, (thanks
to the diode) the PNC is "invisible" to the microcore. So you basically
just start the circuit up and it automatically sets one process going
after a brief delay. So much for theory.

I fired it up, and had three LED's on steady (ie, not moving around the
microcore). When the PNC reached it's threshold, the three processes
began to cycle around in a saturated state. I tried many variations,
including reversing the diode (in case I screwed up). Nothing worked.
Tested the PNC independantly -- it worked OK.

Finally, in a fit of inspiration (blind luck), I put a second diode in
parallel to the first. Something different happened just for a moment.
Then back to the same problem as before.

So I tried a third diode, and NOW I duplicated your problem -- except it
was more visible with 6 neurons. Basically the neuron the PNC was
connected to stayed high and "absorbed:" the processes, but then a
second process appeared 2 neurons downstream. This was while the PNC was
still active. When the PNC shut itself out of the picture, these two
processes chased themselves around the loop.

Since adding diodes was making things better, I put in a forth diode in
parallel with the other three, and now the PNC circuit behaved as
advertised.

So if you're using a diode to connect your PNC to the circuit, try
putting one or two more in parallel with it and see what happens.

Regards,
Bruce

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

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



Home