Alt-BEAM Archive
Message #02273
To: Sean Rigter rigter@cafe.net
From: Wouter Brok w.j.m.brok@stud.tue.nl
Date: Tue, 13 Apr 1999 10:29:08 +0200
Subject: [alt-beam] Re: Linked BiCores; How Does The Slave Cycle?
Hello Wilf,
>Actually I send those waveforms to you privately but as long as we are
>discussing them here I have attached them to this reply.You will find one
>additional circuit which exploits the overvoltage at the input of the
>"isolated" suspended BiCore to generate some additional supply voltages
>(+7.5V and -2.5V).
Whoops, sorry, didn't see that; smart circuit !!!
>The graphs are approximations I made with a CAD program. They don't have a
>calibrated time axis. You are correct that the 74HC04 and 74HCU04 would have
>the same exponential curve using the same RC parts but a different period
>because of the difference in thresholds. The 74HC04 inverter has a voltage
>gain of about 100 and a narrow (50mV) gap around 2.5V between the upper and
That's were I start lacking knowledge (I still am surprised how little
electronics is in the standard curriculum of physics (I'm a graduate
student physics and everything I know about electronics comes from
experiments and rational thinking):
Voltage gain of about 100, is that the voltage-amplification factor from
input to output, bounded by supply voltage levels, or am I totally wrong?
The exponential part of a suspended bicore, with a symmetric inverter
(threshold +V/2) I calculated to be:
V1 + V2 = A
V1 - V2 = B exp(-2t/RC)
with V1 the input-voltage if one inverter and V2 the input-voltage of the
other.
This results in e.g.
V1 = A' + B' exp(-2t/RC)
With A' = A/2 and B' = B/2
I figured that A should be +V and B the threshold-voltage +V/2. With these
assumptions V1 would never reach the threshold-voltage, so I concluded that
for a symmetric inverter noise would have a large influence.
I was still thinking about inverters with an asymmetric set of threshold
voltages (I mean different slope up and slope down threshold-voltages),
since one will change state before the other does. All this is quite hard
to put in a set of analytic equations, so I wanted to find a digitizing
oscilloscope at uni to make some pictures. It's a great miss not to have
that equipment if you want to study something in a deeper way
>lower thresholds. The 74HCU04 inverter has a much smaller gain and widely
>separated upper and lower thresholds. In this description, threshold means
>the input voltage level at which the output voltage starts to change. The
>graphs for the HC/AC240 are very similar to the 74HC/AC04 so the example
>holds for both cases. However there are important effects related to output
>load current and obviously the 74HC04 has much less current capability
>(higher output impedance than the 74AC04 and 74HC/AC240. Power supply noise
>and "ground bounce" can change the period of a master or slave BiCore by
>50% which could be suppressed by good supply decoupling and supply bus
>layout for more repeatable results. Of course, this all falls into the
>category of "feedback" so you can explore ways to the modulate BiCore pulse
>width in response to change in load by deliberately allowing a certain
>amount of noise to influence the BiCore period. This is the way in which
>BEAM design exploits electronic circuit behaviour on the edge of chaos
>giving rise to such complex and surprising behaviour in BEAM bots.
Exactly what intrigues me in BEAM; and more escpecially in the suspended
bicore circuit.
To which extend do you think this noise-feedback can be controlled or made
usefull in more complex BEAM-robots?
When I build my first two-motor walker I made it with a microcore and I
thought it was a great bummer: it didn't do anything except for walking in
a straight line or on a circle. Thinking about it I came up with a theory:
a walking robot needs one degree of freedom (DOF) for lifting its body and
the DOFs that are left can be used to span a space in which it walks. The
two-motor-walker has two independent DOFs so it has one left to create a
walking path, which will be one-dimensional and thus a straight line or a
circle or so. My house-mates doubted that theory though, because if you
make a two-motor robot with two wheels and replace every wheel by for
example 5 spokes, it still could move on a two-dimensional area. (we than
had a discussion about what exactly a walking robot was, since, obviously I
didn't think the one they made up really walked). Then I replaced the
microcore by a bicore and it actually seemed to be able to span the whole
two-dimensional area (very large radius , so my house-mates were happy, and
so was I since I was amazed by the complexity of the circuit.
Now I am wondering actually about this: of you create a fractal, which
consists of a line (I guess you will have seen those standard pictures in a
beginners fractal and chaos book) you can cover quite a bit of a
two-dimensional area with that one-dimensional line-fractal. That's where
they usually introduce a non-integer fractal dimension, like 1.7 or so. 'Is
the two-motor bicore walker walking on such a fractal?' my thought was.
Just a thought; slightly off the original topic, but well ...
>Hi Wouter, Here is something I posted to the list in Jan 1999 (when I
>was still learning the basics) and I updated the attached article to
>include results of more recent experiments.
I read your article then, and I was at approximately the same level of
understanding the circuit. (I remember your anti-lock bicore). But, as Chiu
(I think it was) mentioned once: you think up things faster than I can try
them out by experiments. I'm looking forward to read the attached article
later today.
>Your observation of "noise" as the trigger for high gain inverter (ie
>240) bicores is spot on! The program is AutoSketch by AutoCad and the
>cad drawing is captured on thr screen and converted to a GIF with Paint
>Shop. The subject BiCore is surprisingly complex and it is easy to omit
>some of the more detailed circuit behaviour. A good article makes a
>subject understandable and exciting for beginners as well as interesting
>and challenging to more advanced readers.
In a couple of weeks I think I have something to send, unless everything I
have in mind will be in what you wrote of course.
Regards,
Wouter Brok.
------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/list/alt-beam
Free Web-based e-mail groups by eGroups.com
Home