Alt-BEAM Archive
Message #09182
To: beam@sgiblab.sgi.com
From: Richard Piotter richfile@rconnect.com
Date: Fri, 14 Jan 2000 03:22:22 -0600
Subject: [alt-beam] Re: BEAM LEGO was microcore experimentation kit
Exactly! Making legos from these will result in a scaled back version of
the hextile. I for one would not be interested in a more restrictive
form. I like the DIP idea best since it can be breadboarded, and that is
the ultimate requirment for EASY expirimentation. For simple loops, use
a hextile and jumpers to set up the loops. Hextiles are great, and it
sounds like Dave is trying to make them. The thing I personaly desire is
a tiny device that can act as an individual neuron and be wired into
many crazy patterns. A more advanced device that can also serve quite
well for the simple loops and networks too.
The latest idea I had was to have the pin that I labeled "Polarity"
switched with "Bias". Then relabel it "Nu In". The In would be re
labeled "Nv In". You attach the unused pin to Ground or Vcc, depending
on whether you want a positive or negative neuron. You then have output
and bias on the other side. The (+) and (-) power can be provided by a
power rail above and below the device. If desired, the new location of
Bias could be Out, (+) and (-) above and below, and place Bias where (+)
was. THis places ALL the inputs on one side, Out on the other, and
easily accessible power rails on top and bottom.
______
| |
Bias o| |o (+)
| |
Nu In o| |o Out
| |
Nv In o| |o (-)
|______|
Ben Hitchcock wrote:
>
> Why not use cable with plugs and sockets - that is, make a whole heap of
> little veroboard units, each with two cables coming out of it.
>
> One cable has a plug on it, the other cable has a socket. You could then
> make whatever shape you liked - 2 neuron ring, 5 neuron ring, whatever. I
> think solid units would be a bit too limited in the topology for true
> experimentation.
>
> You could then have other optional units that could act as summing points -
> so it would be a unit with one plug (output) and two sockets (input).
> People could then design their own units based on this idea so that you
> could do that tetrahedron idea someone was talking about, all based around
> the same bus.
>
> A 9-pin serial connector seems to be pretty common, and very cheap. If you
> don't want to shell out the money for heaps of backshells then you could
> make your own - just drown the wire side in epoxy.
>
> In fact, you only need one side to be flexible - you could solder one
> connector directly to the veroboard, and only have ONE cable coming out of
> it.
>
> On the veroboard could be terminals for power, input, output, and a pot to
> adjust the delay.
>
> You might also want put more than one neuron in each unit - no problem.
> Since we are using hex inverters anyway, we have six inverters doing nothing
> - so it's pretty easy to add a cap and a pot to it.
>
> Now another unit could be a power unit - it just plugs right into the chain,
> and the pulse wires go straight 'through' it without this unit being part of
> the chain.
>
> I think the most important thing here is standards. If we make all our
> units to conform to the same standard then we could have a meet somewhere
> and everyone's units could plug into everyone else's units, to make a
> monster nervous net!
>
> Anyway, enough dreaming,
>
> Ben
--
Richard Piotter The Richfiles Robotics & TI web page:
richfile@rconnect.com http://richfiles.calc.org
-- Make Money by Simply Surfing the Net or responding to E-Mail!!!
-- Click below!!!
http://www.alladvantage.com/go.asp?refid=ATL147
http://www.spedia.net/cgi-bin/dir/tz.cgi?run=show_svc&fl=8&vid=329630
http://www.utopiad.com/member/getPaid?refId=284400
http://www.itadsup.com/index.cgi?ref=richfiles
http://www.desktophorizon.com/index.html?referrer=richfiles
http://www.gotoworld.com/getpaid/default.asp?rid=1040486426
http://www.dotad.com/Default.asp?ID=474867
http://www.moneyformail.com/home.asp?ref=richfiles
9183 Fri, 14 Jan 2000 03:28:02 -0600 [alt-beam] Re: does anyone know where those single inverters can be found? beam@sgiblab.sgi.com Richard Piotter I have, but you get VERY TIRED of it. It's also very dificult an easy to
break the stuff or short it out!
David Perry wrote:
>
> you can try free forming something that is 3mm by 3mm :-)
>
> -----Original Message-----
> From: Meabadboy@aol.com
> To: beam@sgiblab.sgi.com
> Date: Friday, 14 January 2000 2:18
> Subject: Re: does anyone know where those single inverters can be found?
>
> >I found this link for them ~
> >
> >http://www.fairchildsemi.com/search/search.cgi/design?keywords=TinyLogic&x=
> 49&
> >y=18
> >
> >not a bad price
> >
> >dont understand why you say they cannot be free formed though??
> >
> >Steve
--
Richard Piotter The Richfiles Robotics & TI web page:
richfile@rconnect.com http://richfiles.calc.org
-- Make Money by Simply Surfing the Net or responding to E-Mail!!!
-- Click below!!!
http://www.alladvantage.com/go.asp?refid=ATL147
http://www.spedia.net/cgi-bin/dir/tz.cgi?run=show_svc&fl=8&vid=329630
http://www.utopiad.com/member/getPaid?refId=284400
http://www.itadsup.com/index.cgi?ref=richfiles
http://www.desktophorizon.com/index.html?referrer=richfiles
http://www.gotoworld.com/getpaid/default.asp?rid=1040486426
http://www.dotad.com/Default.asp?ID=474867
http://www.moneyformail.com/home.asp?ref=richfiles
9184 Tue, 11 Jan 2000 18:40:39 +0800 [alt-beam] Re: icq? "Chia" 37878331
--------------------------------------------------------
Daniel Chia:
danndom@mbox5.singnet.com.sg
Robotics Web Page(unfinished):
http://home.crosswinds.net/~danstryder/
"It takes 99% sweat and 1% ingeniuty to have success" - Thomas Edison
---------------------------------------------------------------------
9185 Fri, 14 Jan 2000 04:01:18 -0600 [alt-beam] Re: pricing and design specs for modular nv beam@sgiblab.sgi.com Richard Piotter The output of the inverter is the output. the capacitor and resistor
work together to form an RC network.
The bias point is where you have the output now, and the other end of
the resistor is also usable as another bias point to increase the time
constant (if the bias is in series).
Input is between the output of the previous inverter and the capacitor.
I didn't even think of having the LED, or having a jumper to turn it
off. Problem is, if you attach it to one fixed polarity, you can't
switch the neuron polarity without the LED working in reverse. Maybe an
8 pin DIP would work better (and be more standard).
Bias (+)
o o
| |
| |
| |\|
| | \
Nv In o------||----*----| \o----*----o Out
| | / |
| | / _|_
| |/| /_\
Z | |
Z<--. | |
Z | | Z
| | | Z
*---' | Z
| | |
| | |
o o o
Nu In (-) Led
With LED
______
| |
(-) o| |o LED
| |
Bias o| |o (+)
| |
Nu In o| |o Out
| |
Nv In o| |o (-)
|______|
Without LED
______
| |
Bias o| |o (+)
| |
Nu In o| |o Out
| |
Nv In o| |o (-)
|______|
If you want to bias the pot, turn it down and bread board or design the
part in outside of the module, attached to the pins. for a negative Nv
(traditional setup), you attach the LED pin to (+) and the Nu Bias to
(-). For positive polarity attach LED to (-) and Nu In to (+). For Nus,
you attach Nv In instead of Nu In. Input is on the in designed for the
mode of opperation. Selectable with no jumpers on the board, and none
even for the LED!!! Out is output, and (+) and (-) are power supply.
This would work with a DIP style design. Another note on the design. the
one withe th LED has a second (-) connector. You could plug a no LED
version into a LED version socket, and if you can make it fit, even plug
a LED equipped board into a no LED socket. Point ism, they are
interchangable. Both equaly configurable. The (-) is cause the LED has
to attach to the opposite polarity as the Nv/Nu polarity. If you decide
to do a positive Nv for example, Nu In goes to the (+), but that blocks
LED from contacting the (-), and the second (-) make it possible to
attach easily with no jumpers on the board.
This info is in regards to a DIP version, which I'd greatly love to see,
more than hextile wannabees.
> David Perry wrote:
>
> Okay after working all day (in this bloody 35 degree heat wave) i have
> come up with some estimated pricing and designs.
>
> First the pricing, this includes
> PCB
> SMD LED
> SMD resistor
> SMD pot (2 meg)
> .22uf monolithic cap
> nc7s14 fairchild tiny logic single gate scmitt inverter
> 4 pin in connector
> 4 pin out connector
> 2 pin header
> 3 pin header
> two 1 pin header
> two jumpers.
>
> The total cost comes to about $4.50 which is more than i expected, but
> i can get it lower using slightly different parts.
>
> It runs as a normal NV neuron but there are a few things, jumper A is
> designed to give bias, you can adjust the pot to the value you want (0
> ohms if you want) then you use the two pins to add whatever bias you
> want, also handy cause you can just plug it in using a connector. When
> you aren't using the bias, set the pot to the value you want and use
> the jumper to join the resistor to ground.
>
> Jumper B has three headers. The first is to connect your output to
> (motor driver). The next two allow you to turn on or off the LED.
>
> The 'board in' Connector (male) is located on the side of the board,
> it provides power from the previous neuron. It also gives the output
> from the previous neuron to the input of the current one. The spare
> connector (which has a jumper) can be used for any cross board
> application, it's just an auxillary to add more functionality, use
> your imagination on just how to use it (bias, sensors, data bus).
>
> The 'board out' connector (female) is pretty self explanatory. It has
> two power connectors to pass power to the next neuron. Also has the
> output from the current neuron to give to the input of the next. The
> spare connector is the same as above.
>
> Boards can either be connected directly side to side or with a cable.
> Power only has to be provided to one neuron, which could be done by
> soldering wires onto the power pads, or i could add connector (but
> theres enough already).
>
> The whole thing would be surface mounted to a small PCB, in the shape
> of a triangle of hexagon. That means that eventually cables will be
> required or connectors used on multiple sides of the PCB. You could
> also have blank connector boards but that just adds to cost and
> complexity.
>
> What do you think?
>
> David
>
> Name: nvmodule.gif
> nvmodule.gif Type: GIF Image (image/gif)
> Encoding: base64
--
Richard Piotter The Richfiles Robotics & TI web page:
richfile@rconnect.com http://richfiles.calc.org
-- Make Money by Simply Surfing the Net or responding to E-Mail!!!
-- Click below!!!
http://www.alladvantage.com/go.asp?refid=ATL147
http://www.spedia.net/cgi-bin/dir/tz.cgi?run=show_svc&fl=8&vid=329630
http://www.utopiad.com/member/getPaid?refId=284400
http://www.itadsup.com/index.cgi?ref=richfiles
http://www.desktophorizon.com/index.html?referrer=richfiles
http://www.gotoworld.com/getpaid/default.asp?rid=1040486426
http://www.dotad.com/Default.asp?ID=474867
http://www.moneyformail.com/home.asp?ref=richfiles
9186 Fri, 14 Jan 2000 05:37:06 EST [alt-beam] Re: Photovore Performance? beam@sgiblab.sgi.com JVernonM@aol.com In a message dated 1/13/00 8:59:55 PM Eastern Standard Time,
bshannon@tiac.net writes:
> People race sollar rollers, why not race photovores?
I do it all the time. I place a 100w halogen desk lamp on a table and mark a
black dot to indicate the center of the light pool. Then, I draw a line just
beyond the edge of the light pool. I put two photovores side by side behind
the line. With the help of a friend I short both caps and let them go. The
first one to cross the dot with a tactile sensor wins. I've done this with a
lot of different designs and component values and I find that the Chiu type
bot with a mid sized cell and a 4700uF cap and a roller on the front, is the
fastest by far. Another fun variation is to put them backwards, facing away
from the light at the starting line. The bots then have to turn around before
charging to the dot. The wider based bots do pretty badly in this type of
race.
See ya,
Jim
http://www.geocities.com/SoHo/Exhibit/8281/beamart.html
ICQ# 55657870
9187 Fri, 14 Jan 2000 22:28:38 +1100 [alt-beam] Re: BEAM LEGO beam@sgiblab.sgi.com Rob If the BEAM circuit was designed to fit in a Lego block it would be an
individual's decision to use it that way avoiding any legalities.
Rob
Home