Alt-BEAM Archive

Message #11191



To: beam@sgiblab.sgi.com
From: "Ben Hitchcock" beh01@uow.edu.au
Date: Wed, 01 Mar 2000 22:11:06 +1100
Subject: [alt-beam] Re: mini walker




>From: Bruce Robinson



>> ... supply the +V pin on the 1381 from the last neuron in the chain.
>> That way if the sun is REALLY bright the process keeps going around
>> and around the loop while there's voltage available, rather than going
>> through once and stopping.
>
> Very nice, Ben. Elegent.
>
> Also, you would probably want to eliminate startup impulses in the rest
> of the chain (possibly a varation on Wilf's single process startup
> circuit.

I wouldn't bother. Unless you're the sort of person who hooks up a big
momma solar cell (that can drive the motors directly, negating the need for
an SE) and a fully charged cap to the circuit all at once and then hops
onboard a jet and follows the sun around the world for a couple of weeks, in
which case you might have a problem.

What I'm getting at is that the circuit will charge slowly, and when you
apply power the cap will be below switch-on voltage so all processes will
fall off the end of the loop.

>
>> ... No need to enable or disable the 240. And have the process fall
>> off the end of the core - four neurons in a chain. This means that you
>> get to gang two gates together per coil lead - two gates ganged up would
> ? give you enough juice to drive the walker, methinks.
>
> Ben, do I have this right? Are you suggesting a '240 for the microcore
> chain? Because I'm not sure a non-Schmitt inverter will work in an Nv
> net (at least not without some extra external components.

Yep you got me there. I meant the schmitt trigger one (Is it the 245?).

Well spotted,

Ben



11192 Wed, 01 Mar 2000 22:18:08 +1100 [alt-beam] Re: mini walker beam@sgiblab.sgi.com "Ben Hitchcock" Wilf,

As usual, you manage to beat me to the punch.

One question though, on the schematic you posted I couldn't figure out how
the second and fourth pair of motor drivers from the left worked.

On the first and third pair, the inputs for each pair are connected
together. The second and fourth pair seem to share their inputs. Is this a
typo? Or are you doing something very funky? Is this part of the reverser
circuit? If it's not a typo then what prevents the drivers from driving
against each other?

Ben

----------
>From: Wilf Rigter
>To: "'beam@sgiblab.sgi.com'"
>Subject: RE: mini walker
>Date: Wed, 1 Mar 2000 16:26
>

> Hi Ben,
>
> Here is a blast from the past (June 99) : the LIGHT WALKER 1.0 design as
> shown in the attached schematic. In this design the SE function has been put
> inside the microcore loop. From another perspective this is a pentacore
> design. The 1381 acts like a combination Nu, PNC and SE trigger. Note that
> the 74HC14 and AC240 are powered up all the time. Normally the 4Nv outputs
> will all be high, the motors will be stopped and no current will flow. The
> solar panel will charge up the main cap and also the memory cap of the 1381.
> The memory cap ensures that the 1381 input voltage will not drop during the
> first process interval when the motor current starts to pull down the main
> capacitor voltage. When the 1381 fires it "injects" a process into the first
> Nv with a rising edge. This process ripples through the other Nvs and when
> the process reaches the last Nv it resets the memory cap of the 1381. This
> sequence is a 4Nv process cycle which causes the bot to advance by one step
> on each leg. Next the solar panel charges the main capacitor and the 1381
> memory cap back up and when fully charged the 1381 triggers the next process
> cycle. If there is enough charge on the main cap the SE fires almost
> immediately and the walker continues uninterrupted. This embedded SE design
> can be extended to 6Nv or 8Nv designs as well.
>
> The reverser circuit is quite a nice solution to the problem of early
> reverser time out when the LIGHT WALKER only walks intermittently. To avoid
> this problem, the reverser memory capacitor is discharged in steps, one
> process cycle at time, with about 7 or 8 cycles required to return to
> forward motion regardless of the time between steps. In addition the
> reverser circuit switching is synchronized with the start of each cycle to
> avoid the typical gait recovery time associated with asynchronous reverser
> circuit switching. This synchronous reverser idea can be applied to many
> walker designs and imho is a nice contribution to the state of the art of
> BEAM tech.
>
> The LIGHT WALKER still needs tactile inputs. Perhaps it can be left as an
> exercise for the reader to complete the full LIGHT WALKER design.
>
> enjoy
>
> Wilf Rigter mailto:wilf.rigter@powertech.bc.ca
> tel: (604)590-7493
> fax: (604)590-3411
>
>
> <>
>
>
>
>> -----Original Message-----
>> From: Ben Hitchcock [SMTP:beh01@uow.edu.au]
>> Sent: Tuesday, February 29, 2000 7:38 PM
>> To: beam@sgiblab.sgi.com
>> Subject: Re: mini walker
>>
>> Here's an idea:
>>
>> You don't need a SE!
>> Just attach the 1381 to power and ground, andconnect the output of the
>> 1381 to the input of the microcore via a normal microcore cap. The 1381
>> gets to its switch-on voltage, introduces a process, and bang off it goes.
>> No need to enable or disable the 240. And have the process fall of f the
>> end of the core - four neurons in a chain. This means that you get to
>> gang two gates together per coil lead - two gates ganged up would give you
>> enough juice to drive the walker, methinks. And the nice thing about this
>> is that you only need one chip, one 1381, four caps and resistors, plus
>> your coils and solar panel and storage cap and that's it!
>>
>> Too easy.
>>
>> Ben
>>
>> > I was thinking about using a microcore instead of a master slave bicore.
>> > Either way, can pulses be stopped and resumed from their last position?
>> > I'm wondering if an Nv chain that is initiated by the SE firing would
>> > work. Switch the source to a different coil and polarity on each of 4
>> > steps, and then fall off the end. Allow the SE to charge to full. It'd
>> > store enough power for a complete step cycle. If there were enough
>> > power, I assume the SE could continuously fire the start of the Nv
>> > chain, till light levels drop and it would then rely on firing
>> > intermittently, till it looses it's charge again.
>> >
>> > Has anyone made such a circuit before???
>> >
>> > Would the Chloroplast be suitable for this purpose, and how can I get a
>> > single pulse out of it, with a second sustained pulse going to the
>> > driver (I suppose 2 chips would be acceptable, if necessary). If the
>> > driver powers the coils, then would it be possible to attach it to the
>> > steady voltage output of the Chloroplast SE and the pulse to the
>> > trigger, and have a very high dropout level, so it shuts down right
>> > away, waiting for the next charge before refiring???
>> >
>> >
>> > Ben Hitchcock wrote:
>> > >
>> > > Hi,
>> > >
>> > > > What
>> > > > would be best? I thought of an SE type system with a timed trigger.
>> Is
>> > > > there a way to create an SE that triggers different pulse outputs in
>> > > > chain? I'll play with circuits. That's what we're here for anyway.
>> New designs!
>> > >
>> > > How about using a shift register? Attach the SE to an enable pin on
>> your
>> > > motor driver (74xx240) and also to the clock pin on the shift
>> register.
>> > > You'll also need to attach the fourth (or is it fifth?) pin of the
>> shift
>> > > register to the data in
>> > > pin, and somehow start a pulse when there aren't any in the stream...
>> > > you'd need four diodes (one from each gate on the shift register) a
>> > > resistor and an inverter to create a pulse if none existed in the
>> chain.
>> > >
>> > > Maybe a shift register isn't such a good idea after all... 2 IC's
>> instead
>> > > of one... Perhaps ye olde microcore could be adapted to do it.
>> > >
>> > > Good luck!
>> > >
>> > > Hmm maybe the bot won't be able to steer using this setup. no matter,
>> it
>> > > will look pretty cool.
>> > >
>> > > 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
>> >
>>
>>
>> --
>> Nothing is foolproof to a sufficiently talented fool.
>



11193 Wed, 1 Mar 2000 12:53:40 +0100 [alt-beam] Re: BEAMAnt circuit "Thomas Pilgaard" > --- Thomas Pilgaard wrote:
> > I am on the BEAM-ant cirquit as we are speaking and
> > have a breadboarded
> > version running also. Your ant should run both
> > motors at the same time
> > although if you are exposing only one side to light
> > it should spin madly,
> > sometimes leaving the other not moving.
>
> This is totally dependant on the resistor values that
> you are using.. My beamant has a left-right waggle
> with about the same timing as a standard bicore with
> .22uF caps and a 1.5 M resistor, and it is
> aggressively phototropic in a wide variety of light
> conditions.
> if you'd lke, I can post the resistor values on my bot
> when I get home tonight..

I'd love that. Thanks a lot :)

- Thomas


> Otherwise, experiment with
> lots of different values and use the combination that
> is best for the conditions that your bot will be
> operating in.
>
>
> darrell
>
> =====
> ____________________________________
> BICOREEOS... They're BEAMtastic!!
> http://www.geocities.com/beamtastic/
> __________________________________________________
> Do You Yahoo!?
> Talk to your friends online with Yahoo! Messenger.
> http://im.yahoo.com
>



11194 Wed, 1 Mar 2000 12:53:47 +0100 [alt-beam] Re: BEAMAnt circuit "Thomas Pilgaard" Justin also has a 'Bicore Microcore' that much resembles a Beamant. However,
this has 4 photodiodes (and optional tactiles) which enables it to back up
and even pivot.

The cirquit uses both a 240 as well as 245 motor drivers. I'm not sure if
anyone has had a go at this cirquit.

http://www.geocities.com/soho/lofts/1121/

Thomas

> -----Original Message-----
> From: owner-beam@sgiblab.sgi.com [mailto:owner-beam@sgiblab.sgi.com]On
> Behalf Of Phillip A. Ryals
> Sent: Tuesday, February 29, 2000 10:02 PM
> To: BEAM List
> Subject: BEAMAnt circuit
>
>
> ok, I followed the suggestions of some people on the list to go with the
> BEAMAnt circuit. It's confusing me though...
>
> Remember the thread about a photovore that can turn on both
> motors at once?
> I got the impression from some of the posts that the BEAMAnt cicuit does
> that. I've put it together though, and it runs one motor at a time. Is
> this right?
>
> I'd like to find a fairly simple photovore circuit that uses the 240 or 14
> instead of individual transistors, mainly to make building it easier.
>
> I would appreciate any help.
>
> -phillip
>
>

Home