Alt-BEAM Archive

Message #09361



To: beam@sgiblab.sgi.com
From: "David Perry" davidperry@geocities.com
Date: Wed, 19 Jan 2000 21:17:39 +1100
Subject: [alt-beam] Re: Pertaining to the Hextile things. You should really make them hextiles and not DIP


>The Nv/Nu deal being
>configured by the board also prevents you from putting the wrongly
>configured neuron in a socket (yeah, you'd just switch the resistor and
>cap, but if it's standardized, it's simpler to work with).


Fair enough

>below, where i talk about cutting the DIP out, I'm refering to cutting
>the hextile's outer PCB, leaving only the core, and soldering headers to
>it, thus CONVERTING it into a DIP package INSTEAD of a hextile.

I get that, but i propose that you have two PCB's, one DIP and the hextile
base with a IC socket so that it can be converted to and from DIP and
hextile.

>trace cuts are easier to work with. they can be resoldered easy. you
>have it configured by default to be an Nv, and by cutting the existing
>traces and soldering the pre opened pads. Desoldering such tiny pads is
>more dificult than cutting a trace. Believe me, I know!


I've never had any problems doing this with a desoldering pump. It just
seems to me that cutting the trace isn't a very elegant solution.

>What exactly are you talking about with headers and cables???

Okay, imagine on the end of each DIP that you have a 3 pin header (maybe two
for branched NV nets), each header has 2 power lines and the signal line.
You use a 3 core cable to carry the signal (and power) from neuron to
neuron. The point of which is so you don't have to wire up the power lines
for each chip.


David

>> >Who has a bread board on their final robot anyway? I designed the pinout
>> >of that DIP chip so you have the positive power runnnig above the chip
>> >and ground below the chip. The ouputs can lead straight to the inputs of
>> >the next neuron.
>>
>> The original intent was experimentation, these aren't designed to make it
>> easier to build a NV net on your robot (although the DIP design would be
>> very good at that).
>>
>> >As you can see, you can create a 4 Nv Microcore with NO JUMPERS. I'm not
>> >about to draw it in ASCII, but it'd be a simple task to physicaly have
>> >the chips oriented at angles to each other, literaly forming a physical
>> >loop!
>>
>> But you couldn't do that with my breadboard...
>>
>> >On each board, a .22uF capacitor could come soldered on (SMT or through
>> >hole, whatever works). The resistor (and optionaly, BOTH the resistor
>> >and capacitor) can have small machined sockets so you can switch out
>> >various values of resistors. It'd be nice to have a variable resistor,
>> >but the way I see it, it can always be added later.
>>
>> You could use the sockets and instead of having the two inputs you could
>> just swap the resistor and cap around.
>>
>> >For breadboarding, the chips can easily be placed in rows and groups
>> >that indicate loops, branches, etc. It can't be that bad! If you have a
>> >decent bread board with more than one row for chips, then it's even less
>> >of a complaint!
>>
>> One of the major hassles for the DIP chips will be wiring up the power
which
>> i really don't want to do, personally i would like some kind of mechanism
to
>> transfer power and signal from neuron to neuron. Maybe you could have the
>> boards and use some kind of 3 core cable that clicks onto some headers,
it
>> would add to the cost but would be very helpful.
>>
>> >Plus, the fact that if you need a socketable DIP circuit (which is
>> >smaller as well), you merely cut off the portions of the hextile that
>> >aren't necessary for the DIP package! So simple!
>>
>> I wouldn't want to cut it out in case i had to put it back in, i'd much
>> rather have it in a DIP socket.
>>
>> >An "x" is a configuration point. There are configuration points to
>> >configure the Hextile for Nv and Nu opperation, and to disable the LED.
>> >You cut the point with an x-acto knife or something and this configures
it.
>>
>> I don't like the idea of cutting this and cutting that, i'd rather use
>> jumpers or even solder pad that can be linked and then desoldered
>>
>> Don't get me wrong, i love the design, and if i wasn't developing it i'd
buy
>> it.
>> Perhaps another compromise using headers and optional cables is in order.
>>
>> David
>
>--
>
>
>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

Home