Newsgroups: comp.robotics
Path: brunix!uunet!think.com!sdd.hp.com!ux1.cso.uiuc.edu!news.cso.uiuc.edu!ehsn5.cen.uiuc.edu!rg9538
From: rg9538@ehsn5.cen.uiuc.edu (Ronald E Garnett)
Subject: Re: 68HC11EVBU fanout data
References: <1992Oct12.163640.25052@hemlock.cray.com>
Message-ID: <Bw1p0J.42F@news.cso.uiuc.edu>
Sender: usenet@news.cso.uiuc.edu (Net Noise owner)
Organization: University of Illinois at Urbana
Date: Tue, 13 Oct 1992 05:38:42 GMT
Lines: 65

kilian@cray.com (Alan Kilian) writes:


>rg9538@ehsn8.cen.uiuc.edu (Ronald E Garnett) asks:
>> Does anyone out there in net land know how many SLAVE units that the 68HC11
>> EVBU board can drive from a master.

>The synchronous connections are generally daisy chained so that each CPUs
>output only drives one other CPUs input. This way you can connect an infinite
>number of CPUs together. Like this:

>probably be O.K. These things have CMOS inputs so it doesn't take too much
>current to get them to switch. This and the fact that the SPI does not
>switch THAT fast means that one CMOS output can drive several CMOS inputs
>just fine. I'd say ten would be O.K. The problem with this version is
>that it is only a one-way connection. You cannot connect the ten other CPU's
>output lines back to the one CPU's input line. That would be bad news.

>The asynchronous (SCI) channel should be able to drive several other CPUs
>inputs the same as the synchronous (SPI) channel can. You have the same 
>problem with it being a one-way connection. You cannot connect several
>CPU's Transmit lines to one CPU's recieve line.


I tried posting an intelligent response to this, but I'm logging in from home,
and the computer ate my post!!   I'll try again in less detail!


The task which I am trying to do is create an expandable network of HC11's
to be used for control (robotics) projects for which a single processor
just won't hack it.  I would like to have the ability to synchronize (not
by clocking by the same crystal) each processor at the same time.  I was
hoping to use a bus structure to accomplish this.  Restraints on the 
multi-controller network are as follows:

	1)  Simplicityto the programmer is a key.  (So that users of this
	system won't have to get too heavily into the communications 
	between nodes.  I will put as much as possible in ROM and write
	simple (to use) subroutines for the programmers.

	2)  Easy to expand for difficult projects, easy to contract for
	simpler projects.

	3)  Minimum use of ports, pins, memory, timeslices, etc.  In other
	words, low cost in terms of the 68HC11's resources.

	4)  A rough similarity to living organisms' nerve systems is desired.
	each bus of controllers could be a seperate subsystem such as an arm
	or leg or whatever.  The seperate buses could be connected to form
	a body.  Possibly with a "spinal cord" and a "brain" when needed.

The limits on the chain are:  not as easy to globally synchronize each 
controller, longer delay (tied to first one).

If a bus gets used, I will probably connect the MOSI and MISO outputs together
using a bus arbitration scheme and open drain outputs on port D (SCI).
But tif the number of HC11's I can connect iis only two, then I have gain       nothing!  10 might be acceptable....

Ron Garnett,
Grad student in Electrical Engineering (M.S. program)
University of Illinois, Champaign-Urbana

rg9538@ewshp.cen.uiuc.edu

.
