🚀 go-pugleaf

Greetings!  We are going into the development stage of a
LAN-based motion microcontroller.  The system will be a single-board
combination of an 80386SL CPU and an Analog Devices' 2105 DSP.

The controller will be communicated to via a high-speed network
connection.  My question is:  What system (both hardware and
software) would be best?  Some requirements include:

  o  Relatively 'dumb' nodes.  Since the CPU will be in charge
     of each microcontroller, and can require real-time response
     to (possibly) a *bunch* of stimuli, we can't devote a lot
     of CPU resources to network communication.  What systems
     require lots of processor overhead; and which ones are
     relatively automatic?  Furthermore, because several robots
     may need to coordinated their actions, nodes would have
     to be able to communicate with each other, without host
     intervention.

  o  Large host availablity.  We need to choose a system which
     can be communicated with by any of a bunch of host systems.
     Often times, PCs, Sun workstations, or other systems may
     be the most desirable hosts.  Fortunately, almost all hosts
     will be PCs, but we don't want that restriction.

  o  Bandwidth (of course).  Sometimes a machine will be in 'close'
     contact with something else.  Or a network may before long
     wind up with several hundred nodes.  This is an impossible
     question to answer, but after how long would we reach bandwidth
     limits?  I say impossible because it's entirely a function of
     our customers' programs.

  o  Real estate.  Board space is important, so the hardware involved
     needs to be minimal.


Whew!  I've just described the ideal network...wouldn't that be nice?
Anyway, I've investigated the LON system (Eschelon(sp?)/Motorola).  It
seems ideal...except for the development cost.  Does anyone have experience
with it?  Does anyone know of alternative ways to develop with it?  (I am
imagining renting a development system, or hiring a consultant;  both
these must be less expensive.)    Are there any other industrial standards
already in existence?  Are we just off on the wrong track altogether?

Please e-mail your responses, and thanks!
-patrick