> +->Rx.PC1.Tx---->Rx.PC2.Tx--->Rx.PC3.Tx--->Rx.PC4.TX---...-->Rx.PC36.Tx--+
> |                                                                        |
> |                                                                        V
> +------------<--------------------<-------------------<------------------+
>
> Congratulations!  You've reinvented the ring network.

That should work, but you've got to be really careful about ground loops and
think about noise.

Here's a slightly different solution to this problem.
I've got a current project that includes connecting up to a few hundred
advertising signs to be driven from a single COM port.  I used a 17C PIC
with two UARTs as a communications controller on each sign.  Each unit has
an upstream and downstream port, with the upstream port of the first unit
connected to the PC COM port.  The protocol has been designed to support
nodes with more than one downstream port (tree structure instead of linear
chain), but so far we have no hardware that does this.  All communication is
wrapped in packets with checksums, and use ACKs and retries to get reliable
delivery.  At wakeup, the host software assigns each node a unique address
based on the topology.  It can then send messages to specific nodes or do
broadcasts.  So far we've tested this with ten nodes and all is working very
well.

> I supposed that it would be worthwhile to read up on the history of token
> ring networks (like - why do they need a token in the first place?)

The short answer is so that only one node at a time is allowed to inject a
message onto the ring.  The other nodes are only allowed to pass on the bits
from upstream to downstream.


********************************************************************
Olin Lathrop, embedded systems consultant in Littleton Massachusetts
(978) 742-9014, olin@embedinc.com, http://www.embedinc.com

--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu