The ring configuration has other characteristics that most don't realize right away. You can actually move more traffic on a ring than you can on a hub/peer contention configuration. Most at first though would think that wrong.

To do a ring with a differential signal pair, the ring costs you an additional transceiver since there is no separate TX and RX signal. If your application warrented it, you could survive a single cable cut or lockedup node by using a reverse direction protocol. It would be similar to SONET rings in telecom in that when an obstruction is discovered, the traffic starts flowing in the other direction. You could do this via message protocol and not require the second loop.

Doing a loop would be fun but I don't know what the ideal application would be around an 8051. It should be in a highly servicable environment, programmed to shovel messages on the fly (no full buffering first) and require frequent messages and many nodes.

With a contention hub/peer, the more nodes, the more nodes must wait. A loop increases speed because several messages can be in route at the same time and no contention is used. You'd be adivsed to simulate expected traffic flows before chosing rings though because at some point, chatter begins to get omnipresent. You can't rely upon strategic wiring to shorten paths, better to design in sub loops.

I wish I could think of a good application to do this with. It could be fun.

aka J