| ??? 02/05/08 01:31 Read: times |
#150340 - Common solutions Responding to: ???'s previous message |
Multimaster RS485 presents the problem of multiple masters trying to send at one time as we all know. If we look at ethernet which implements carrier sense and collision detection - we can detect transitions on the bus to see if someone is sending (remember two of more masters might do this at the same time and even having a bus busy signal doesn't avoid collisions) and we can receive the data we send to detect a collision. Physically, a collision isn't much of an issue - the RS485 bus drivers should have current limiting - in fact I've seen schemes where the current is sensed to detect a collision. Then the issue is what to do in the event of a collision - ethernet implements a binary backoff algorithm so the colliding nodes back off for a random amount of time the idea being that when they retry, that they don't retry at the same time. To implement this adds a lot of complexity. The other method is use token passing where a token packet is passed from master to master. Upon receiving the token, the master either passes the token on or sends its data. Failing that, use ethernet instead of RS485 - its getting cheaper ! |
| Topic | Author | Date |
| Multi Master communication RS-485 ? | 01/01/70 00:00 | |
| how fast do you need to go? | 01/01/70 00:00 | |
| Use another cable pair - for bus busy | 01/01/70 00:00 | |
| that will not work | 01/01/70 00:00 | |
| How's about using with MAX1490 | 01/01/70 00:00 | |
| what good will that do ? | 01/01/70 00:00 | |
| Common solutions | 01/01/70 00:00 | |
| Multi master my way | 01/01/70 00:00 | |
| Still on progress | 01/01/70 00:00 | |
| wrong chip? | 01/01/70 00:00 | |
| Maybe Token Ring | 01/01/70 00:00 | |
| ARCNET | 01/01/70 00:00 | |
| re:multimaster | 01/01/70 00:00 | |
| what if two masters send 'busy' at the same time? | 01/01/70 00:00 | |
| re:collision | 01/01/70 00:00 | |
| all good and well; however ... | 01/01/70 00:00 | |
| separate the masters for network | 01/01/70 00:00 | |
sure, but the problems are the same | 01/01/70 00:00 |



