Email: Password: Remember Me | Create Account (Free)

Back to Subject List

Old thread has been locked -- no new posts accepted in this thread
???
11/04/08 16:50
Read: times


 
#159647 - Token-passing requires explicit or arbitrated mast
Responding to: ???'s previous message
I have never said that you should run multiple masters without protocol synchronization.

But you do not get multiple masters to be synchronized unless they are asymetrical in some way. And the time to synchronize nodes is affected by the frequency of the arbitration loop.

If you turn on 10 identical nodes, you have to figure out what nodes are on the cable, and allocate an initial "token" for the synchronization. If they start without any address in them, they would all start to send a message (without hearing that the message gets destroyed by everyone else). They would then all wait for an answer, fail and try again. Having a serial number in each unit to affect random delays do help but only on a small scale unless the delays are long in comparison to a message transmission.

Collission detect capabilities - while not perfect - can greatly help to speed up this arbitration phase.

Being able to hear a collision means that you can let the serial number (or noise on an ADC channel) create quite short delays where you have a good chance of establishing a token owner within a limited number of character send times. Then you can continue to pick up the remaining nodes and one-by-one and assign each node a unique address and/or time-slot for the following communication.

And being able to hear collisions helps a node to figure out if it is broken - or alone - on a cable.

Don't be so convinced that people have to shoot themselves in the foot.

List of 24 messages in thread
TopicAuthorDate
Interfacing RS485 (SN75176) with AT89c51            01/01/70 00:00      
   Select your components with care            01/01/70 00:00      
   Compnents are irreplacable !!            01/01/70 00:00      
      What do you miss in the datasheet            01/01/70 00:00      
      It is a simple chip            01/01/70 00:00      
   do not do rs485 without ...            01/01/70 00:00      
   Here's the info for starting            01/01/70 00:00      
   Controlling the sn75176!!            01/01/70 00:00      
      use the \"industrial strangth\" LT1785 instead            01/01/70 00:00      
         Easy to control the RS-485 transeciever            01/01/70 00:00      
            this is one of the biggest mistakes in RS485            01/01/70 00:00      
               Not so much specific for RS-485            01/01/70 00:00      
               might be???            01/01/70 00:00      
                  These are NOT RS485            01/01/70 00:00      
                     Lack of dominant bit state            01/01/70 00:00      
                        that's exactly why on RS485...            01/01/70 00:00      
                           Token-passing requires explicit or arbitrated mast            01/01/70 00:00      
                        the amazing thing is ...            01/01/70 00:00      
                           ... that it matter            01/01/70 00:00      
                              I am NOT talking about multiple masters, ...            01/01/70 00:00      
      follow the link            01/01/70 00:00      
         the two REAL 485 sources            01/01/70 00:00      
            Preview?            01/01/70 00:00      
               I do not know why            01/01/70 00:00      

Back to Subject List