Why does the turnaround time matter that much? If you connect together two locations with fiber, aren't both RS-485 networks treated as separate networks with the fiber transceivers acting as a node on the local wired network and just acting as a relay in between?
My RS485 is a bit rusty. but I'll take you here for a moment.
http://www.bb-elec.com/tech_articles/sendd...llustration.asp There is a lot of information concerning how Rs485 operates.
I'm going my memory. It's a master-slave type of protocol and transmit and receive can take place over the same two wires, or actually in some cases 4 wires if the receivers are differential 485 which in most cases they are. So it's kind of like a one way street multi-drop RS-422 kind of connection. The R422 takes care of the ground loops etc.
Somewhere at the end of the bus there is this elusive termonator. It just reduces reflections because it has the same characteristic impeadance of the cable.
OK, now what. We can hang about 32 (don't take my word for it) devices on this bus and one master.
So, they all are in a High Z state listening, so everybody is listening at the same time.
The Master speaks and then someone says, that's me and decides to talk. Everybody hears, but only the master does something. Now how does the bus reverse direction? Ahh, say 3 character times of silence and everyone's back listening again.
Again, details may not be right. It's just to give you a sense of how the protocal operates.
Flow control can be a bit tricky and hardware flow is a possibility. Hardware flow control over fiber is a possibility.
So, good data requires:
1. A slow speed
2. If you up the speed, you need good noise immunity (twisted pair shielded cables)
3. You want better data and slow processors, incorporate hardware flow.
So, if there is significant latency in getting the data, some the device may miss some data.