When using a Trio Q Series full duplex radio (eg a QB450 or QH450) as a repeater, the radio will receive the entire packet before transmitting it back out. All Ethernet packets are transmitted intact, and processed intact. This ensures highest packet integrity, and allows for routing of messages when necessary.
Now you’re probably going to say “Well, if that is the case, what is the advantage of a full duplex repeater, if it can’t transmit what it is receiving instantly?” (as does our older EB450 full duplex repeater, providing very low latency) The answer lies in the nature of Ethernet traffic.
The concern is the possible number of different applications that may be trying to access the channel, without any synchronisation between them. Even in the most basic Point to MultiPoint setup you might think all that is happening is polling of devices by the SCADA protocol, but look at the traffic with Wireshark and you may be surprised.
There may be one application polling devices via (eg) DNP3, Windows maintaining an ARP cache and deciding at random 15 to 45 second times to refresh the cache, and then add some SNMP diagnostics or Telnet browsing and you have a lot of different devices trying to access the channel.
The need for effective collision avoidance (requiring a full duplex ChannelShare+ Master) is much higher than on a serial system. The full duplex nature of a QB450 or QH450 Entry Point or Repeater allows it to both receive a message and at the same time transmit a Channel Busy flag when necessary.
There are several separate things to consider in the following scenarios:
(a) We are talking about radio system topologies that are two frequency (simplex systems can only make use of Carrier Detect collision avoidance)
(b) You can only enable “Collision Avoidance Master” in full duplex radios
(d) In a PTMP system, the entry point is a collision avoidance master
(d) In a PTMP via Repeater system, the repeater is a collision avoidance master
(e) If you want to use ChannelShare+ collision avoidance, you MUST have a collision avoidance Master
(f) If you have a two frequency system but only half duplex radios (no full duplex Entry Point or Repeater), you can only operate in Carrier Detect collision avoidance mode
Here are a few examples of the operation of ChannelShare+ mode, describing the activity of the Master radio and a typical Remote:
Scenario A:
(a) CS+ Master: Has data to transmit, but Rx channel is clear (mute not broken) -> Data is transmitted but the Channel Busy flag is clear (not set)
(b) Remote: Mute breaks as CS+ Master keys up. Remote initially assumes channel is busy, but once the data stream starts it sees the Channel Busy flag is clear. Remote declares channel is clear and that it is free to transmit.
Scenario B:
(a) CS+ Master: Has data to transmit, and its Rx channel is busy (mute broken) -> Data is transmitted and the Channel Busy flag is set
(b) Remote: Mute breaks as CS+ Master keys up. Remote initially assumes channel is busy and backs off. Once the data stream starts, it sees the Channel Busy flag is set. Remote declares channel is busy and continues to back off. Remote waits until Channel Busy flag is clear or its Mute re-asserts. (no signal heard)
Scenario C:
(a) CS+ Master: Has data to transmit, and Rx channel starts off clear (Mute not broken). In middle of transmission, Mute breaks. -> Data is transmitted and Channel Busy flag is initially clear, but when Mute breaks the Channel Busy flag is set.
(b) Remote: Mute breaks as CS+ Master keys up. Remote assumes channel is busy, but once data stream starts, it sees the channel busy flag is clear. Remote initially declares channel is clear and will transmit if it has data to send. When after a short time Remote detects that Channel Busy flag has now been set, it declares channel is busy and backs off.
The time taken for the CS+ Master to detect its mute breaking and set the Channel Busy flag is approx. 5 mS. This time is known as the collision window. During this brief time it is possible for collisions to occur.
Now you’re probably going to say “Well, if that is the case, what is the advantage of a full duplex repeater, if it can’t transmit what it is receiving instantly?” (as does our older EB450 full duplex repeater, providing very low latency) The answer lies in the nature of Ethernet traffic.
The concern is the possible number of different applications that may be trying to access the channel, without any synchronisation between them. Even in the most basic Point to MultiPoint setup you might think all that is happening is polling of devices by the SCADA protocol, but look at the traffic with Wireshark and you may be surprised.
There may be one application polling devices via (eg) DNP3, Windows maintaining an ARP cache and deciding at random 15 to 45 second times to refresh the cache, and then add some SNMP diagnostics or Telnet browsing and you have a lot of different devices trying to access the channel.
The need for effective collision avoidance (requiring a full duplex ChannelShare+ Master) is much higher than on a serial system. The full duplex nature of a QB450 or QH450 Entry Point or Repeater allows it to both receive a message and at the same time transmit a Channel Busy flag when necessary.
There are several separate things to consider in the following scenarios:
(a) We are talking about radio system topologies that are two frequency (simplex systems can only make use of Carrier Detect collision avoidance)
(b) You can only enable “Collision Avoidance Master” in full duplex radios
(d) In a PTMP system, the entry point is a collision avoidance master
(d) In a PTMP via Repeater system, the repeater is a collision avoidance master
(e) If you want to use ChannelShare+ collision avoidance, you MUST have a collision avoidance Master
(f) If you have a two frequency system but only half duplex radios (no full duplex Entry Point or Repeater), you can only operate in Carrier Detect collision avoidance mode
Here are a few examples of the operation of ChannelShare+ mode, describing the activity of the Master radio and a typical Remote:
Scenario A:
(a) CS+ Master: Has data to transmit, but Rx channel is clear (mute not broken) -> Data is transmitted but the Channel Busy flag is clear (not set)
(b) Remote: Mute breaks as CS+ Master keys up. Remote initially assumes channel is busy, but once the data stream starts it sees the Channel Busy flag is clear. Remote declares channel is clear and that it is free to transmit.
Scenario B:
(a) CS+ Master: Has data to transmit, and its Rx channel is busy (mute broken) -> Data is transmitted and the Channel Busy flag is set
(b) Remote: Mute breaks as CS+ Master keys up. Remote initially assumes channel is busy and backs off. Once the data stream starts, it sees the Channel Busy flag is set. Remote declares channel is busy and continues to back off. Remote waits until Channel Busy flag is clear or its Mute re-asserts. (no signal heard)
Scenario C:
(a) CS+ Master: Has data to transmit, and Rx channel starts off clear (Mute not broken). In middle of transmission, Mute breaks. -> Data is transmitted and Channel Busy flag is initially clear, but when Mute breaks the Channel Busy flag is set.
(b) Remote: Mute breaks as CS+ Master keys up. Remote assumes channel is busy, but once data stream starts, it sees the channel busy flag is clear. Remote initially declares channel is clear and will transmit if it has data to send. When after a short time Remote detects that Channel Busy flag has now been set, it declares channel is busy and backs off.
The time taken for the CS+ Master to detect its mute breaking and set the Channel Busy flag is approx. 5 mS. This time is known as the collision window. During this brief time it is possible for collisions to occur.
Released for:Schneider Electric Canada
