Modbus Dual Ethernet Communications

[Anonymous user]

📖 Home  Back  
Modbus can be configured such that two communications paths to a PLC can be configured on a "switched" scheme, meaning that all of the attached scanners (on a channel) will either use the first configured path or the second.

The driver architecture that the Modbus driver is built on uses one thread per channel. Scanners attached to a channel are scanned in sequence, one scan must end before the next can start. Even if the Modbus scanners on a single channel in the configuration had the connections spread between addresses one and two, the scans would still be synchronized, this meaning that even though requests are being shared between two connections, the overall time it takes to complete all the scans would be unchanged. The network usage would look something like this:

'#' means in use'.' means idle1: ##...#..##...2: ..###.##..###

Bandwidth is shared between the two links, but overall Modbus scan performance is not improved.

To improve the performance and throughput to a PLC, there are a couple of things that can be done. The first option involves using multiple channels. This immediately allows multiple concurrent connections to the PLC allowing more data to be transferred between the SCADA and PLC.

To allow load sharing on the network, for every second channel to a PLC, switch the primary and secondary IP addresses around on these scanners, so that address 1 becomes address 2. This way one channels primary address would be on one link, the other on the other link. So by default, the scanning load would be shared between the two networks. On link failure, all of the scanners would switch onto the same link, it would then be a manual exercise (or a FBD or ST Logic program to perform this automatically) to switch the channel back to primary when the network issues have been resolved.

Using 2 channels as described above will give the following:

Channel 1 path A (Address 1): ##########Channel 1 path B (Address 2): ..........Channel 2 path A (Address 2): ##########Channel 2 path B (Address 1): ..........

You can have scanners for different PLC's on one channel. A channel will make one TCP connection for every unique address configured in it's scanners, scanners with the same address sharing the same connection.

To get multiple connections to the same PLC, split the scanners between multiple channels. This will give one connection per PLC per channel. The more channels you have, the more connections you get.

Path switching is determined by how many scanners on a channel have failed. This is configurable in the channel as "Switch after n scanners have failed", zero meaning that only switch if all scanners fail. Switching doesn't care what IP addresses the scanners are connecting to, it just goes from A to B (or back) when N have failed. The idea was that N PLC's may be configured on just two physical networks, and if one network fails, all the PLC's need to be switched to the other network. Finer granularity of switching can be achieved by introducing more channels and splitting the PLC's between them.

Automatic switching will only occur if at least one of the paths was determined "healthy". This is determined as "More scanners were polled successfully than the "Switch after n scanners have failed" of the channel". This is to prevent the link continuously switching between the two links.

Go: Home Back