The whole idea of the newly structured Ethernet Node counting system
and increase in Connections for the newer Logix controllers was to make it simpler to configure and to eliminate the possibility of running out of Connections too quickly as a project grows. This happens all too often with the older CompactLogix controllers.
For the older Ethernet CompactLogix controllers (L32E and L35E) they are very restricted. They only support a maximum of 32 CIP Connections. Imagine how quickly some I/O Nodes, which can use double figure Connection counts, will eat up Connections on these controllers.
Also, this limit of 32 is further restricted by the RPI set per Connection. The higher the RPI the lower the supported maximum CIP Connections becomes. For high speed applications, this can absolutely murder the limit. If you have 2 Connections set to 2ms RPI each, then those 2 CIP Connection are all that are allowed for the controller. This would leave nothing for RSLinx Classic communications, a HMI, or the like.
However, if all Connections are set to 35ms RPI or higher, then the full 32 CIP Connections will be available. But with these controllers you always have to be careful to leave enough spare Connections for programming workstations and HMI. Each RSLinx Classic driver instance uses 1 Connection (not Node). Each PanelView HMI can use up to 5 Connections (not Nodes).
The new CompactLogix controllers now have Ethernet Node limits, which does somewhat restrict the number of devices you can add to the I/O tree, but this is relative to the controllers intended application market. They now restrict the number of Nodes allowed, in the software, so as not to saturate the specific controller with I/O Connections. This leaves plenty of overhead with the remaining Connections for non I/O tree devices. i.e. There should be plenty of Connections to go around for "non I/O" communications, such as RSLinx Classic.
Ethernet Nodes and devices using Connections are one and the same. They all use Connections. The difference now being that they have a much greater overall Connection count available to them (256), compared to their previous counterparts (32). This allows a much larger scope for new and temporary devices to be simultaneously provisioned onto the controller's network.
The feature, of restricting Ethernet Nodes per controller, is basically one intended to limit, or reserve, a portion of the overall Connections for I/O communications. The Node limit, while fixed, does not dictate the final CIP Connections used for all of the I/O Node devices added to the I/O tree. The type of devices added will determine this.
Example:
If we add 3 PowerFlex 40 drives as Nodes to an L16ER I/O tree, using 22-COMM-E adapters, they can use up to 16 CIP Connections each. So that's a possible 48 CIP Connections in total.
If we add a 1734-AENTR as a 4th Node, it can use up to 20 CIP Connections.
So now, for the 4 allowed Nodes, we have a potential 68 CIP Connections used off the total of 256 for the controller, leaving 188 Connections free for non I/O devices. This should be more than adequate for most applications.
It is all about counting Connections at the end of the day. Ethernet "Nodes" is just a new buzzword for the limitation in Logix Designer for the I/O Configuration in these controllers. Once there are plenty of Connections available, then using RSLinx Classic, Explicit Messaging, HMI communications, peripheral devices such as barcode readers and printers, should not be such an issue, as it was in the past.
Rockwell Automation said:
IMPORTANT - With the increased abilities of the CompactLogix 5370 controllers, where it should no longer be of concern, it is still recommended that the control system limits should still be checked and verified for each specific application.
What they are basically saying here is that once you have selected the correct controller for your application, and because they are now much more capable, you should not have to worry about Connections any time soon.
If your selected controller's I/O Node count seems "tight" then you move up in specification.
Another improvement made on the new CompactLogix controllers over the old is the maximum Packet Rate for the Ethernet port. The old controllers can achieve up to 4000 PPS (Packets Per Second) whereas the new can achieve up to 6000 PPS. Also, for Unconnected Messages, the old allow 32 across the backplane and 32 out to the Ethernet network. The new allow a total of 400 for everything.
GaryS,
I'm always open to learning new things and I do not presume to know everything on a given subject, but I am seriously doubting the advice given by Rockwell here as being credible?
Perhaps the Support Engineer was misinformed or something, I don't know?
I'm still open to being further enlightened here, but so far I cannot see that...
...the Ethernet Node count...
...which is
only based on the devices added to the I/O tree...
...of which a "computer", "RSLinx Classic" or "RSLinx Enterprise" are not...
...can be reduced by 2 Nodes...
...on a controller spec'd for 4 Nodes...
...that is half gone already...
...just by communicating from RSLinx Whatever...
Why would they only spec it for 4 Nodes if such a trivial thing can eat up half its available Nodes before you even start to build the I/O Configuration?
Also, this Node limitation is in the software, not the controller's firmware. When you try to Verify or Download a project with an exceeded I/O Node count it will error with...
The maximum number of nodes on the local Ethernet port has been exceeded - reduce the number of Ethernet nodes
When a valid project is downloaded to the controller and running, all the controller knows about is Connections. It is not counting Nodes. This is purely done back in the software before the Download. Regardless of whether in the I/O tree or not, the Controller keeps opening Connection requests until it has no more. If the limit is exceeded then the controller will not be able to open any more requested Connections until some are freed up.
With so many Connections now available to the new Logix controllers, and with the added restriction on the amount of devices that can be added to the I/O tree, preventing the Connection total from being I/O saturated, there should be more than enough Connections to go around, if you have selected the right controller for the job.
What ever was wrong in your case above (you didn't say how you fixed it?), I would be very surprised if it had anything do with a programming workstation using up "Nodes".
Regards,
George