Kinetix 6500 Ethernet/IP communication issues

Hello, we currently have a Kinetix 6500 application with a Controllogix 72 series processor system. There are 7 Servo Drives connected in a linear topology to a 1756-EN2T (rev 4.3) module (Slot 2). In addition, there are 9 Powerflex 40 VFD's, 3 x FlexIO Enet modules, 1 x Generic Touch Screen, 1 x Murr Cube 20 on the Control network to a 1756-ENBT (Slot 0).

Both networks are separate, but with identical subnets (no IP overlap of course).

The problem we are having is that as soon as we connect to the Servo motors, all communications becomes painfully slow. Additionally, when we activate the Kinetix system we seem to lose communication consistency with our non-servo network, which is causing some timing of vfd accel / decel curves to vary between cycles.

We have adjusted the percentage of communications time up significantly, but it is getting to a limiting point.

All network switches are managed with IGMP snooping enabled.

Do both networks use the same switch at any point? Without VLAN's this would be your issue.

Rant On / Also IMHO it's bad design practice to have more than 1 network with the same subnet schema on the same process line or piece of said equipment. In my book this is one of those things that just because you can it still does not make it a good idea./Rant Off

The first three tools I recommend are the embedded Web pages in the 1756-EN2T and -ENBT, and the ControlLogix Task Monitor utility.

The web pages will help you get an idea of the CPU utilization and the packets-per-second for each Ethernet module separately, and you can see if these change when both networks are connected.

The ControlLogix Task Monitor will let you make similar measurements for the ControlLogix CPU.

The web pages will also let you check out the error counters; if there is a media or grounding problem, you need to fix that first.

The system as described doesn't seem overloaded, but I really can't tell without details like RPI values.

A couple words about PowerFlex 4-series on EtherNet; the 19200-baud 'DSI' serial interface between the network module and the drive is much slower than the PowerFlex 7-series CAN-based 'DPI' interface.

I got deeply into this a few years ago on a system that had about 50 PowerFlex 40 drives on EtherNet/IP.

We measured that the time between when an EtherNet/IP frame with the 'Run' command set true arrived and when an EtherNet/IP frame with the 'Active' bit set true was produced by the drive. This was a consistent 70 milliseconds.

So the fastest RPI I use for PowerFlex 4-series drives in 35 milliseconds. Anything faster is just wasting bandwidth.

Also check your start/stop logic on the PowerFlex drives. I've seen a lot of logic that holds 'Stop' true whenever 'Start' is false, instead of setting 'Stop' back to false once the drive has been verified to be stopped. This is correct from a pure logic point of view, but because ControlLogix can produce output data to the network card at any point during the logic scan, it's possible to accidentally send out both the Stop and Start bits true at the same time, which will result in a 'the drive was commanded to start but did not start' sort of problem.

sorry for delay in response

Hey, I just wanted to say, thanks for the quick response, I've been checking my phone for emails showing I had a response, and finally realized that I probably was missing the emails for some reason. I'll respond with some quick answers :

Do both networks use the same switch at any point? Without VLAN's this would be your issue.

No. The Servos are direct to the EN2T card. There is only Kinetix 6500's wired in Lineal connection on this port.


The first three tools I recommend are the embedded Web pages in the 1756-EN2T and -ENBT, and the ControlLogix Task Monitor utility.

The web pages will help you get an idea of the CPU utilization and the packets-per-second for each Ethernet module separately, and you can see if these change when both networks are connected.

The ControlLogix Task Monitor will let you make similar measurements for the ControlLogix CPU.

The web pages will also let you check out the error counters; if there is a media or grounding problem, you need to fix that first.

The system as described doesn't seem overloaded, but I really can't tell without details like RPI values


Thank you. There is definitely a notable change in response when the servos are connected vs when they are disconnected, even in just Saving the Program / Uploading to the Laptop, etc.

The RPI values are not adjustable for each of the cards (defaulted to 500ms) in the PLC.

I have to collect the RPI values on the other items here shortly.

There do not appear to be any errors on either of the two network cards status monitor in RSLogix.

I will check the RPI settings on the Powerflex 40's thanks for that, and thanks for the Start command suggestion. We've started a switch to two / three wire control on the drives (from unfortunately a Ethernet/IP FlexIO module beside the drive, so i'm not sure that makes a significant improvement on communication latency issues and stopping inaccuracy, but we'll give it a shot.).

Hello,
I had some problem to you.
I have a 1756 L72S plc and a 1756 EN2TR Ethernet card with 7 Kinetikx 6500.
Sometime one of this 7 drives fault with communication lost.
I solve the problem with firmware upgrade to release 2.18

By

What is the coarse update period for the servo motion group? If this is set too low, it will slow everything down - especially laptop communications of getting online and saving.

If you have 7 servos, try using an update rate of 7msec or greater. See if it makes a difference.

I have only 6msec but now with the new firmware (2.18) everything seems to be going well.