EtherNet IP on Positioning Devices

Just curious how people are making out with their applications using EtherNet/IP for motion control. Any issues with Latencey and Lost Packets or Wrong Position Information???

What about laser distance measuremetn for positioning applications in Automated Storage Retrieval Systems?

Cheers,
Chris

Few if any problems

Our customers the Control Logix and Ethernet/IP with our controller. There have been some traffic problems when there are a lot of PLCs and motion controllers all on the same net. Smart switches take care of this problem.
Whole sawmills are tied together with Ethernet/IP.
By default the Ethernet/IP updates every 10ms. This is much faster than required in our case because our controllers can operate autonomously except to get the position data for evert piece of wood.


Wrong positions are simply a programming problem. Not an Ethernet/IP problem. The packet gets there with a checksum or doesn't. There are no packets that get there screwed up.

Few if any problems

Our customers the Control Logix and Ethernet/IP with our controller. There have been some traffic problems when there are a lot of PLCs and motion controllers all on the same net. Smart switches take care of this problem.
Whole sawmills are tied together with Ethernet/IP.
By default the Ethernet/IP updates every 10ms. This is much faster than required in our case because our controllers can operate autonomously except to get the position data for evert piece of wood.

Wrong positions are simply a programming problem. Not an Ethernet/IP problem. The packet gets there with a checksum or doesn't. There are no packets that get there screwed up.

Hey Peter,

Thanks! I figured as much. I was just participating in a seminar on the EtherNet communication protocols and there was a lot of (Heated Discussion) on the issues with latency and lost packets and collisions and such... I figured that with Smart Switches and the shear speed at which information is communicated that this would be a relatively mute point.

BUT... alas... there are those who are still hesitant.

Cheers,
Chris

I don't mean to speak for Peter but keep in mind he is just referring to sending position commands and other high-level information. He is not talking about servo loop closure or transport of feedback information. EtherCat can pull this off. Ethernet/IP cannot.

Keith

Why will ethernet ip not do this?

Is ethercat faster? Or is it in the way it handles packets?

We are evaluating EtherCat

kamenges said:

I don't mean to speak for Peter but keep in mind he is just referring to sending position commands and other high-level information. He is not talking about servo loop closure or transport of feedback information. EtherCat can pull this off. Ethernet/IP cannot.

Keith

Click to expand...

We still wouldn't close the loop over EtherCat in a motion control system. There is too much jitter. We looked at using Ethercat for a back plane, that back plane is not high enough. It still looks good for an alternative to Sercos.

EtherNet/IP is good for communicating between modules and it is fast and reliable when the network is properly setup. It is not fast enough to do closed loop motion control. The key is to have isochronous and deterministic data transfers and this requires a special FPGA before the normal Ethernet chip. The Ethernet/IP does not have this special FPGA that provides an isochronous and deterministic data transfer. Also, EthernetCat uses a daisy chain which avoids the collisions that may occur in switches and the EtherCat packets are raw Ethernet packets. There is no IP address, IP layer etc. That is what makes it deterministic. This is very close to the way I would do it. Now if EtherCat only had an option to got to 1GBs.

Making this extra FPGA is an extra expense. The dumb devices need only this special FPGA.

Thanks Keith... I am looking for good converstaion on this topic.

PLC Kid... EtherCAT isn't faster it's how it uses a read and write on the fly principle... so a packet is created and transmitted from a master (controller) and as the packet is processed by each slave (device) on the network information is read and written to the packet... then it is sent to the next device on the network... kinda like the old Token Ring process... if you're old enough to remember those days... if not... wiki has lots of good pictures and info.

Great info Peter... Thanks Again!

Peter Nachtwey said:

Now if EtherCat only had an option to got to 1GBs.

Click to expand...

+100. Once again, I could only wish there was the same kind of feedback with Beckhoff designers as we have here with the Red Lions'...

CBowman said:

Thanks Keith... I am looking for good converstaion on this topic.

PLC Kid... EtherCAT isn't faster it's how it uses a read and write on the fly principle... so a packet is created and transmitted from a master (controller) and as the packet is processed by each slave (device) on the network information is read and written to the packet... then it is sent to the next device on the network... kinda like the old Token Ring process... if you're old enough to remember those days... if not... wiki has lots of good pictures and info.

Click to expand...

CBowman

I am not old enough to remember those as i am only 18 but thanks for the info i will check out the wiki's.

I try to learn about the older stuff as i feel like it gives me a better understanding all around. To me i can understand the newer technologies better if i know where the came from and what they are replacing so to speak.

I having been teaching myself slc 150 and plc 2 programing we only havve a couple of those running but i just wanted to understand the system so i play with them in the shop when i can.

PLC Kid... that is awesome!!! I almost remember being that age... and if I recall correctly... I was trying to put a MicroLogix into a Mustang to control the NeoN lights... OMG... that was a little bit ago! hahaha

Keep the learning alive!

Originally posted by Peter Nachtwey:

We still wouldn't close the loop over EtherCat in a motion control system. There is too much jitter. We looked at using Ethercat for a back plane, that back plane is not high enough. It still looks good for an alternative to Sercos.

Click to expand...

Just out of curiousity, was the jitter inside the published spec or don't they have a published spec for jitter? I have to assume one or the other or you guys wouldn't have even tried in the first place.

The guys I used to work for are currently doing a rotary knife (my kind for paper, not your kind for metal) using EtherCat for feedback and drive current commands. They seem to be pretty happy so far. My guess is the issues that jitter is causing them are currently less of a problem than pushing a wet noodle with wings forward at 1200 FPM.

Keith

Upcoming Webcast: Deterministic Ethernet for Distributed Motion Control

EtherCat sure seems to be an up and coming network. I've already signed up for this webcast.

If you think that using Ethernet for motion control is a recipe for a non-functional system, then this Webcast is just what you need.
Engineers have recognized that basic, commercial Ethernet is not inherently deterministic. Solutions have been developed, and EtherCat is a predominant one. Speakers on this Webcast will discuss what constitutes a deterministic Ethernet network and show examples of how it actually works.

Click to expand...

http://www.automationworld.com/webcast-5917

CharlesM said:

EtherCat sure seems to be an up and coming network.

Click to expand...

It may or may not be an ominous sign, but Yaskawa recently came up with an EtherCAT version of their flagship Sigma 5 servodrive. Despite of and in addition to their own motion bus system.

Apparently there was enough customer demand or marketing forecasting to justify the development work.