Talk:EtherCAT

Copyright of EtherCAT Material on Wikipedia

The material initially provided about EtherCAT is copyrighted by the EtherCAT Technology Group and/or the company Beckhoff Automation GmbH. The re-use of this material in Wikipedia was permitted expressly by both organizations in an email by the Executive Director of the EtherCAT Technology Group and Head of Technology Marketing of Beckhoff directed to permissions(at)wikimedia(dot)org. —Preceding unsigned comment added by Fieldbusguy (talk • contribs) 11:48, 20 October 2006.

Fieldbusguy, I have not seen the e-mail, but you'll want to ensure the permission is not just confined to use by Wikipedia, since Wikipedia in turn makes all of its articles available for reuse and modification by others. As a result, Wikipedia requires that such permission apply to everyone not just Wikipedia in accordance with its legal terms. (Caveat: that's my layman's possibly incorrect, summary interpretation -- refer to the links on the article's copyright violation notice for the actual legal terms). --A. B. 16:51, 23 October 2006 (UTC)

A.B., you are right - you did not see the e-mail, which contains this: "... and permit the re-use of this material under the GNU Free Documentation License". I hope one does not have to employ a lawyer to contribute to Wikipedia. Fieldbusguy 20:54, 23 October 2006 (UTC)

No you don't need a lawyer! (You just can't keep any strings attached to edits or uploaded files.) --A. B. 20:57, 23 October 2006 (UTC)

Notability and reference tags

I added notability and reference tags. It appears that EtherCAT is probably notable, however, this must be documented using English-language references that meet the Wikipedia:Reliable sources guideline. Once that's done, delete the tags. --A. B. 17:02, 5 November 2006 (UTC)

EtherCAT is notable, however this page is currently an advertisement.

I tried to remove the advertisement.Fieldbusguy 14:13, 2 April 2007 (UTC)

I have access to the IEC 61158 documents, and I can verify that the entire technical description of EtherCAT (That is the "Introduction", "Functional principle", "Protocol", "Performance", "Topology", "Synchronization" and "Device profiles" chapters) of the article are consistent with the IEC 61158 documents mentioned at the end of the article. I assume that IEC standard documents are considered to be reliable sources by wikipedia? I can't think of a better source.Brolin 21:25, 17 April 2007 (UTC)

I would expect key statements to be linked to the IEC document, perhaps by section number. Since most people don't have access to IEC, it would be cool to find open sources such as trade magazines, or perhaps company whitepapers.Feldercarb (talk) 14:35, 9 August 2016 (UTC)

Speed and availability

I am working with drive application since almost 30 years, and I can not imagine, how the drive controller for 100 drives which is able to update all drives in 100 microseconds looks like. Please give more precise information. Ethercat is not the fastets Industrial Ethernet Solution, as Ethernet Powerlink announced to use Gigabit Ethernet in the future. It is not possible, that Ethercat is used in series production since 2003, because I was not able to get some components from different suppliers like Beckhoff, Kübler, Fraba Posital and so on until today. Where should other companies have bought their components for series production three years ago. DriveExpert 18:20, 8 November 2006 (UTC)

You are right: there is no drive controller around that is able to control 100 drives with a control loop of 100 microseconds. But this article discusses a communication system, and not the drive controller. And the point I wanted to make is that the communication system is so fast that it does not slow down any controls application: there is no communication bottleneck. With EtherCAT, you can send new command values to 100 drives every 100 µs (in fact every 80µs), and also receive actual values from each of those drives every 100µs. Depending on the network topology and physical layer chosen, the entire cycle may take a little longer since each device introduces a small delay - but since EtherCAT is full duplex one does not have to wait for the response before the next frame is sent.

EtherCAT is used in series applications since end of 2003, as can be seen this application report published at the beginning of 2004: http://www.pc-control.net/pdf/012004/pcc_schuler_e.pdf. As with every technology, first applications were pilot applications. But the application described in this article was started mid of 2003, and the company is shipping EtherCAT equipped machines since end of 2003. At the recent SPS/IPC/Drives show in Nuremberg, Germany, 48 vendors showed over 90 different EtherCAT products. It is possible, though, that not all of them are shipping worldwide, and some may still be in prototype phase. Please compare the acceptance rate of EtherCAT with any other fieldbus technology. Fieldbusguy 20:35, 22 December 2006 (UTC)

Yes, Ethernet Powerlink has announced the possibility of using Gigabit Ethernet in the future, but 100Mbit/s EtherCAT is actually faster than Gigabit Powerlink due to the large Ethernet overhead (minimum packet size of 46 bytes, and long required interframe gaps) which affects Powerlink and most other industrial ethernet solutions. The claim that EtherCAT is the fastest industrial ethernet solution is very much valid. Brolin 23:03, 21 February 2007 (UTC).

It is still valid that Gigabit is 10 times faster than 100MBit. The statement that 100MBit EtherCAT is faster than Gigabit Powerlink due to the long overhead only applies to very special applications i.e. the way Beckhoff uses EtherCAT in it´s I/O slices (that is not even Ethernet by the way) and therefore calculating only a few bytes per node. --Fieldbusboy 04:26, 27 July 2007 (UTC)

Well, it is not quite that simple. Gigabit may sound like 10 times faster than 100Mbit, but "speed" in a industrial ethernet network is not measured in Mbit/s, it is measured in cycle times. The cycle times on Ethernet powerlink are limited by the master and slave response times. Those times will be completely unaffected by a switch to gigabit, so the switch to Gigabit will not really affect the main bottleneck at all.

Secondly there is the minimum frame length. 100Mbit Ethernet has a minimum frame length of 46Bytes, but Gigabit Ethernet has a minimum frame length of 512Bytes. Which just happens to be about 10 times the minimum frame length of 100Mbit Ethernet.

Calculations have shown that a performance gain of 1.38-2 can be expected if you switch from 100Mbit to gigabit powerlink. That is assuming that you use gigabit hubs and not switches. If you use gigabit switches, then the performance will be worse than 100Mbit Ethernet due to the switch latency. Do you know where to buy gigabit hubs these days? I don't.

I don't think there is a single realistic application where Gigabit powerlink will have a shorter cycle time than 100Mbit EtherCAT. Only thing I can think of which might be just slightly faster than EtherCAT would be a very small gigabit powerlink network with just a few nodes and each node has a absolutely huge amount of I/O data (512Bytes or more per node), but that is not a realistic application. If you can show me an realistic example where Gigabit powerlink would be faster than EtherCAT, please do.

There are other aspects as well.. You cannot mix 100Mbit and Gigabit EPL nodes without using switches, and switches will reduce performance significantly. So you get a compatibility issue.

Another problem is the high power consumption of Gigabit interfaces.

I'am not sure what you mean when you say that the long overhead only applies to very special applications. It applies to every single industrial Ethernet node with less than 46Bytes of I/O data. Drives for example typically have 4-8 bytes of I/O data.

Beckhoffs "I/O slices" (Called E-Bus terminals) are not really any different from other EtherCAT slaves. The only difference is that they use a much cheaper short range LVDS based physical layer to keep costs down. The protocol is still Ethernet.

EtherCAT does indeed use Ethernet in a somewhat "unusual" way in order to get as high performance as possible at the lowest possible price, this is no secret. All the other high performance industrial Ethernet protocols (including powerlink) does the same, simply because it is necessary in order to get high real time performance out of Ethernet.

Brolin 19:20, 31 July 2007 (UTC)

EtherCAT and Open Source

EtherCAT is often praised as an open standard with open source implementations, but it is not so open at all. If you want to write and publish your own EtherCAT master you are not allowed to distribute it publicly unless approved by Beckhoff (holder of the EtherCAT standard). To get approval you need to meet certain criterias regarding completeness of implementation and absence of bugs, which usually contradict the "release early philosophy" of open source software. Even worse, you have to fix all bugs yourself (since you cannot distribute), killing one of the main advantages of collaborative open source software design.--Tobox (talk) 10:28, 3 August 2009 (UTC)

There are EtherCAT Open and Shared Source Implementations since 2006 (see also: EtherCAT Technology Group (ETG) website on open source), used successfully in many industrial applications. ETG and Beckhoff support those implementations actively, but have taken the position that "Open Source" should not jeopardize the compatibility and interoperability of the technology: implementing incompatible variants of the technology is not supported, regardless of the license model of the implementation. Therefore there are rules which aim to ensure compatibility and interoperability in the best interest of the openness of the technology. ETG does not share the view that "open" has to include that anybody may change the technology in any way. ETG holds the view that "open" shall mean that each and every company, university or institution may implement and use the technology in a compatible way at no charge (EtherCAT master) or at very low costs (e.g. there is, just as with CAN, a small license fee embedded in the costs of the EtherCAT slave controller chips). ETG membership is free of charge and ETG also provides implementation support free of charge. By the way, I would like to suggest that the discussion "pros and cons of open source" is held in more suitable panels. Fieldbusguy (talk) 09:42, 7 August 2009 (UTC)

Background/Context

This article has a lot of information, but lacks context. When and where was this system first developed? by who? Who uses this system? Clerks 15:34, 1 December 2006 (UTC)

EtherCAT was developed by Beckhoff Automation: http://www.beckhoff.com.

It was first presented by Beckhoff in 2003: http://www.embeddedstar.com/press/content/2003/4/embedded8141.html.

The EtherCAT Technology Group (ETG) was formed later the same year.

It can be used in pretty much all traditional fieldbus applications (see wikipedia keyword: fieldbus), EtherCAT is part as "Type 12" of the IEC 61158 international fieldbus standard described in the wikipedia "fieldbus" article.

EtherCAT can also be used in new applications with higher performance requirements than what traditional fieldbusses can provide. Motion control servo drives is one such application.

Examples of products and companies which use EtherCAT can be found here: http://www.ethercat.org/en/products.html. Brolin 23:03, 21 February 2007 (UTC)

Good article...but

Lots of information, however the article would benefit from being put in standard English. — Preceding unsigned comment added by 107.10.155.15 (talk) 03:06, 7 March 2014 (UTC)

The Synchronization section says specifically that Ethercat is a Ring Topology. However, the Topology section says many topologies are supported. This discrepancy must be addressed. — Preceding unsigned comment added by Kenbrey (talk • contribs) 02:42, 21 September 2016 (UTC)

Data Security

The Section EtherCAT#Application_Examples says EtherCAT is used for DataSecurity and Integrity. But the whole Article and a short google search give no hint how EtherCAT could achieve this. It does not implement any crypto as far as I know. --Wuestenschiff Dune-talk 07:57, 11 February 2016 (UTC) — Preceding unsigned comment added by Wuestenschiff (talk • contribs)