I hear a lot about ProfiNet, Modbus/TCP and Ethernet/IP as options for industrial ethernet. Can anyone comment on which one is the most popular and what are the pros and cons of each?
I do not like any of them for peer to peer PLC communications.
Much prefer Omron Controller Link but it is proprietary.
In automation there is no standard for communication up to now. Each vendor uses it's own protocol. Now they begin using TCP/IP as a means for data exchange. They do handycrafts for creating their protocols on top of TCP. I don't know which one will be wide spread. One criteria will be openness.
- Modbus/TCP is documented publically
- ProfiNet is only visible for members of the Profibus Nutzer Organisation
- Ethernet/IP don't know if Rockwell opens their documentation
In most cases you will have to use the protocol your PLC can speak. You will have no choice.
You could eventually use gateways for converting different protocols.
The legal situation for anyone wanting to implement Ethernet/IP is quite complicated. There was a long and involved discussion on this last year. "http://www.control.com/thread/1026244726" Read that if you want the details.
I had a brief look just prior to replying to this, and so far as I can tell the situation hasn't changed since then. To put it briefly, getting your hands on a copy the documentation doesn't give you the rights to implement their "Licensed Technology". You have to be an approved vendor and sign a contract with them. That contract puts your relationship with ODVA under contract law, so consult a qualified lawyer and give it some careful thought before signing anything.
ODVA will license their protocol to approved "partners", but by no stretch of the imagination is Ethernet/IP an open protocol.
>... but by no stretch of the imagination is Ethernet/IP an open protocol. <
I've seen so many people try to use the word "Open" and "Free" interchangeably, or actually, they think the word "Open" means "Free".
EtherNet/IP is OPEN - anyone can get a spec and anyone can build EIP products. You just have to pay for it and conform to their rules. And the rules are there for a REASON - to ensure interoperability. EIP is a complicated protocol, and interoperability problems in the field are a huge reason why a user will choose one network over another. If the world was allowed to do whatever they wanted with EIP, I doubt anyone would stick to the spec, meanwhile interoperability problems would increase. For a protocol as complex as EIP to be successful, someone has to police it.
Modbus is FREE - anyone can get a spec and anyone can build Modbus products, but there are no rules. And of course there is no price on anything. And guess what, you end up getting 100 different flavors of Modbus out there. For the most part people stick to the spec, but that's also because Modbus is so darn simple. Registers and Coils - that's it. Obviously EIP has a ton more to it than registers and coils.
Now let's talk about a CLOSED specification. Obviously CLOSED is the opposite of OPEN. So when you say EIP is not open, you must mean it is closed? Uh, no. Suppose I invent a new protocol and call it SuperNet. It's the best thing since sliced bread. My company makes a bunch of controllers and devices that talk to each other using SuperNet. Customers love it. But I'm the only one who makes SuperNet devices. I don't give away my spec, nor do I sell it to anyone. It is a CLOSED protocol. Even if I pick and choose a handful of companies and allow them to make SuperNet devices, it's still CLOSED, or NOT OPEN.
Obviously ODVA is not following the model for a CLOSED protocol. EtherNet/IP is OPEN.
"Modbus is FREE - anyone can get a spec and anyone can build Modbus products, but there are no rules. And of course there is no price on anything. And guess what, you end up getting 100 different flavors of Modbus out there. For the most part people stick to the spec, but that's also because Modbus is so darn simple. Registers and Coils - that's it. Obviously EIP has a ton more to it than registers and coils."
Now come on. A person reading your post would expect Ethernet/IP to be much easier to integrate and commission than Modbus/TCP, which is the exact opposite of my experience. If EIP is a lot more than registers and coils, that's probably the problem.
In reply to John: You're trying to square the circle here. It's easy to define "open" - anyone can implement it without getting permission from anyone else. If you have to get permission, then it isn't open.
>EtherNet/IP is OPEN - anyone can get a
>spec and anyone can build EIP products.
>You just have to pay for it and conform
>to their rules.
So which is it then? Anyone can build EIP products, or only those who get permission are allowed to do so?
I'm reminded of an old Russian joke from back in the days of Stalin. An American was telling a Russian that the US had more freedom than Russia. "In America I can stand in front of the White House and shout 'down with America', and nothing will happen to me". The Russian replied "Well we can do the same thing in Russia. I can stand in front of the Kremlin and shout 'down with America' and nothing bad will happen to me either".
>And the rules are there for a REASON - to ensure
>If the world was allowed to do whatever they wanted with EIP, I doubt
>anyone would stick to the spec, (...)
>someone has to police it.
Here's the amazing thing. Here we are with people from all over the world communicating across the Internet in order to see this discussion on this web site. We're using different operating systems and web browsers. Some people have DSL, some people have cable modems, some people are using dial-up. Some people are reading this via a web interface, some are getting e-mail digests, some are monitoring via RSS feeds.
What's more there are tens of thousand, possibly hundreds of thousands (millions even?) other web sites out there doing more or less the same thing, but with different content. Some are built using PHP, some are collections of Perl scripts, some use the latest hot things like Django or Ruby on Rails.
There's all these amazingly complicated things happening all over the world, yet it all just seems to work. And guess what, nobody, but *nobody* is policing it. Everybody just does whatever they feel like. There's no association saying what you can or cannot do, no single company who controls all the software, there's nobody running it. And yet it all just works. That's pretty amazing when you think about it.
Here's the thing. People look at the Internet, and then they look at the effort and expense required just to get two PLCs to talk to each other (if it can be done at all), and they start to wonder if something is wrong here.
The Internet analogy of open/closed is seriously flawed. There are tons of things that have to be done in a specific way or none of it works. if people just did what they felt like, none of it would work. HTML works in a certain way, as does TCP/Ip. If you don't abide by things liek that, you get nothing.
I might agree that Ethernet I/P is not as open as one might want it to be, but it is about as "open" as it gets in the industrial control arena.
"I might agree that Ethernet I/P is not as open as one might want it to be, but it is about as "open" as it gets in the industrial control arena."
Both pro-EIP and anti-EIP posters in this thread have described it as less open than Modbus/TCP, and it's more plentiful as well.
In reply to bob peterson: People really can do what they want on the Internet. There is nobody enforcing any rules. Usually, what people want is for their stuff to work, so they do tend to follow standards and conventions. It's entirely voluntary however. That voluntary cooperation however works a lot better than anything we've seen come out of industry.
As for whether AB Ethernet I/P is "'open' as it gets in the industrial control arena", there are other common protocols that unlike EIP you are free to implement without restriction. At best you might be able to say that EIP isn't as bad as the worst of the others, but then some of the others are a pretty bad lot.
We had a discussion on this a couple of years ago. Here's the URL:
Go have a look at that discussion, which includes numerous quotes from the ODVA contract. How is any of that "open"?
Let's go back to the Internet analogy. How come I can view any web site anywhere in the world with any web browser that I have, but I can't plug an AB PLC into a Siemens PLC and have them talk to each other?
How come I can write my own web server (which I have done more than once), and it will talk with any web browser, but writing a server that can talk with an AB or Siemens PLC is almost impossible?
I'm not saying that the Internet is perfect. I'm just saying that if industrial networks were as good as the Internet people would be satisfied with that.
I think some of the reasons people say it is not open [on this forum] is that it doesn't comply with the Open source software model, which is to say anyone can use it for free with no restrictions on use (that is a real paraphrase, but you probably get the idea).
You can add FoundationT Fieldbus HSE to that list. For process control FF-HSE is the best option.
To fully understand FF-HSE, take a look at the book "Fieldbuses for Process Control: Engineering, Operation, and Maintenance" (buy online in hardcopy or download immediately in softcopy): http://www.isa.org/fieldbuses
If you can't buy the book now, you can download chapter 1 (overview) for free in softcopy form. It's free, but you must register an account. If your email does not support this hyperlink feature correctly, please copy the entire link and paste it into your Internet browser. Mind the line wrap, make sure to get the complete path all the way to the 4585:
Learn fieldbus at your own pace: www.isa.org/fieldbuses
ProfiNet is too new to know much about. I've never seen it in action. At least in the US it has not yet had a significant impact. Personally, I hate the fact that it is based on COM technology from Microsoft, but that should be transparent for end users.
EtherNet/IP (I HATE that name; it's so confusing) is an Allen-Bradley / Rockwell product. Yeah, yeah, ODVA and "open networks" and all that. Betting against Rockwell in the US is a risky proposition. EIP is pretty impressive, and there are already a fair number of products for it, even if those products are mostly Rockwell or Rockwell Encompass Partners. It is complicated, although Rockwell hides some of that complexity from end users if you're using their products. The benefit is flexibility.
Modbus/TCP has been available the longest, and is by FAR the simplest. You can implement Modbus/TCP support on any system with a TCP/IP stack in a day. It really is open; download 30-odd pages from modbus.org and off you go. There are two downsides; it's a stupid protocol and there's no driving industry force. By stupid, I mean that you send or receive a bunch of bytes. Are they supposed to be ints, floats, strings, what? You have to handle that manually. The upside of Rockwell's stake in EIP (similar with Siemens and ProfiNet) is that they have their marketing guys out there pushing its adoption. Modbus/TCP doesn't really have that same push. So even though Modbus/TCP has a modest lead right now it could easily get passed by EIP or ProfiNet.
The good news is that all three work. I don't think you can go wrong here. I would start with Modbus/TCP because it's so dead simple. Then you can add EtherNet/IP or ProfiNet later as the market shakes out. Or use my standard advice; if you have a vendor you trust and work with a lot then go with whatever solution they use.
Sage Automation, Inc.
The communication aspect of PROFInet is not very impressive. One "channel" uses DCOM in a way strikingly similar to the OPC-DX and the other implements a new media which makes it non-Ethernet and also removes TCP/IP to make it more like a Fieldbus.
I'm surprised that they did not simply put Profibus-DP or Profibus-FMS frames on UDP/IP and Ethernet.
Having said all that, perhaps we are all missing the point. The interesting part of PROFInet is not the communication. The way I look at PROFInet it is not really meant for communication from PLC to PLC or from configuration tool to PLC. As per documents from PND all this will still be done using Profibus-DP.
The interesting point of PROFInet is the IEC 61499 stuff which does not get much mention yet. To me, PROFInet is a scheme for integrating different machines and package units to a central system. I.e. it is more of a system-to-system link, not really a PLC-to-PLC link. The cool stuff here is that IEC 61499 appears to be implemented using some XML files that exposes the very small fraction of internal system data (thousands of objects) that are actually interesting to other subsystems. In my view, PROFInet is at the next higher level, above EtherNet/IP, Modbus/TCP, and FF-HSE. It's not
really competing with those. It is a sophisticated form of OPC-DX, a more clever way of passing data about the underlying system to the upper connection tool. The higher level system need not worry about nitty-gritty details within the subsystems. The IDA effort also mentioned IEC 61499 but I'm not sure what happened to it after they merged with Modbus.
Integrating package units is very different from, and have other difficulties, than integrating PLCs, so all bus organizations better look at this.
Learn fieldbus at your own pace: www.isa.org/fieldbuses
Many of the reasons for implementation are political not technical. Engineers can qualify their choice with technical merits but the real reason for one over the other is the PLC supplier and their support or percent of installed PLCs in their plant.
Rockwell is for the Rockwell part of the world. (Big industry and North America)
Profinet is for Europe and european suppliers i.e. Siemens, Modbus TCP/IP was the "other choice" as it was open from the begining and supported by Beckhoff and others who lied its simplicity.
MODBUS® TCP/IP has became an industry de facto standard because of its openness, simplicity, low cost development, and minimum hardware required to support it.
There are more than 200 MODBUS® TCP/IP devices available in the market. Process industry liked the origional modbus for it's simplicity and the early adopters who used it.
http://www.ethernet-ip.org/ for info on the "Rockwell" ethernet offering.
My opinion of ProfiNet is that it is still the future and not really in the present. it is not widely used in North America yet...
Of course all my thoughts are based on North American usage and knowledge.
Depends on what you are trying to accomplish. Generically, they are all based on TCP but the difference is the application layer. For EtherNet/IP it is CIP (common industrial protocol) which is the same for DeviceNet and ControlNet (see www.odva.org). I think ProfiNet is in the same situation but I am not sure (see www.profibus.com) they have ProfiNet seminars going on that you can attend.
In the end, these NEW Ethernet based protocols are just a way for Siemens and Rockwell - Allen Bradley to addrest the Ethernet question and maintain their market share. I just did a formal review on this for our company and found only one company to be able to support ALL THREE (Modbus/TCP, EtherNet/IP, and ProfiNET)and that is WAGO (www.wago.com). The beauty of their system is that the cpu and network comm are all in one front end module. In addition, the I/O cards are universal for all front end modules (CPU's). The 750-841 meets the spec for both TCP and EtherNet/IP. The 750-840 is their new ProfiNet CPU.
Then Click on Programmable Fieldbus Controller.
Modbus/TCP is about 5 years old and is widely used in PC-PLC communications, and with hundreds of other devices as well. It uses only TCP/IP frames making it universal in application, but unsuited for any real-time work. Look at this web site: www.modicon.com
EtherNet/IP is an open network supported by ODVA. It has been commercial about 2 years and is used by Rockwell Automation and their
Encompass Partners for their products. Data transfers use UDP/IP and are suitable for use in soft real-time applications. See this
web site: www.odva.org
PROFInet is relatively new and was developed by the Profibus organization. It has versions using TCP/IP for Master-slave (PROFINET
I/O) communications and a version (CbA)using UDP/IP for soft real-time applications. By next year, a version using a special ASIC
will be available for (IRT) hard real-time applications, typically in motion control. See this web site: www.profinet.net
For a full discussion please purchase my book Automation Network Selection, published by ISA that discusses all of these networks.
See this web site: www.isa.org/books
Richard H. Caro, CEO
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
E-mail: RCaro@CMC.us <mailto:RCaro@CMC.us>
Buy my books: Automation Network Selection
Wireless Networks for Industrial Automation
The Consumer's Guide to Fieldbus Network Equipment
for Process Control
I've been working with Ethernet/IP for over two years and can tell you a few things:
1. Class 1 uses UDP. Its good for general I/O when losing a packet every now and then is OK. Otherwise, it is not the best technology in my opinion.
2. Class 3, otherwise known as explicit messaging, uses TCP connections and is ideal for applications when lost data is not acceptable. This mode is peer-to-peer and works very well.
Adding Ethernet/IP to a device is a big deal. Very expensive to buy the stacks and half a man-year to develop one yourslf. Then there is ODVA approval and that costs $5000 too...
None of the Allen Bradley PLCs are capable of millisecond time frames over Ethernet/IP due to the nature of the backplane on their PLCs. But they are an excellent choice for many applications in my opinion.
Modbus/TCP. That can be made to run peer-to-peer, at least according to the Modbus/TCP spec (but not their model). The problem is that many PLCs cannot handle the networking processes and very few devices can.
Adding Modbus/TCP server (slave) mode to a device is easy and cheap.
Profinet. There is an interesting one. CBA was a misfire in my opinion. Way to complicated for average PLC programmers to use. I/O is different and should have legs. Expect it to take more market share as it gains acceptance.
Adding Profinet is easier than Ethernet/IP, but harder than Modbus/TCP. The stack costs about $6000 which is a fraction of a good Ethernet/IP stack.
Just my opinions.
Ethernet IP and Modbus TCP are really only addressing I/O connectivity from the PLC to the remote I/O. The question to ask is what (other than possibly simplified cabling) does this give me that the existing I/O networks from Modicon and AB do not. The real question is based on enterpise connectivity, connectivity of multiple brands of PLC, connectivity of stand alone machines into on integrated common ethernet network. Profinet with Component Based Automation is the only network addressing the whole picture.