S
Is TCP/IP a real-time protocol?Could it be used in field layer of an automation system?
Thanks
Thanks
Scroll to continue with content
S
Sergey
Yes. Stack TCP/IP can be used in the physical layer at your automation system. Many companies do this. For example Schneider Electric. They use modbus over IP and on the second layer is ethernet.
A
Anonymous
Real time, hmmm, well, nothing is real time but 100 Mbps is fast enough.
For field layer it's gonna be a bit difficult since most of the devices do not support this protocol.
For field layer it's gonna be a bit difficult since most of the devices do not support this protocol.
K
Ken Muir
Which really poses the question of 'what is real-time?'
Real-time is whatever is adequate for the task in hand. If you're measuring the level of water in a reservoir, then a response time measured in hours may be perfectly good. On the other hand, if you're trying to do synchronisation between two high-speed moving machines, then milliseconds may be much too slow.
There's no such thing as a real-time protocol in isolation - it all has to be judged in terms of the application. What are you trying to do, how often, what are your process constraints etc ...?
Regards
Ken
Real-time is whatever is adequate for the task in hand. If you're measuring the level of water in a reservoir, then a response time measured in hours may be perfectly good. On the other hand, if you're trying to do synchronisation between two high-speed moving machines, then milliseconds may be much too slow.
There's no such thing as a real-time protocol in isolation - it all has to be judged in terms of the application. What are you trying to do, how often, what are your process constraints etc ...?
Regards
Ken
A
Anonymous
> Is TCP/IP a real-time protocol? <
No.
> Could it be used in field layer of an >automation system? <
Not if you do require guarenteed response time limits (normally, you do).
No.
> Could it be used in field layer of an >automation system? <
Not if you do require guarenteed response time limits (normally, you do).
R
Robert Scott
On a lightly-loaded dedicated network, TCP/IP can and does give adequate real-time response. So practically speaking it is real-time, even though it does not have the guaranteed maximum latency of something like CAN.
Robert Scott
Real-Time Specialties
Embedded Systems Consulting
Robert Scott
Real-Time Specialties
Embedded Systems Consulting
D
Dick Caro
No, Sabrina, TCP/IP is very definitely not real-time. All errors detected by TCP/IP are corrected by retransmission. Routing of messages may take different paths for each message or even parts of messages called segments with indeterminate time delays.
That being said, TCP/IP is used for higher layer communications in control systems and even some for the field layer where tight real-time requirements are not needed. The most popular use is Modbus/TCP. EtherNet/IP also uses TCP/IP for its non-real-time data transfers. However, EtherNet/IP and Foundation Fieldbus HSE both use UDP/IP for time-critical data transfers.
For more information about industrial networks, please read my books available from ISA.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
That being said, TCP/IP is used for higher layer communications in control systems and even some for the field layer where tight real-time requirements are not needed. The most popular use is Modbus/TCP. EtherNet/IP also uses TCP/IP for its non-real-time data transfers. However, EtherNet/IP and Foundation Fieldbus HSE both use UDP/IP for time-critical data transfers.
For more information about industrial networks, please read my books available from ISA.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
A
August
TCP/IP defines the network and transport layers of the communications. Very often the real-time is met by the MAC layer. Often it is argured that Ethernet cannot be suitable for real time applications but that is when Ethernet is used as a bus. Most Ethernet is now on Ethernet switches and they are argued to be good for real time. See http://www.industrialnetworking.co.uk for some articles.
On the other hand field level protocols cater for short bursty data not something that TCP/IP is good at. So fieldbus protocols like Hart, FF and profibus do not use TCP/IP at the field layer.
For fast applications in 5-10 ms or even <1ms response TCP/IP cannot provide the response required. Protocols like Profinet directly accesses the Ethernet layer from the application in SRT or IRT mode.
On the other hand field level protocols cater for short bursty data not something that TCP/IP is good at. So fieldbus protocols like Hart, FF and profibus do not use TCP/IP at the field layer.
For fast applications in 5-10 ms or even <1ms response TCP/IP cannot provide the response required. Protocols like Profinet directly accesses the Ethernet layer from the application in SRT or IRT mode.
S
Shailesh
which protocols are real time?
shailesh
shailesh
J
Jonas Berge
I agree with Dick that UDP/IP is a better platform for control network.
I also want to stress that TCP/IP or UDP/IP are not complete protocols. They are just two of the layers. There are many industrial protocols based on TCP/IP and UDP/IP over Ethernet media.
If you are doing process control you may wish to look at the Foundation fieldbusT HSE technology. If you are doing factory automation you may wish to look at EtherNet/IP. If you are doing motion control then TCP/UDP/IP is not for you.
When you say "field layer" I assume you are asking if you can connect transmitters, analyzers, and valve positioners using Ethernet. You can't - today. These devices do not support Ethernet and UDP/IP today. The field-level has very unique requirements and Ethernet is not quite there
yet. For example, at the field-level you want to run long wires from control room to the field, usually in excess of several hundred meters, but Ethernet is limited to 100 m for copper and fiber is kind of expensive. Since you have many instruments you want to multidrop many of them on one wire, Ethernet does not do this as it connects one device per wire and needs an active LAN switch. Due to the many devices, you want to provide power over the bus using the same two wires - and it shall often be intrinsically safe
- which Ethernet does not do. And so on. Therefore Ethernet is used at the remote-I/O level and the control level. At the field-level you use fieldbus. That is, fieldbus and Ethernet complement each other to provide a complete system architecture.
To understand the different network layers and the difference between fieldbus and Ethernet, take a look at chapter 1 and 2 of the yellow book "Fieldbuses for Process Control: Engineering, Operation, and Maintenance"
buy online:
http://www.isa.org/fieldbuses
Jonas Berge
SMAR
===========
[email protected]
www.smar.com
Learn fieldbus and Ethernet at your own pace: www.isa.org/fieldbuses
Learn OPC and automation software at your own pace: www.isa.org/autosoftware
I also want to stress that TCP/IP or UDP/IP are not complete protocols. They are just two of the layers. There are many industrial protocols based on TCP/IP and UDP/IP over Ethernet media.
If you are doing process control you may wish to look at the Foundation fieldbusT HSE technology. If you are doing factory automation you may wish to look at EtherNet/IP. If you are doing motion control then TCP/UDP/IP is not for you.
When you say "field layer" I assume you are asking if you can connect transmitters, analyzers, and valve positioners using Ethernet. You can't - today. These devices do not support Ethernet and UDP/IP today. The field-level has very unique requirements and Ethernet is not quite there
yet. For example, at the field-level you want to run long wires from control room to the field, usually in excess of several hundred meters, but Ethernet is limited to 100 m for copper and fiber is kind of expensive. Since you have many instruments you want to multidrop many of them on one wire, Ethernet does not do this as it connects one device per wire and needs an active LAN switch. Due to the many devices, you want to provide power over the bus using the same two wires - and it shall often be intrinsically safe
- which Ethernet does not do. And so on. Therefore Ethernet is used at the remote-I/O level and the control level. At the field-level you use fieldbus. That is, fieldbus and Ethernet complement each other to provide a complete system architecture.
To understand the different network layers and the difference between fieldbus and Ethernet, take a look at chapter 1 and 2 of the yellow book "Fieldbuses for Process Control: Engineering, Operation, and Maintenance"
buy online:
http://www.isa.org/fieldbuses
Jonas Berge
SMAR
===========
[email protected]
www.smar.com
Learn fieldbus and Ethernet at your own pace: www.isa.org/fieldbuses
Learn OPC and automation software at your own pace: www.isa.org/autosoftware
M
Max Power
NONE - you must select a protocol and transceiver speed to suit your process' reqirements!
S
Sabrina
Speaking of Hart, the data transmission rate is 1200 bit/s. Is it too slow for polling data from the devices in the field?
D
Dick Caro
Error correction: Sergey - TCP/IP is not a Physical Layer protocol. Schneider Electric originated Modbus/TCP and gave it to Modbus-ida.org for distribution, but Modbus/TCP is not intended to be for low level (I/O) communications. It is intended for Level 2 automation between computers and PLCs or between PLCs on a network. Ethernet is used at the Data Link Layer and Physical Layers of a Modbus/TCP network. Layers are defined by this international standard:
ISO-OSI 7 layer Reference Model, officially known as ISO Standard 7498-1:1994
Levels of automation are defined best by this standard:
ANSI/ISA-95.00.01-2000, Enterprise-Control System Integration, Part 1: Models and Terminology.
See this url: http://www.isa.org/MSTemplate.cfm?MicrositeID=285&CommitteeID=4747
It is too easy to confuse the term "Layers" with the term "Levels."
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
ISO-OSI 7 layer Reference Model, officially known as ISO Standard 7498-1:1994
Levels of automation are defined best by this standard:
ANSI/ISA-95.00.01-2000, Enterprise-Control System Integration, Part 1: Models and Terminology.
See this url: http://www.isa.org/MSTemplate.cfm?MicrositeID=285&CommitteeID=4747
It is too easy to confuse the term "Layers" with the term "Levels."
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
D
Dick Caro
There are three true requirements for real-time message delivery: 1) the time delay must be less than the time required for action,
2) the transport delays of the network must be deterministic - that is the maximum possible or worst-case network delay is predictable, and
3) every message must be delivered to its intended destination.
TCP/IP is the Network and Transport layer protocols used on the Internet and most LANs. The IETF defines the goal of TCP/IP to make a "best effort" to deliver every message without error. There are no guarantees. This fails requirements 2 and 3 above since messages can be lost on the Internet, but probably not on a LAN.
All of the protocols included in the IEC 61158 Fieldbus standard meet the requirements for being real-time by design. The includes the UDP/IP-based Foundation Fieldbus HSE (High Speed Ethernet) and PROFInet. EtherNet/IP defined by ODVA is also a real-time network based on Ethernet and both TCP and UDP over IP. DeviceNet and most of the CAN-based protocols are also real-time.
All of this is discussed in my book Automation Network Selection published by ISA.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
2) the transport delays of the network must be deterministic - that is the maximum possible or worst-case network delay is predictable, and
3) every message must be delivered to its intended destination.
TCP/IP is the Network and Transport layer protocols used on the Internet and most LANs. The IETF defines the goal of TCP/IP to make a "best effort" to deliver every message without error. There are no guarantees. This fails requirements 2 and 3 above since messages can be lost on the Internet, but probably not on a LAN.
All of the protocols included in the IEC 61158 Fieldbus standard meet the requirements for being real-time by design. The includes the UDP/IP-based Foundation Fieldbus HSE (High Speed Ethernet) and PROFInet. EtherNet/IP defined by ODVA is also a real-time network based on Ethernet and both TCP and UDP over IP. DeviceNet and most of the CAN-based protocols are also real-time.
All of this is discussed in my book Automation Network Selection published by ISA.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
J
Jonas Berge
TCP/IP is not the physical layer.
TCP is transport layer 4, UDP is also at this level IP is network layer 3.
TCP/IP is usually used on top of Ethernet as a physical layer 1 (and data link layer 2).
Modbus, HSE, and CIP etc. are application layer 7.
For layers, take a look at chapter 11 of the yellow book "Fieldbuses for Process Control: Engineering, Operation, and Maintenance" buy online:
http://www.isa.org/fieldbuses
Jonas Berge
SMAR
===========
[email protected]
www.smar.com
Learn fieldbus and Ethernet at your own pace: www.isa.org/fieldbuses
Learn OPC and automation software at your own pace: www.isa.org/autosoftware
TCP is transport layer 4, UDP is also at this level IP is network layer 3.
TCP/IP is usually used on top of Ethernet as a physical layer 1 (and data link layer 2).
Modbus, HSE, and CIP etc. are application layer 7.
For layers, take a look at chapter 11 of the yellow book "Fieldbuses for Process Control: Engineering, Operation, and Maintenance" buy online:
http://www.isa.org/fieldbuses
Jonas Berge
SMAR
===========
[email protected]
www.smar.com
Learn fieldbus and Ethernet at your own pace: www.isa.org/fieldbuses
Learn OPC and automation software at your own pace: www.isa.org/autosoftware
C
Curt Wuollet
No and Yes!, respectively.
Regards
cww
Regards
cww
J
Jonas Berge
HART is too slow for closed loop control. That is, you cannot use HART for the PLC to read process value from a transmitter and write the output to the valve positioner. That will be too slow. That is why HART retains the 4-20
mA signal for the closed loop control.
HART uses digital communication only for retrieve identification information, for calibration, diagnostics, and device parameterization - which is done as and when required, often through a hand-held connected ad-hoc. You can have a permanent HART communication infrastructure if your control system has IO modules supporting HART, or if you use a stand-alone HART multiplexer to a separate software.
For HART architecture and use, take a look at the yellow book "Fieldbuses for Process Control: Engineering, Operation, and Maintenance" buy online: http://www.isa.org/fieldbuses
Jonas Berge
SMAR
===========
[email protected]
www.smar.com
Learn fieldbus and Ethernet at your own pace: www.isa.org/fieldbuses
Learn OPC and automation software at your own pace: www.isa.org/autosoftware
mA signal for the closed loop control.
HART uses digital communication only for retrieve identification information, for calibration, diagnostics, and device parameterization - which is done as and when required, often through a hand-held connected ad-hoc. You can have a permanent HART communication infrastructure if your control system has IO modules supporting HART, or if you use a stand-alone HART multiplexer to a separate software.
For HART architecture and use, take a look at the yellow book "Fieldbuses for Process Control: Engineering, Operation, and Maintenance" buy online: http://www.isa.org/fieldbuses
Jonas Berge
SMAR
===========
[email protected]
www.smar.com
Learn fieldbus and Ethernet at your own pace: www.isa.org/fieldbuses
Learn OPC and automation software at your own pace: www.isa.org/autosoftware
D
Dick Caro
Sabrina,
No, HART is not too slow, but it depends upon your process requirements. Most installations use the 4-20mA signal provided by HART for the PV (Process Variable), and that is virtually instantaneous. If you must use the digital signal of HART, then it would not usually be good for fast control requirements such as flow or some level applications. However, the digital signal of HART is usually fast enough for temperature control.
To learn how to pick the right fieldbus for Process Control, I recommend reading my book:
Consumer Guide to Fieldbus Network Equipment for Process Control -- referenced below.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
No, HART is not too slow, but it depends upon your process requirements. Most installations use the 4-20mA signal provided by HART for the PV (Process Variable), and that is virtually instantaneous. If you must use the digital signal of HART, then it would not usually be good for fast control requirements such as flow or some level applications. However, the digital signal of HART is usually fast enough for temperature control.
To learn how to pick the right fieldbus for Process Control, I recommend reading my book:
Consumer Guide to Fieldbus Network Equipment for Process Control -- referenced below.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
M
Michael Griffin
I researched Ethernet I/O for a recent project, and Modbus/TCP seems to be the most widely supported Ethernet based I/O protocol. Examples of supporting companies are Phoenix, Beckhoff, and Opto-22. It seems to be the closest
thing to a universal standard.
I don't see why you feel it wouldn't be suitable for I/O. The simplicity and low overhead off Modbus should make it suited for Ethernet devices that don't have a lot of computing power available.
thing to a universal standard.
I don't see why you feel it wouldn't be suitable for I/O. The simplicity and low overhead off Modbus should make it suited for Ethernet devices that don't have a lot of computing power available.
D
Dick Caro
Good question Mike!
When thinking of I/O, the model is remote I/O for a PLC. Plain simple "dumb" remote I/O. The problem is determinism - when you need for the PLC to react in "real-time." If there are no time-critical I/O points and remote I/O is not being used for a safety application, then determinism is not required and Modbus/TCP is OK. Otherwise, it would be better to use a deterministic and fast communications bus such as Modbus/RTU, PROFIbus, Interbus, ASi, DeviceNet, ControlNet, or EtherNet/IP, depending on the networks supported by the remote I/O supplier and the PLC supplier.
Modbus/TCP is also OK when the remote I/O unit is "smart" such as for Opto22 I/O equipped with a "Brain Board," or the Schneider Momentum for which it was designed. Here the time-critical functions, such as interlocks, should be performed by the smart I/O unit and not wait for reaction by the PLC.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
When thinking of I/O, the model is remote I/O for a PLC. Plain simple "dumb" remote I/O. The problem is determinism - when you need for the PLC to react in "real-time." If there are no time-critical I/O points and remote I/O is not being used for a safety application, then determinism is not required and Modbus/TCP is OK. Otherwise, it would be better to use a deterministic and fast communications bus such as Modbus/RTU, PROFIbus, Interbus, ASi, DeviceNet, ControlNet, or EtherNet/IP, depending on the networks supported by the remote I/O supplier and the PLC supplier.
Modbus/TCP is also OK when the remote I/O unit is "smart" such as for Opto22 I/O equipped with a "Brain Board," or the Schneider Momentum for which it was designed. Here the time-critical functions, such as interlocks, should be performed by the smart I/O unit and not wait for reaction by the PLC.
Dick Caro
===========================================
Richard H. Caro, CEO
CMC Associates
2 Beth Circle, Acton, MA 01720
Tel: +1.978.635.9449 Mobile: +.978.764.4728
Fax: +1.978.246.1270
E-mail: [email protected]
Web: http://www.CMC.us
Buy my books:
Automation Network Selection http://www.isa.org/networkselect
Wireless Networks for Industrial Automation http://www.isa.org/wirelessnetworks
The Consumer's Guide to Fieldbus Network Equipment for Process Control
http://www.spitzerandboyes.com/Product/fbus.htm
Attend my seminar: Fieldbus Network Equipment for Process Control Seminar
http://www.spitzerandboyes.com/Seminar/semfbus.htm
===========================================
