A ControlGlobal article, Industrial Networks: The Ethernet vs. Fieldbus Cage Match, spawned a discussion in the PROFIBUS and PROFINET LinkedIn group. I wanted to share Emerson’s Jonas Berge‘s posts on this topic:
- Ethernet is slowly taking the place of “H2 fieldbus”, but;
- “H1 fieldbus” is taking the place of 4-20 mA and on/off, and;
- I do not see Ethernet taking the place of 4-20 mA and on/off, hence;
- Ethernet and “H1 fieldbus” will coexist
Let me explain the important difference between “H1 fieldbus” or “H2 fieldbus” which is often overlooked in the debate on Ethernet and fieldbus, but first, let’s look at 4-20 mA and on/off signals.
Ethernet is great. But I don’t really have a problem with RS-485 either. However, for automation to move to the next level, we really got to have digital communication all the way down to the sensors and actuators. We cannot rely on analog 4-20 mA and discrete on/off signals. So to me, the important discussion is how do we get from hardwired signals for sensors and actuators, to digital communication; be it Ethernet or fieldbus, on wires or wirelessly? Sensors in particular, but also actuators, are the basis for industry initiatives like “Internet of Things”, “Industrial Internet”, “Industry 4.0”, or “M2M” etc. but we cannot debate pervasive networking if we still use analog signals for the sensors etc. The future is digital, completely digital all the way from sensors and actuators up to the controllers. Fieldbus is the underlying sensor/actuator level communication technology that can bring about these initiatives.
Keep in mind this difference:
“H2 fieldbus” connects at level 1-1/2 of the Purdue reference model such as Modbus/RTU, PROFIBUS-DP and DeviceNet:
- I/O subsystems
- Wireless gateways
- Package unit PLCs
“H1 fieldbus” connects at level 1 of the Purdue model such as PROFIBUS-PA, FOUNDATION fieldbus, IO-link, CompoNet, ASi, and to some extent HART:
- Control valve positioners
- On/off valves
- Electric actuators
Really what we have been seeing is Ethernet taking the place of old digital communication networks; control network (level 2 in the Purdue reference model) and remote-I/O (H2 fieldbus at level 1-1/2 in the reference model), that is, where legacy protocols (often proprietary) have initially paved the way.
Most sensors/transmitters and positioners/actuators/valves (at level 1 of the Purdue model) use on/off and 4-20 mA hardwiring, so we are really talking about is the possibility of Ethernet taking the place of hardwiring, not really about Ethernet taking the place of fieldbus. What would it take?
That is, in previous evolutions Ethernet has been taking the place of other digital communication, but now it would largely have to take the place of analog and discrete signals – an entirely different beast. Moving from RS485 to Ethernet is no big deal, one communication media to another, associated work practices remain the same. This is why Ethernet has been able to make such transitions with relative ease. But moving from hardwiring to digital communication, this is a large step. So will Ethernet be able to directly displace 4-20 mA and on/off, or do we need to go through an intermediate step where an H1 fieldbus first takes the place of 4-20 mA and on/off in a majority of plants, and then, in the next evolution, Ethernet in some form takes the place of H1 fieldbus for sensors and actuators? That is, can Ethernet take the place of hardwiring, and in what form?
I’m a strong advocate of replacing 4-20 mA and on/off with digital communication, and I would be glad to see Ethernet in some form take the place of hardwired signals. Indeed there are some flowmeters and gas chromatographs available with Ethernet. Gas chromatographs and flowmeters are examples of devices that have sometimes been integrated using RS-485, so these are also examples of where Ethernet has taken the place of RS485 protocols rather than taking the place of 4-20 mA and on/off.
So what will it take for Ethernet to take the place of 4-20 mA and on/off signals for sensor/transmitter and positioner/actuator/valve? Industrial Ethernet development so far follows developments in the consumer products which then become industrially hardened. So I guess we would first see perhaps some form of Ethernet taking the place of USB, for devices like mouse, keyboard, and ThumbDrive etc., and then get industrially hardened?
Apart from the communication and power on the same cable, very long cable, as well as intrinsic safety someone already eluded too; there are some major challenges moving from 4-20 mA and on/off signals to digital communication. RS485 in the past, or Ethernet today, at the “H2 fieldbus” level (level 1-1/2), is used with maybe up to a hundred or so nodes (IP of course supports millions) because there aren’t that many PLCs, I/O subsystems, wireless gateways, or drives/MCC in a plant. One of the challenges with sensors and actuators is that there are thousands of them in a plant; lots of Ethernet switches, and power, would be required. Also, PLCs, I/O subsystems, gateways, and drives/MCC don’t fail much because they are protected indoor and have redundancy, so device replacement is not an issue. Moreover, if they fail, they will invariably be replaced by a new node of the same model. The I/O subsystem, wireless gateway, drive/MCC, or PLC is not switched to another brand or model if it fails, so no issues – and it might be the system integrator doing it. These devices do not need periodic calibration, so no issue there either. However, sensors and actuators (at level 1 of the Purdue model) are in the harsh outdoors, in direct contact with the process (extreme temperatures, corrosion, abrasion, vibration, and humidity etc.), and therefore see more failures, and have no redundancy, so ease of replacement becomes critical. Replacing a sensor or actuator with digital communication must be as easy as replacing one using 4-20 mA or on/off. This is not as easy as plugging in a connector. This requires an extremely well defined application protocol, far beyond for instance what Modbus/TCP does since an instrument technician cannot map Modbus registers or change the decoding of data in the DCS every time they replace an instrument with a new type or version. Over and above the physical aspect, the application protocol would be a critical piece for Ethernet to take the place of 4-20 mA and on/off.
Digital communication for sensors and actuators make sense. Young engineers ask me what the difference between two-wire and four-wire signal, sink and source, active and passive, normally open and normally closed etc. The new generation of engineers take digital communication and software for granted, finding hardwiring and multimeters old fashioned.
That is, H1 fieldbus and Ethernet are not competing. They are complementing each other:
Axiom: “For every fieldbus there is a corresponding and complementing industrial Ethernet application protocol”
The way I personally perceive their direction, the bus organizations are all aligned on the application of fieldbus and Ethernet: Neither organization says that sensors/transmitters and positioners/actuators should use Ethernet. At Purdue level 1 they use fieldbus (H1, PA, IO-link, and CompoNet) to take the place of 4-20 mA and on/off signals. Since these devices don’t come with Ethernet, there is no overlap or competition here.
All organizations suggest that Ethernet can be used as an alternative to RS-485 at level 1-1/2 of the Purdue reference model; drives, remote-I/O, gateways, package unit PLC etc. There is some overlap at this level. This is the “H2 fieldbus”, and it doesn’t really run in the “field”, it runs in the auxiliary equipment room. So the level 1-1/2 and level 1 protocols are:
PROFIBUS / PROFINET:
- PROFINET / PROFIBUS-DP
- PROFIBUS-PA and IO-Link
- EtherNet/IP \ DeviceNet
- 4-20 mA/HART and WirelessHART
- Modbus/TCP \ Modbus/RTU
So I think Ethernet and fieldbus complement each other in all these architectures. It is just like your PC; at the lowest level your mouse and keyboard etc. use USB or Bluetooth (they are the input sensors) but at the higher level it has Ethernet and Wi-Fi carrying TCP/IP. You need both. So fieldbus and Ethernet are complementing each other; fieldbus at level 1, and Ethernet at level 1-1/2 and above. Together they make the plant digital from sensor to actuator. You need both.
Indeed you could even speculate that it is the adoption of H1, PA, IO-link, CompoNet, and WirelessHART for sensors/transmitters and positioners/actuators at level 1 which generates so much more data than analog 4-20 mA and discrete on/off, that this in turn drives the need for Ethernet at the next level to aggregate all this data. One is FUELING the other! Hardwired devices have 1 I/O signal, but devices on a digital bus have an average of 3 real-time I/O signals with status signal plus diagnostics and setup information. The future is digital; so ultimately I personally see the combination of fieldbus and Ethernet as the path forward to “Digital Everywhere”, with intelligence in every device and digital closed loop control, from sensor to actuator without analog signals, and intelligent device management software part of the asset management system enabling a “Check the software first” mindset.
The Ethernet or fieldbus debate has been going on since 2000 or earlier: