By P. Hunter Vegas, Avid Solutions
This is Part II of a two-part article. Part I in the Q1 2011 issue detailed what fieldbuses can do compared with the claims made about them. You can read both parts now at www.industrialnetworking.net/fieldbus11q1.
Based on Part I of this article, you might wonder if fieldbuses are worth the effort. The answer is a resounding "Yes"—in the right application. It also can be an equally resounding "No" in the wrong situation.
This time we'll explore the major fieldbuses and address situations that favor and disfavor each. The high-speed network communications buses (Profibus-DP, EtherNet/IP, ControlNet and others) are not included because this article concentrates on the more common, low-level buses that communicate directly with the field devices.
Actuator-Sensor Interface (AS-i)
AS-i is among the simplest networks to implement. It is generally used to communicate with digital devices (solenoids, pushbuttons, on/off valves, etc.). It has some limited ability to transmit analog information, but rarely is used for that. It is a simple network, has a reasonably high speed (167 Kbps, <5 ms cycle time), and is insensitive to electrical interferences. The major limitation of the network is length. The total cable length of a network must be limited to 100 m. However, the recent introduction of low-cost network repeaters and power conditioners allow a single AS-i power supply to power multiple 100 m segments of the same network.
AS-i favors any process that involves numerous digital devices in a relatively small area. A batch reactor is often a great application for AS-i because the reactor usually has a large number of on/off valves controlling the jacket and raw material charges in close proximity to the reactor. Because each valve has at least one solenoid and two limit switches, wiring savings can be significant.
AS-i is best used when the control system has an integrated AS-i card. It is possible to install an AS-i card in third-party remote gateway racks (via Profibus-DP, DeviceNet and others), but that arrangement makes engineering more difficult and slows valve response time.
It is good for non-classified, Class I Div 2, and Class I Div 1 areas. Recent introductions in AS-i wiring equipment make conduit routing simpler and more cost-effective, especially in electrically classified areas.
Network "tuners" also have been introduced recently to extend the network beyond 100 m by automatically optimizing the network capacitance and termination resistance. Its network cable has no particular topology requirements, so it can be branched as necessary to accommodate field equipment.
AS-i Profile files don't change often, so old and new equipment are generally interchangeable, I/O modules can be wired to pushbuttons and/or power lights to handle local operator stations, and the networks are rated to 8 A. This provides much more available power than most fieldbus networks. Configuration of a device is simple; just assign it a node number.
It's Not Always Right
AS-i is designed for on/off devices and is ill-suited for analog instrumentation. If the process equipment is scattered across a plant, the 100 m limit will prove too constraining even with repeaters. And due to the 100 m limit, you must know where the valves are going to be installed, often before the plant is built. If equipment moves, the network design must be rechecked.
AS-i valve heads cost more than standard on/off heads with limit switches. The valves tend to use low-power solenoids, so if the instrument air supply is poor, the solenoids could become a maintenance issue.
Like any network, getting the first valve or device to communicate can be a challenge. However, once that valve is established, configuration only involves hooking up an AS-i handheld to the device and giving it a node number. Checkouts are extremely quick.
Spotlight on DeviceNet
DeviceNet is based on the CIP protocol and typically is used for digital devices and motor drives. It is a deterministic, high-speed (500, 250 or 125 Kbps) network that also provides power on its multi-conductor cable for field devices. It is a more complex network and will be more difficult to set up than AS-i, but it handles a much larger variety of data. The allowable length of a DeviceNet network is speed-dependent, but the length limitations are not nearly as constraining as AS-i.
The best applications for DeviceNet are motor control centers (MCCs) or discrete manufacturing lines where a single bus cable can run along the system with short drops for each device. Cable distances usually are not a constraint, so knowing the precise equipment location in advance is not critical, and equipment can be moved without impacting the network significantly.
DeviceNet 120 Vac motor modules can be a good way to solve issues with NEC Arc Flash requirements. Running the 110 Vac motor control leads to a cabinet with the DeviceNet devices lets electricians make voltage checks on live motor systems without exposure to 480 Vac.
I/O modules can be wired to pushbuttons and/or power lights to handle local operator stations. This network is generally rated to 8 A and provides much more available power than most fieldbuses.
An Auto Recovery feature allows a DeviceNet device to be replaced and automatically downloaded with the proper configuration when it is put on the network. There are some limitations, but this feature can shorten system downtime—if the replacement device is an exact duplicate.
Numerous vendors have introduced many new DeviceNet-compatible devices. This generally has driven down the prices of the field devices considerably.
I know of two users who have experienced "jabber' from a single DeviceNet device that can interrupt communications on the entire network. This is not a common occurrence, but designers should separate the network into manageable areas that will not badly impact production should a failure occur. In addition, very few DeviceNet devices are rated for hazardous locations, so DeviceNet is not often used in classified areas.
On most DeviceNet systems, you cannot add a device without downloading the network and taking it out of service. Generally this is not a problem on a discrete manufacturing line, but it can be completely unacceptable to take down an entire MCC worth of motors in a continuous chemical plant to add one motor to the network. Separate the networks appropriately to avoid downtime issues. In the case of an MCC, consider installing several spare MCC modules in advance so they can be wired into new motor buckets without network outages.
DeviceNet is a more complex network and it requires special network software to establish communications. It also involves special data files (EDS) that can complicate startups. Despite marketing claims of "plug and play," most users find very few startups go that easily, especially if untested devices are involved. It is important to establish communications with new equipment in advance of startup to avoid delays during commissioning.
Device EDS files are being updated constantly, so a replacement device might not be recognized or configurable until the proper files are installed on the system. Therefore, engineers might have to be called in for maintenance.
DeviceNet topology is a single line with drops of 6 m or less, so extensive branching is not allowed. However, given the allowable trunk line lengths, you usually can extend the main network out to an area and double back.
Troubleshooting this network can be very difficult and there are few means of diagnosing problems other than eliminating one device at a time if severe network problems occur. Engineering involvement usually will be required and this adds to lifecycle costs.
Foundation fieldbus (H1) and Profibus-PA
Foundation fieldbus (FF) and Profibus-PA are lower-speed networks (31 Kbps) that mainly target the analog process control industry. FF and Profibus-PA have been competing for years, but recently the two fieldbuses have been merging toward a common process control bus and device-definition file.
Both use the same physical layer and provide some limited power on the network lines and support low-speed, fault-tolerant communications between the control system and field devices. FF also supports device-to-device communications and allows control in the field devices, though this feature is rarely used.
Unlike FF, Profibus-PA uses a Master/Slave protocol and does not support device-to-device communications. Profibus-PA is a low-speed version of Profibus-DP and usually runs as a sub-network to a Profibus-DP communications system. DP/PA links and/or couplers bridge the two networks. DP/PA couplers reflect PA communications into the DP network so the field devices are nodes on Profibus-DP. DP/PA links separate the networks so the DP/PA link becomes a node on the Profibus-DP network and is the master of the Profibus-PA network. Links are preferable to couplers because couplers force the Profibus-DP speed to match the much slower PA network speed. The DP/PA couplers provide power for each Profibus-PA network, so additional power supplies are not needed.
Although both can support digital devices, most installations do not include them.
Advantages and Favorable Situations
Foundation fieldbus and Profibus-PA allow multiple transmitters/control valves on a single wire, and some cable savings are possible if the devices are clustered in small areas and located a long way from the control system. Both will work in non-classified, Class I Div 2, and Class I Div 1 areas. However, Class I Div 1 areas require intrinsically safe wiring, which limits the number of devices to four or five per segment, and eliminates any cable savings. Recent developments in the fieldbus intrinsically safe concept (FISCO) allow more devices on an intrinsically safe segment with the proper field connection blocks and network equipment. Cable distances usually are not a constraint on either network, and the network topology requirements are minimal.
FF support in the U.S. is strong, but support in other areas of the world is varied. Profibus-PA support in Europe is generally strong, but its support in other regions also is variable.
FF is best used where the control system has integrated FF card and system diagnostics. These capabilities can be extensive if they are integrated into the control system. If they are not, many of the advantages will be underutilized.
Similarly, Profibus-PA is best used when the control system has Profibus (DP and PA) fully integrated into the system. Profibus-PA can be run through a Profibus-DP communications card and third-party DP/PA links and couplers, but configuration is much more difficult.
FF and PA Disadvantages
Recent changes in pricing raised the software and hardware cost of FF considerably. When you include the cost of power supplies, field blocks, fieldbus cards, licensing and configuration software, and the field devices themselves, the cost can equal or exceed traditional I/O.
The installed cost of Profibus-PA is often as high as or higher than most 4-20 mA installations when you include the cost of the Profibus-DP communications, DP/PA links and/or couplers, along with field blocks, licensing and configuration software, and the field devices. However, Profibus-PA seems to currently enjoy a slight cost advantage over FF.
Process areas with scattered field devices or devices located close to the control room do not favor FF or Profibus-PA. The installed cost in such situations almost certainly will be higher than traditional I/O.
Both Profibus-PA- and FF-capable field devices are typically more expensive than 4-20 mA devices. Also, most field devices are higher-end devices with all of the options. Vendors offer very few low-end devices that are Profibus-PA-capable or FF-capable.
Designing either of the networks can be difficult because of competing priorities. Maximizing the number of devices on a network reduces installed cost but greatly increases the likelihood of plant disruption if a segment is lost. If a segment is limited to one control loop and some indicating devices, the exposure drops significantly but installed cost will be substantially higher.
Strongly evaluate failure modes when you consider an FF or Profibus-PA installation. The loss of a single communications card, power supply, network cable or link/coupler (in Profibus-PA) can take out the entire segment. If multiple control loops are on that segment, a plant shutdown could be unavoidable. There are options for redundant FF and PA communication cards and redundant power supplies, but this can be significantly more expensive.
Most FF and Profibus-PA segments must come out of service in order to install a new device on the network. This can cause serious problems for plants in continuous operation.
Both networks require much more engineering than 4-20 mA networks, and each segment must be evaluated for data transfer rates, power loading, reflections, etc.
Officially, the largest number of devices on either network is 32. In reality, the number is a great deal less than that. Failure considerations, data communication rates and power constraints reduce that figure significantly.
Installation for both networks is more technically challenging than 4-20 mA systems. Special care must be taken to make terminations properly and terminate the network in the proper locations.
FF device file revisions are common and replacement instruments might not work without some effort to find the new device file, put it in the system, and update the network.
Profibus-PA device definition files (GSD) are revised frequently, so when a device is replaced, it will often require an updated GSD definition file in order to be recognized by the system. This can require a rebuild of the network. (A Profibus-PA network is slot addressed and the network configuration file is made up of a list of the GSD files of each node. If the GSD file changes, then the network configuration file must be recompiled and downloaded to restore operation.) Both cases require a much higher technician or engineering skill set to address maintenance issues.
Troubleshooting these networks can be very difficult, and there are few means of diagnosing problems other than eliminating one device at a time if severe network problems occur. Engineering involvement will usually be required, and this adds to lifecycle costs.
P. Hunter Vegas is senior project leader at industrial process control system integrator Avid Solutions (www.avidsolutionsinc.com), based in Winston-Salem, N.C. Vegas and his colleagues have engineered and installed many fieldbus systems in many different industries. This article reflects their experiences plus those obtained from interviews of numerous end users throughout industry.