This article was printed in CONTROL's November 2009 edition.
Is wireless easier to integrate with legacy systems than fieldbus? Certainly, one of the criticisms of Foundation fieldbus and Profibus was that users had to replace wholehog their 1990's-or-earlier vintage DCSs or PLCs to integrate them with one of the newer bus protocols. A lot of sophisticated users have very capable, robust and well-supported platforms that predate the widespread use of fieldbus. Since wireless and especially, WirelessHART have emerged as viable options for integrating intelligent devices, some users have been pleased to find that wireless connects easily with their old system.
[pullquote]This has not always been the case, but today, Modbus/Modbus-over-IP and OPC-based implementations are common. Emerson has released a proprietary solution that connects its Rosemount 1420 wireless gathering points directly to the DCS Ethernet backbone, and has a proprietary redundant solution as well for WirelessHART. Honeywell's OneWireless R120 serves both WiFi 802.11a-b-g clients and the ISA 100.11a field device network, but integration with non-Honeywell PCNs would appear to be limited to Modbus or OPC. Meanwhile, developers at the Fieldbus Foundation are working on a secure, open "backhaul" solution—also using Ethernet/Fieldbus HSE—that would serve WirelessHART, ISA 100.11a-based solutions, and open, wired, remote I/O solutions common in Profibus-capable platforms.
So, it's not surprising that a user might write, "I think wireless will continue to grow because it integrates so easily with existing platforms." I read that the other day on the Foundation Fieldbus LinkedIn group, and it made me think: Is integrating fieldbus that different?
You can't buy a fieldbus H1 I/O card for your Honeywell PM (Process Manager) I/O, nor for legacy APACS, Bailey (ABB), Centum-XL orProvox. Wireless connects to these systems typically by way of Modbus or a serial interface card. End users can roll their own redundancy solutions in some cases, but the serial data is still subject to the vagaries of unknown latencies and Modbus mapping/scaling. It's like remote I/O, and many users in the large process industries wince at the thought of using any of it in process control—wired, wireless or otherwise. Nonetheless, some users are seizing the opportunity to add wireless gathering points and installing numerous wireless field devices.
The WirelessHART gathering point, the 1420, has a fieldbus-capable brother that's been around for years, the 3420. Although they are limited to four H1 segments, or 64 total field devices, many of these devices can be multivariable, so the total number of measurements or indications can be nearly an order of magnitude higher. The 3420 has web-server, OPC-server and Modbus-RTU capability just like the 1420. The web browser is used for configuration and Modbus mapping (if using the Modbus interface) just like the WirelessHART appliance.
Unlike wireless, you can add control valve positioners to any segment and perform closed loop "control on the wire" with macrocycles of 500 ms or less, depending on segment loading. The greater the function block population, the greater the required macrocycle, just like host-configured FF segments. These field-executed segments can be configured to keep running as long as there's power on the segment, even if the gateway ceases to function.
You can sample measurements at sub-second rates without concerns about battery life. The batteries employed by wireless devices are not inexpensive, especially those packaged for hazardous area applications.
Competing products, such as the Softing FG100, Smar DF51 and its kin available from E+H and Rockwell, are out there as well, which makes me think that claims of "easier integration" for wireless devices may be more hype than reality. Users should examine whether wired fieldbus solutions have comparable capability for connecting modern intelligent field devices to their aging host platforms.