Ethernet adoption in the test and measurement and general-purpose data acquisition markets, no doubt, is increasing. The number of devices with built-in Ethernet and related web-browsing capability also is increasing.
But is Ethernet ready to wholesale replace traditional serial (RS-232, RS-422/485) solutions? Also, with the introduction of both CAN-based device-level buses and high-speed peripheral buses including the universal serial bus (USB)/USB2.0 and IEEE 1394 (FireWire), where does Ethernet really fit?
Serial Ports Still King
In spite of the introduction of faster CPUs and the availability of new, high-speed communications and networking options, many market research surveys have confirmed that the good old serial port remains the most widely used interface for data acquisition today. It also is likely that the use of the serial bus will not significantly diminish with the sharp increase in Ethernet adoption. Why? Because for many test and measurement applications, Ethernet still is more an enabling technology today than a replacement technology.
For example, adding legacy serial devices to Ethernet is possible using inexpensive gateways. The result is a hybrid Ethernet/serial architecture extending the application range of serial devices by making them Ethernet-enabled. Now a PC with built-in Ethernet, COM ports, and USB functions as the center of a powerful, open-architecture system that integrates a wide range of hardware including dataloggers, distributed (DIN-rail) I/O modules, and a host of other data acquisition peripheral devices.
Next-Generation Buses
Common as the serial bus is, there are clear limits to its application range. As computer manufacturers move toward slotless or closed PCs, the need for an external bus with a broad application range is obvious. When bandwidth, distance, diagnostics, power, and other factors are taken into consideration, it is prudent to explore other available communications options. The payoff of more open-system architectures is the capability to pick the best bus/network and data acquisition hardware to match application requirements.
The USB standard was introduced in 1995 and now is supported by a consortium of hundreds of developers who are part of the USB Implementers Forum. Initially, it was developed with the backing of Intel and Microsoft for conveniently connecting peripheral devices such as printers, monitors, and modems to PCs.
The USB has some significant advantages over conventional serial and parallel communications. These include higher bandwidth (up to 12 Mb/s), easy system expansion with plug-and-play devices, multiple devices on a single port, and available power for peripheral devices.
These benefits suit USB for data acquisition applications needing a local (5 meter max without hubs/repeaters) network that is fast and both easy to connect and expand. USB also is ideal for portable data acquisition since, unlike less robust PCMCIA card solutions, most of today’s commercially available USB-based data acquisition devices feature seamless channel expansion and resident screw terminals. With the introduction of USB2.0, most notably with higher bandwidth at 480 Mb/s, the already strong adoption of this bus is likely to increase.
IEEE 1394 also is an emerging standard for connecting peripheral devices to PCs. Like USB, it allows multiple devices to be connected to the PC, supports plug-and-play and hot swapping of devices, and provides power.
The major benefit of IEEE 1394 is its speed (400 Mb/s), making it suitable for high-bandwidth applications. Common applications for IEEE 1394 include data transfer to mass-storage devices and support for imaging/vision devices for production-quality monitoring.
Why Ethernet?
It is not likely that a single bus will soon meet every application requirement. However, Ethernet is best positioned to meet the broadest application range. It has the best chance to function as the open network with potential for universal connectivity networking all levels of an enterprise.
Ethernet’s specific benefits are well documented, but here are a few of the more important ones that relate to general-purpose monitoring and control:
- Low Cost—If not already built in, adding a network card to a PC is inexpensive, and prices continue to drop. Hubs, routers, and other commercial accessories also are universally available.
- High Bandwidth—10 Mb/s, 100 Mb/s, and recently introduced 1 Gb/s are provided.
- Flexible Wiring—Options are available for virtually any application including twisted pair, coax, fiber-optics, and wireless.
- Interconnectivity—Ethernet can support multiple protocols and network alternatives simultaneously (based on IEEE 802.3).
- Software Support—Windows/NT-based SCADA/HMI software like CiTect provides easy-to-use Ethernet system configuration and expansion.
Building a System Architecture
Traditionally, most PC-based data acquisition systems ran as stand-alone applications with limited networking capability. Basic systems consisted of a single PC connected either to local serial (RS-232/422) devices or populated with PC plug-in (ISA/PCI) I/O boards (Figure 1). This is the classic data island. However, as applications have increased in complexity with the need for higher I/O count, more distributed I/O, higher throughput, and increased data sharing, this architecture is inadequate.
The economical and practical approach is to find a way to enable existing installations to gracefully evolve into more enterprise-level systems. Since most facilities already have an installed Ethernet network and since today’s PCs include Ethernet and the associated drivers, this integration task is greatly simplified.
By adding an inexpensive and easy-to-configure Ethernet card to the PC, data acquired from this station now can be shared with other PCs on the network. The result is a hybrid Ethernet/serial architecture that preserves your initial investment.
This is an adequate solution for many applications, but certainly not for uses characterized by low channel count (0 to 32) or a high variety of signal types per node. For these applications, an Ethernet backbone is the foundation for a highly distributed architecture.
For example, with IOtech’s PointScan I/O with embedded Ethernet, you can add distributed I/O modules to the Ethernet backbone (Figure 2). Since these modules also feature a fully isolated RS-485 port, up to 512 PointScan distributed I/O modules that extend Ethernet’s effective distance from 100 meters to half a mile can be configured from a single IP address. This is a good example of a best-of-both world’s hybrid Ethernet/serial system.
What about high channel-count systems typically with 64 to 500+ channels on a single data acquisition node? Here Ethernet is assisted by gateway devices used to convert Ethernet to RS-232 or RS-422/485. Gateways make it possible to place legacy devices that normally have a local serial connection to a PC in an isolated location and communicate remotely with the PC via the Ethernet (Figure 3).
What About Performance?
Ethernet is neither truly real-time nor deterministic, but with a little creativity you can overcome these limitations. Let’s start with a benchmark example. A system composed of five stations with 200 analog and 400 discrete I/O points each can meet a polled throughput rate of 50 ms while still using only one-fifth of the 10-Mb/s network. However, to guarantee this performance, you must decouple the data acquisition and control subnet (lightly loaded with less opportunity for collisions) from the business-wide network.
For large channel-count applications where values do not change frequently, PointScan’s report-by-exception mode can be used to achieve faster response rates, typically in the 20 to 30 ms range—independent of module count.
Another way to guarantee real-time deterministic performance is to add remote processing capabilities. For example, a cPCI chassis configured with a CPU runs a real-time operating system/RTOS controlling resident I/O. Since cPCI CPUs typically feature dual Ethernet, USB, IEEE 1394, serial ports, and a parallel port, it is possible to support an additional system.
Hybrid system architectures are becoming the standard in factory monitoring, production test, and laboratory data acquisition systems. They provide the versatility to meet current and future application requirements while preserving the initial hardware investment (Figure 4).
Consider This
The Ethernet standard, in spite of its clear benefits, is not perfect. Table 1 (see below) summarizes some of Ethernet’s characteristics to consider when comparing Ethernet to your application needs. Many of these can be overcome, and others are being addressed by the further evolution of Ethernet-related standards.
Table 1. Characteristics of Ethernet
The Future of Ethernet
Ethernet is an evolving standard, and continuing enhancement will focus on current limitations, increasing its addressable applications and subsequent adoption rate. Examples of recent enhancements include the following:
IEEE 802.3x—Supports bidirectional, simultaneous transmission of data over full duplex supported media.
IEEE 802.3z—Accommodates high-speed, gigabit Ethernet.
Even with all its benefits and planned enhancements, Ethernet is not without some limitations. For example, for harsh industrial applications, more robust connector options are required, and there still is no resolution regarding which protocol (Modbus/TCP, EtherNet/IP, ProfiNet or Foundation fieldbus/HSE) will emerge as the standard for the application layer of the seven-layer OSI model. However, existing communications standards now are finding a common ground based on Ethernet connectivity.
Conclusion
Ethernet can function as a viable backbone for data acquisition either with Ethernet-only devices or in conjunction with legacy serial or other interface-based devices. Also, the present limitations of Ethernet can be overcome with a little creativity and the use of commercial off-the-shelf hardware and software. The best course of action is to understand your current and future application requirements, look for the first justified Ethernet application, and dive in to integrate your existing and new applications into an Ethernet network.
In short, match the right bus to your needs, and make certain Ethernet is in the equation. If Ethernet is not everywhere today, it very likely will be in the next two to three years.
About the Author
Ron Chapek is the new business development manager at IOtech. He has more than 15 years experience in the test and measurement and data acquisition industries, including positions in sales, product management, and product marketing with companies such as Gould Test and Measurement and Rockwell Automation/Allen-Bradley. IOtech, 25971 Cannon Rd., Cleveland, OH 44146, 440-439-4091, e-mail: [email protected].
Return to EE Home Page
Published by EE-Evaluation Engineering
All contents © 2001 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.
April 2001