Preparation neutralizes network problems before they grow into a full-blown crisis.
If you wait until a crisis strikes before you consider how to reduce the frequency and effects of network problems, you'll be too late. Instead, take a proactive approach and implement a strategy geared toward preventing network problems before they have a chance to shut down your system. If you pay attention to the physical aspects of the system and the way it transfers data, you'll be closer to achieving a failure-free network.
Start with the physical installation.
Improper grounding is a major contributor to network failure. In an effort to reduce noise, many installers devise grounding schemes based on unsound theory. Regardless of what the manufacturer's installation manual says, you must ensure your network conforms to IEEE Standard 142 and NEC Art. 250. If your warranty is bound to a configuration that violates these standards, work with the manufacturer — and possibly your legal department — to resolve the issue. Otherwise, you not only risk personal liability but loss of warranty, as well.
Unbonded grounding electrodes (misapplication of “isolated ground”) and ground loops (often due to neutral-ground bonds at the equipment) are the two most common grounding errors in these systems. If these design errors exist in your system, you have a personnel danger that needs immediate repair. And even a properly designed system can have broken wires and poor connections, so give the whole system a thorough inspection. Cable connectors are the most common weak points in a networking installation. Because cable connectors expose unshielded wire, they leave communications susceptible to environmental factors such as radiated noise and corrosive chemicals.
Environmental noise can enter a system through several key points, such as the chassis, I/O connections, or communications lines. While walking down your network, list the potential noise sources and note their severity. You must determine what noise sources are potentially most damaging to the system operation. High-frequency interference and transients, such as lightning and electromechanical interference (EMI), are examples of noise that can cause or complicate problems. It's possible to solve noise problems with such mitigation techniques as special shielding and careful routing, or you may cope with noise problems by de-rating system performance.
Moving beyond the physical.
The physical installation is not always the source of network troubles, but it must be correct to avoid problems. Once you've ensured it's correct, you are ready to address other potential problem sources.
Consider performance and bandwidth issues, such as event-basing communications, messaging format optimization, detecting and handling communications loss, and splitting networks. When sending data, base the transfer on an event that indicates fresh data are available. For example, if your refresh rate is every 5 sec., it is not necessary to read data three times per second. Reading the same data multiple times wastes the network communications bandwidth and robs other devices and users of this limited resource.
If you effectively manage data flow, you may find it unnecessary to upgrade to equipment that provides faster data rates. It's wise to calculate approximate communications thresholds by knowing the amount of data, how frequently transfers occur, and device-buffering limitations. Once a system is in place, measure the network performance empirically and become familiar with how much bandwidth users actually consume. Being aware of these factors will help determine where a problem originates and how you might resolve it.
Although troubleshooting tools have long been available for networks like Ethernet, similar devices now exist for plant-floor buses. The following tools can help you determine when physical network problems are present, provide information on signal levels of the network, and confirm that devices are sending data that meet the protocol specification:
A time domain reflectometer (TDR) identifies faults or points of discontinuity via a reflection.
Passive network tools (Photo on page 58) go beyond the capabilities of a TDR by offering the ability to detect split pairs, extract cable parameters, and determine wire-mapping errors.
The digital multimeter (DMM) measures resistance on a cable and detects broken wires.
Beyond the cable, devices like an active network tool allow you to measure address-specific signal levels and quality on the network. Such a device also provides feedback on specific nodes that do not like the signal or data being received.
The most sophisticated tool, however, is a network traffic analyzer. This device records exactly what is happening on a network, which devices are communicating, the frequency of their communications, and specific details of their communications for a given period. You can analyze the details off-line to determine problems in network traffic.
Your suspicions about what is happening on the network are often wrong. To provide insight, analyzers show the biggest bandwidth users and display the actual data flow between devices. While one node may claim to send data, the other will reply only if the sent data use the protocol perfectly. In this case, the analyzer provides proof that data are on the wire and are flawless. However, few people use analyzers, unless they are developers or network specialists trained to read the results.
How to get help.
Many resources, such as user manuals, Web sites, and industry associations can provide valuable information when problems occur. You need to look, investigate, and record what you find. Maintenance people often call vendor tech support to say their system has stopped working — but fail to provide any critical information, like error codes or system diagnostics, which burns valuable time and does little good. Before calling vendor support, always view the status indicators on a device — which will clue you in to the nature of the problem. These instantaneous, easy-to-read alarms can help you avoid more detailed and time-consuming troubleshooting steps.
Additional outlets for help.
Network station diagnostics are also useful because they provide a point of reference to characterize problems. A network with several token-related errors, for instance, can mean performance problems exist or a device is dropping off of the network by not accepting the token. Understanding the diagnostics gives direction for expert troubleshooting and helps to identify the problem.
Taking advantage of product updates is important, because more recent revisions of products can offer faster data-transfer rates and support additional communications protocols — giving you more options when trying to solve plant-floor troubles. Keeping information like catalog or model number, series, firmware, and hardware and software revisions readily accessible prepares you for effective troubleshooting that may lie ahead.
Putting it all together.
With the numerous and varied forms of help available today, there are few reasons for you to suffer through a networking nightmare. The key to keeping your plant-floor networks operating is awareness of available resources. Each tool available will help you understand your current network requirements, characterize behavior, and plan for future growth. In short, education and awareness give you the ammunition needed to fire back at networking problems.