Device-level buses such as DeviceNet, Interbus- S, or Sensoplex are well known for their labor-saving qualities. They not only reduce to manageable levels the amount of wiring required between controllers and remote sensors, but also make installation and servicing more practical as well. Nevertheless, a few common ailments occasionally crop up on these systems.
Wiring problems are by far the most typical difficulties that cause buses to go down. Good clean data line connections are mandatory. Most manufacturers of industrial data cable products use gold-plated pin and sleeve construction for connectors. When a connector sees numerous insertions and removals, or is prone to abnormal alignment, the tension in the sleeve should be spring reinforced. The cleaned and tinned wires are either soldered or crimped to the pins and sleeves in a clean factory environment. Then rubberized plastic is molded around the solder or crimp joint that seals them from the environment and restrains mechanical stress.
In most installations, perhaps 90% or more of on-the-machine wiring can take place with premade connectorized cables. But there are some lengths that cannot be set until the installation has been planned. For example, it is difficult to determine much in advance factors such as the length of the cable run from the first node to the controller, or the length between sections of a machine.
Network cabling can also be plagued by radiated electrical noise. In industrial settings, perhaps the main source of such signals today are mobile transmitters such as walkie-talkies.
Most vendors of industrial electronics learned to live with 3-W walkie-talkies in the 1980s. But the first 5-W devices began appearing in 1990, and these began bringing down regulators, switching power supplies, triggered transistors, and other kinds of industrial electronics.
Vendors have again adopted to the higherpower devices, but external sources of electrical noise in the walkie-talkie frequency band are still a major concern. In wiring, an aluminum foil with drain wire is an effective solution against wavelengths that commonly cause problems. The drain wire must terminate to ground at only one point. The theorist will say the termination should be at a midpoint on the bus. The practical person will say termination should be at the source. This usually is in the enclosure where the master node, PLC, and bus power reside. The common and overwhelming argument is that the ground point in the enclosure is easier to establish and maintain than a point on the machine.
As the noise wavelengths increase, the usefulness of aluminum-foil shielding diminishes. The most common impact of electrical noise on a bus is to garble bus data, forcing devices to resend repeatedly. Fortunately, it can be fairly easy to track down the source of the difficulty. Welders and SCRcontrolled equipment are likely culprits. Also, equipment with power MOSFETs or IGTOs could be candidates.
Other sources could be anything that makes an arc or uses electromagnetic radiation, such as RF furnaces. The most obvious place to look first is the equipment having the highest wattage. Turn off the power to the equipment and run the bus. If bus performance doesn’t improve, try the potential source with the next highest wattage, and so on. If such a procedure doesn’t identify the problem, run the bus data and power lines inside ferrous conduit for 5 m before and after the source. Ground the conduit to any available stable ground. This ground needn’t be exceptionally high quality.
Physical problems on a bus — as with connectors and wires, transceivers, or layout — generally manifest themselves in the form of nodes dropping out, the bus bogging down with repeats, or portions of the bus dying. Also, it is likely that such physical problems are present if changes in the actual location of one or more nodes cause miracle cures or mystifying problems. Many such difficulties can be found and resolved with a digital VOM meter, particularly in the absence of extensive built-in diagnostics on bus devices themselves.
For example, it is relatively easy to check for open data lines or shorted transceivers, though this may not be the most common problem. First isolate the two data lines by turning off all power to any nodes. Remove any termination resistors and leave all the nodes in place and connected. Check the resistance between the two data lines. The total resistance R is the parallel summation of all the transceivers (R1, R2, . . . RN,) given by
The CANbus specification upon which many industrial buses are based dictates that a transceiver have a minimum differential resistance of 20 kΩ and a maximum of 100 kΩ. It is then a simple matter to restate this parallel relationship in the form of a spreadsheet that solves the above equation for maximum and minimum total resistance, given the number of nodes.
Similarly, EIA RS-485 type transceivers have a 12 kΩ minimum value, though no published maximum value. The specification also limits the number of unit loads (transceivers) on a segment to 32. It is thus possible to use the same parallel summation relationship to calculate a table of minimum total resistance values for between one and 32 nodes.
When the resistance is less than the values calculated in the spreadsheet, there is probably a short somewhere. Divide the system in half, test again, and check the values. Continue to halve, test, and check the part of the system that is less than the preceding values until the problem is found.
Open data lines are more difficult to find. The values calculated in the preceding procedure may provide a hint, but the minimum and maximum vary so much that on open condition is not conclusive. If you suspect an open, measure the bus at the beginning and at the end. The resistance should be the same. If it is not, halve the system, check, and test until the open is found.
The procedure for detecting grounded data lines starts with turning off all power to the nodes. Disconnect any terminating resistors. Check the resistance to ground on each data line. It should be nearly infinite. Reverse the meter polarity in case the lines are protected by diodes. If the resistance is not nearly infinite, then divide the system in half and test again. As before, continue to halve the system until the ground is found.
It is quite possible to flip-flop the two data lines during installation, add-ons, or a physical relocation. When this happens, there will usually be a large portion of the bus that is dead.
Today, most nodes feature good communication diagnostics and indications that reveal such conditions. You could also find the problem by just walking down the bus from the beginning, and then looking at the LEDs for the first node that wasn’t communicating. Then check the connection just before the node.
In cases where these practices don’t work, begin by disconnecting the nodes from any power source. Disconnect any termination resistors. At the end of the bus, jumper one of the data lines to the drain wire. Test this data line drain wire circuit back in the cabinet at the start of the bus. It should read nearly zero resistance. If it does, test the other data line, after reconfiguring the jumper from the drain wire to the other data line. The resistance should be nearly zero. If either test results in some resistance, then the data lines are probably flip-flopped.
Unfortunately, this test can’t find two flip-flops. If this is a possibility, break the bus into sections (maybe two or three nodes per section), then install the terminating resistor at the end of the first section. Finally, power-up and check to see if the nodes are communicating. Keep adding sections until the fault is found.
Grounded or open shields can cause ambiguous symptoms, but can be checked in just a few seconds. Find the one point where the shield should be grounded, usually back at the PLC. Power down, and disconnect the shield drain wire from the ground. Then check the resistance from the drain wire to ground. The resistance should be nearly infinite. If not, divide the system in half and test again. Continue to halve and test the system until the ground is found. Remember to reconnect the single-point ground when finished.
Though it may seem obvious, momentary drops in power below the specified voltage can cause nodes to fault and go through a start-up procedure. The simplest way to make a voltage check is to install a Tee on the power bus at a point farthest from the power source. Check the bus power specifications to locate the power lines, then use a digital VOM and check the voltage while the bus operates. If the voltage is low, the power source is inadequate for the load or one or more high resistance connections are in the system, or there is too much cumulative resistance in the wire and connections.
If checks of the power source reveal both the voltage and current are below the rated output, replace the power supply. If the current is at or above the rated output and the voltage is at or below the rated output, then you need more power, perhaps a bigger supply or multiple supplies. There are many wrong ways of connecting multiple supplies, however, and such connections require some expertise.
If the voltage is at the rated output but the current is below the rated output, there is too much resistance in the circuit. The hard part comes in determining if the voltage is a gradual accumulation of resistance or just one or two bad connections. Start again using a Tee at a point farthest from the power source. Make voltage checks at each connection, working your way back to the source. Record the voltages. If there are a few bad connections, the voltage drop should stand out. If so, fix or replace the hardware at the bad connections.
If the voltage drop is gradually cumulative, use the same power source to find a way to feed the power bus at multiple points. If it is a straight-line bus, try powering it at both ends.