Without diagnosis no therapy. This is valid not only in medicine but also in engineering. If malfunctions occur at machines, plants or electronics, service technicians are called for. It is their task to re-establish the target state – comparable with paramedics that are concerned about the health of their patients in acute emergencies. In order for a rescue mission to be successful, the emergency diagnosis must be correct. Just as measuring devices provide information on vital statistics of the human organism such as electrocardiogram, blood glucose level and so forth, there are similar "quality values" in industrial networks as well. The technician speaks of physical and logical parameters – depending on whether they refer to network device, network infrastructure or network communication.
Physical vs. logistical network diagnosis
A proper network diagnosis includes both types of parameters. That is why one also differs between a physical and a logical diagnosis. An example: If a malfunction occurs in the PROFIBUS network, the troubleshooting begins with the logic. Suitable diagnostic devices are integrated into the network for that. They allow for an analysis of all status-relevant events. This includes telegram repetitions, error telegrams, device failures and restarts, cycle times, bus speeds and so forth. Also the data of diagnosis-capable network devices – such as repeaters or intelligent measuring points – flow into the evaluation. A professional cable tester is employed as well for the physical diagnosis that checks parameters such as impedance, cross talk, cable length, bending radii and termination and precisely localises the weak spot(s) in the network. At this time point at the latest, the repair technicians and service employees should check the equipotential bonding system of the plant to rule out electromagnetic incompatibilities as sources of interference. Only then can you ultimately take the correct measures for rectifying faults.
At this point, many further examples can be listed to illustrate the process during network diagnosis. We are content though to point out that the relevant quality values deviate from technology to technology and respectively fitting hard- and software are required for the diagnosis. PROFIBUS and PROFINET are rather the rule here than the exception. Whereas in the former the causes for quality problems are predominantly of a physical nature, the focus on the latter is primarily on logical parameters such as update rate, jitter and network load: failed data updates and delayed telegrams are indicators of weak spots in the PROFINET just as fluctuating load peaks that endanger the maintenance of the update rate.
Ignoring the technological differences, four basic questions remain that are asked during every network diagnosis: What should be measured? How should it be measured? Where should it be measured? With what should it be measured?
Network diagnosis as a routine
Network diagnoses serve not only for targeted fault rectification in case of a malfunction but are also routinely conducted, for instance during inspections, revisions or in the course of new installations. The main reason is to determine the respective status quo in order to derive targeted measures from that for planners, installers or service technicians. In a newly installed network the diagnosis is even obligatory so that is can be accepted properly. Included in an acceptance is an online and offline test of the communication quality in which the network is tested physically as well as logically. Following that, the measurement results are evaluated in order to carry out any improvements of the network stability. The concluding measurement certificate certifies the readiness for operation of the system.
Permanent network diagnosis
If planners and operators think of maintenance in longer terms, maintenance does not need to be carried out anymore due to events, but can be carried out according to needs and with foresight. The requirement for such a condition monitoring is that measurement and diagnostic devices remain constantly in the network so that their data on the quality of the physical and logical parameters are kept up to date during ongoing operation. With this permanent monitoring, you make an early diagnosis of weak spots in the network possible and thus make condition-based maintenance plannable. The necessary information can be collected actively and passively:
- Active network diagnosis
The active network diagnosis determines important, network-describing parameters such as IP-/MAC addresses, software/hardware statuses, port statistics/port faults, line lengths, attenuation reserves or the topology. Keep in mind though that broadcast telegrams are often used to collect information (on other words: "queries" to all devices) which cause a relatively high network load as a consequence.
- Passive network diagnosis
The passive network diagnosis determines important, network securing parameters such as jitter, load ratio, bus load, data throughput at the port of the controller and the such; whereby the collecting of the information does not create any additional network load. The diagnostic devices only "listen in" instead of sending actively and thus can be ruled out as potential sources for faults. Important for an optimal diagnosis is that the devices are placed at the backbone connection so that they can monitor the complete telegram traffic between the PLC and the field.
The comparison shows: Only in the interaction of active and passive diagnosis does a nearly complete image of the network status result.