Rocking the EMC boat
[tabs] [tab title="English Article"]
As long as I have been active in industrial networking, cabling problems have played a role in the reliability. In the early years of the industrial networks, most attention went to the technology itself. It was new and unknown, and confidence in 'software' for controlling I/O, instrumentation and peripherals was very low. This lack of confidence was justified - growing pains in the network protocols themselves, errors in deployments with implications for interoperability, and a lack of knowledge among end users and suppliers often resulted in problematic networks.
But all that has changed. The errors in the network protocols have been solved and user groups created test labs to prevent interoperability problems. Today it is much easier to connect products from different vendors together. Also, users have gained experience and know that they can rely on industrial networks. However, there is still one aspect of networks that specification documents, test labs and vendors cannot control: the cabling. Even with the best protocols, best hardware and bug-free software, ultimately the cabling is installed by the user. If this is not done properly, the network will not work correctly and thus also the application will suffer.
Each network dictates how it should be wired: the topology, usage of terminating resistors (or not), electrical properties of the cable, the relationship between speed (bitrate) and length, etc. Some aspects, however, continue to call for discussion. For example: must I use shielded cable? In almost all
networks this is mandatory, however not so in AS-Interface. Must the shield be grounded at one end, or both ends? And how must that be done, fully covered, or is a “pigtail” allowed?
Having no electrical background myself, I attempt to implement the advice given to me by those who know more about this. But it is confusing that the advice of expert A can be completely different than the advice from expert B. Who am I to believe?
During the last Dutch National Profibus event in Eindhoven, the last lecture of the day handled EMC (Electro-Magnetic Compatibility) in industrial networks. On behalf of Prokorment, Rene Heidl of Indu-Sol gave an impressive demonstration about the “do’s” and “don’ts” of various EMC measures. The fact that the subject had the interest of the visitors was clear – usually the last lectures of any seminar are not very well attended, but in this case this was different. The room was packed to capacity with more than 80 visitors, and the presentation + demo was rated very high.
A special test system with 20 metres of ProfiNet cable, accompanied by one additional wire connected to a 600W “noise generator”, was used to show the pros and cons of various EMC measures such as screening, using twisted- pair cabling, and various ways of grounding. Because the presentation had to be done in half an hour, the experiments followed in rapid succession, and Heidl told his story with much flair, keeping the attention of the public until the last second.
It soon became apparent that incorrect implementation of good advice may, after having spent a lot of money and time, result in a network that functions worse than before.
A person in the audience asked about the common advice to avoid pigtails. Heidl replied that this is nonsense, as the pigtail can be seen as a (short) antenna and therefore picks up only disturbances of a very high frequency, related to the length of the pigtail. In his view, therefore, it does not make sense to invest in expensive fittings.
Indu-Sol of course provides the knowledge and equipment that can help to find and solve grounding and shielding problems in an industrial network. Usually this is done with (software-based) network analyzers, but these only show indirect electrical problems: because network messages are damaged, missing, or retried. Indu-Sol tools directly measure at the source – with current clamps on the grounding, the screen, and the signal wires. Four probes can be connected to a laptop, where software runs to show the measurement results. As was shown in one of the demos, a change in the system’s grounding wiring had no positive effect (picture). The fault voltage (red signal) remained the same, the interference voltage (blue) only slightly decreased, but the other interference voltage (yellow) became disproportional larger.
Unfortunately, half an hour was too short for this very interesting topic. This is why Prokorment and Indu-Sol organize free four-hour seminars in which the EMC aspects of industrial networks are explained in more detail.[toggle title="Information" state="open"]
- Email email@example.com