Modellierung und Simulation ausgedehnter Massestrukturen (English shop)

To ensure electromagnetic compatibility within systems, precise knowledge of interference sources, interference sinks and their coupling is required. In systems whose components (devices) share common return conductors or a common ground, these constitute a major coupling path for interference. In aviation, for example, the aircraft fuselage is frequently used as a common ground for entirely different circuits in order to save weight. In other fields as well, mechanical structures are often used as ground.

In order to reduce the costs of building prototypes, complex electronic systems consisting of many components are first simulated. The simulations are often carried out using simulation programs based on networks. It is a great advantage if the interference coupling through the common ground can be included in this simulation, since this means that the electromagnetic compatibility of the mains feedback effects is already taken into account during the design process.

The prerequisite for integrating interference coupling via common ground structures into the system simulation is the availability of a suitable model:
- The model must not be too complex, since a substantial increase in simulation effort (time, resources) will not be accepted by the system developers responsible for carrying out the simulation.
- The model should be relatively accurate. Interference coupling must under no circumstances be underestimated, so that developers do not wrongly dispense with interference suppression measures on the basis of the simulation.
- The model should be flexible enough to allow changes in cable routing or connection points.

In principle, these requirements are mutually incompatible. For a very accurate AC simulation, for example, the routing of all circuits must be taken into account. If an AC model is to permit arbitrary routing, it can therefore either not be accurate, or it must be very complex. This work presents new methods that generate appropriate models depending on the actual weighting of these requirements.

The focus of this work is on modelling existing mechanical structures. Here the electrical properties are frequently not sufficiently known, because only static aspects are considered during design and implementation. For example, components are sometimes first painted and then joined, or vice versa, without this being documented. Classical methods must rely on estimated values here. All the new methods presented here use measured data.

Several methods are based on equivalent networks which essentially provide one or more network elements for each part of the structure. For the DC case, a method is presented that derives a resistance model from a structural analysis and a small number of measurements. For the AC case, this model is extended with inductances. Owing to their underlying principle, these AC models can only provide an estimate of the real interference coupling. In the experiment, however, good agreement between measurement and simulation is found.

A further class of methods is based on specially developed measuring instruments, the modular network analysers. The models can be generated automatically from the measured data. The result is an N-port model that can be mapped onto very simple equivalent networks (macromodels) with little effort. Analysers were developed in both the time and frequency domains.

The well-known PEEC method by A. Ruehli is applied in a simplified form and analysed according to the same criteria as the modelling methods developed by the author.

All methods presented in this work were verified on the same experimental setup by creating a model of the setup according to each method. The models were used for simulations in the time and frequency domains and the results were compared with direct measurements. In all cases, good agreement was obtained.