| Your position: IcestarsEM Analysis > Accurate modelling and representation |

Accurate modelling, parasitic couplings and transition

Inductor loop

The activities of the ICESTARS research on EM simulation and coupled EM/circuit analysis are supportive to the work in Time Domain and Frequency Domain Techniques. Here, in principle, we deal with the ‘communication’ of physical and mathematical layer. The mathematicians supply the electromagnetic circuit simulation to be coupled with the components’ simulation software of an industrial partner. The ever increasing miniaturisation of future circuits realized in physical models resulting in the need to simulate the electromagnetic circuits – an entirely new mathematical undertaking, realized in ICESTARS by coupling electromagnetic simulation to DAE-solvers for transient simulation. 


The research-focus is three-fold:

The first part deals with being able to extract from field-solver computations an accurate modelling of the passive elements in the circuits. Starting from a technology flow and a layout description, the compact model parameters are computed.
 
The second part deals with parasitic couplings. When an RF circuit is designed, parasitic couplings must be carefully controlled. They can manifest themselves as unintentional coupling paths. For example, the substrate is a possible medium for such couplings, and what is even more annoying, such signal can be collected at some locations of the metallic strips and injected back into the substrate at other locations. In other words, the coupling can also depend on the design. Other parasitic couplings can deal with inductor-inductor couplings that are modified by the semiconductor environment..
 
Finally, the third part deals with inclusion of the active devices in the field solver and to make the transition to compact active device models at the earliest stage possible. In order to grasp the idea, one should realize that a full field solver approach for a non-trivial circuit is not within present-day reach as the huge scale jumps inside the circuit (1 nm for gate oxides, ~mm for integrated passives) prevents a frontal approach of the simulation challenge. A solution is found in applying domain-decomposition that effectively can be mapped on replacing small-scale structures by effective compact models. This approach has been shown to be successful up to design tasks into the 60 GHz range.


EM Simulation

EM simulation is needed to model high-frequency and high-speed circuits and devices and absolutely necessary to accurately verify the interactions between devices in the RF front-end at very high frequencies. Regarding the mere scale of EHF band communication front-ends it is impossible to run full simulations at the level of detail an EM analysis offers. Therefore EM analysis only looks at a small part of the total design, while the rest of the circuit is simulated at a higher level of abstraction.

To this end, the state-of-the-art RF circuit simulation has to be adapted in order to deal with the specific demands to exactly capture the interaction between neighbouring devices. In particular the frequency response calculation requires the ability to inject electromagnetic energy in a device model.