Capacitance Analysis
Capacitance: Theoretical Foundations
Overview
Professor, how is capacitance calculated using FEM?
Capacitance is the ratio of charge to voltage between conductors.
In FEM, we solve Poisson's equation for electric potential and calculate the charge on conductor surfaces by integrating using Gauss's law.
Typical Analytical Solutions
Are there any shapes we can compare with theoretical formulas?
| Structure | Capacitance |
|---|---|
| Parallel Plates | $C = \varepsilon A / d$ |
| Coaxial Cylinders | $C = 2\pi\varepsilon L / \ln(b/a)$ |
| Concentric Spherical Shells | $C = 4\pi\varepsilon ab / (b-a)$ |
| Isolated Spherical Conductor | $C = 4\pi\varepsilon_0 a$ |
The golden rule is to verify the validity of FEM results with these before moving on to real-world problems.
Capacitance Matrix
How is it handled for multi-conductor systems?
The capacitive relationship between $n$ conductors is described by the capacitance matrix $[C]$.
The diagonal components $C_{ii}$ are self-capacitance, and the off-diagonal components $C_{ij}$ are mutual capacitance, which take negative values. This capacitance matrix is essential for evaluating crosstalk between PCB traces. Ansys Q3D can extract it automatically.
The Leyden Jar—Humanity's First Capacitor Was Born in the Netherlands in 1745
The Leyden Jar is humanity's first capacitor that embodies the concept of capacitance (capacitance). It stores static electricity inside a glass jar covered with metal foil on the inside and outside. In 1746, Franklin used this to conduct his kite experiment, proving that lightning is electricity. The essence of modern capacitance theory $C = Q/V$—"how much charge can be stored for a given potential difference"—was inspired by observing this jar.
Computational Methods for Capacitance
Details of Numerical Methods
How is the capacitance matrix calculated using FEM?
For a system with $n$ conductors, we solve Poisson's equation by applying 1V to conductor $j$ and 0V to the others. The charge $Q_i$ on each conductor surface directly becomes $C_{ij}$. This is repeated for all conductors.
Doing $n$ analyses sounds tough.
We can reduce the number of runs by utilizing the symmetry $C_{ij} = C_{ji}$. In Ansys Q3D, this operation is fully automated.
Energy Method
Are there methods other than charge integration?
There is the energy method, which calculates capacitance from electrostatic energy.
This can be calculated using COMSOL's "Volume Integration." Calculate using both the charge method and the energy method; if the difference is within 1%, it's reliable. A large difference indicates the mesh is too coarse.
Structures with large fringing fields are prone to insufficient meshing, right?
Exactly. When the trace width and gap are comparable, fringing capacitance can account for 30-50% of the total. The parallel plate approximation $C = \varepsilon A/d$ alone is insufficient.
Why the "Parallel Plate Capacitor" Never Disappears from Textbooks
The formula for the parallel plate capacitor $C = \varepsilon_0 \varepsilon_r A / d$ always appears as a fundamental of capacitance analysis. You might think, "Such a simple shape doesn't exist in reality," but semiconductor gate oxide layers, PCB power planes, and interlayer laminations in flexible boards are essentially parallel plates. In actual design, edge effects (fringing effects) cause 10-30% error, making correction by CAE solvers essential. The textbook formula is merely a starting point.