Power-Integrity Analysis

Home: SimulationPI  Analysis

The aim of a PI Analysis task is to compute the impedance of the power-distribution network (PDN) as seen from the selected component pins. The result of the computation are the self-impedances measured between pairs of power and ground pins. (Off-diagonal, "transfer", impedances will not be stored.) The self-impedance curves are the essential characteristics of a PDN, as they describe the relation between high-frequency voltage drop and current drawn on the component site.

The simulation set-up consists of two steps (see dialog below).

Selection tab

Net Class Manager button

The button provides a convenient way to change the net class of existing nets. This is important in order to get pins from (former) other nets into the "Available power pins" frame.

Available power pins frame

The list of components and their pins that are connected to a power net. Creation of ports happens via double-clicking, drag and drop, or right-clickcontext-menuNew port, or by using the "Create new port" symbol. You can use multi-selection. Please note that the software searches for the nearest ground  pin on the same component and generates the corresponding port.

Ports frame:

Names of ports consist of  three parts:  <component>-<power pin>-<ground pin>. Port names can be changed via context menu "Rename...".

Solver Settings tab

This dialog tab allows to edit all settings which are necessary to control the 3D solver.

In the Frequency settings frame, the following parameters can be defined:

Minimum/maximum frequency

The unit is in [Hz], but the number can be completed with the standard characters for kilo (k), mega (M), or giga (G).

Number of samples

The number of frequency samples to take.

Logarithmic sweep:

Determines whether linear sampling (use "False") or logarithmic sampling (use "True") will be applied.

Mesh settingsMax. mesh step

Specifies the mesh element size.

Component settingsConsider components (R, L, C)

Determines whether the electric models of the passive 2-pin components should be considered or not. The electric models for the corresponding components are defined in Passive Device Modeling.

Simplification settingsRestrict to excited nets

Determines whether all conductive structures of the PCB should be taken into account or not. If the parameter is true, only those nets which are connected to the selected Component pins and Terminals are considered.

Spatial impedance plots visualize the field aspect of the 3D computation on the surface of conductor layers. In order to understand the content of the plots, consider the electrical field computed for port excitation j at frequency f. For a given position p on a conductor layer, we draw a line to the closest reference conductor specified (see below). By integrating the electrical field along this line, we obtain a voltage value, which, in turn, is divided by the excitation current at port j. The result is "voltage divided by current", i.e. the transfer impedance Z(p,j).

Note: The practical content of impedance plots is as follows. Assume PDN impedance is too high at some port j and frequency f. One approach to mitigate the problem is to introduce de-coupling capacitors in appropriate locations, i.e. locations of high Z(p,j). Thus, Z(p,j) indicates best positions for adding "decaps".

The following parameters control creation of spatial impedance plots:

Generate plots

Switch on/off plot generation.

Reference layers

One or more reference layers can be specified, here. Note that for each plot, exactly one reference layer is used.

Restrict reference conductors to power/ground nets

If the parameter is true, on one reference layer, only those nets with either net class power or ground will be considered as reference conductor. This is the default behavior, since signal pads would disturb the plot.

Only plot at impedance extrema

If the parameter is true, plots will only be generated for frequencies that are either minima or maxima of some impedance curve. In that way, the number of plots is greatly reduced.

If, in contrast, the parameter is false,  a very large number of plots will be produced, as given by the following product:

(# plots) = (# ports)   x   (# frequency samples)   x   (# reference layers).

(If, for instance,  we assume a plot-file size of 1MB, 10 ports, 100 frequency samples and 2 reference layers, storing the plots would require 2GB of hard-disk space.)

Extrema contrast (%)

This parameter controls the definition of "impedance extrema", see above. Impedance |Z(f)| is said to have a maximum at frequency f0, if |Z(f)| falls monotonously on both sides of f0, by at least the relative value specified here.

Frequency tolerance (%)

Again, only relevant for finding impedance extrema: definition of close-by frequencies which is used for reducing the number of field computations required at impedance extrema.

DecapsExport decap optimization data

Prepare the S-Parameter results matrix and further data for input to a separate software tool (beta stage) allowing to find optimized de-coupling-capacitor (decap) mountings.

Calculate Impedances:

Pressing this button starts the 3DFEFD solver and the port impedances will be calculated. After the calculation has been finished, the 1D impedance curves will be stored in a separate Transfer Impedances folder inside the Navigation Tree of CST DESIGN STUDIO (according to the listed pins inside the Excitations/ports frame). If no 2D results were calculated the 1D impedance curves will be automatically displayed. If 2D results were calculated the corresponding data will be stored in a separate Results folder inside the Navigation Tree of CST PCB STUDIO and in this case, the 2D result viewer (see 2D Result viewer) will be launched in order to display the 2D results immediately.