STDA033 October 2026 MSPM0G3105-Q1 , MSPM0G3106-Q1 , MSPM0G3107-Q1 , MSPM0G3505-Q1 , MSPM0G3506-Q1 , MSPM0G3507-Q1 , MSPM0L1227-Q1 , MSPM0L1228-Q1 , MSPM0L2227-Q1 , MSPM0L2228-Q1 , MSPM33C321A
Another often overlooked optimization factor is the placement of the capacitors. The preferred placement shown in Figure 3-3 has three capacitors with the lowest value capacitor as close to the device pin as possible to reduce parasitic resistance and filter out higher frequency noise.
The ground connection for each capacitor is connected with a via to the ground plane layer with very short traces from the pad of the capacitor and the via. This is done to minimize the loop inductance and suppress noise that can be a result of larger loops which can function as an EM radiator. Target impedance turns a transient requirement into a design metric. If the load current changes by Delta-I and the allowed rail deviation is Delta-V, the power-distribution network must remain below Delta-V/Delta-I over the relevant bandwidth.
Target impedance is obtained by dividing the allowable rail-voltage deviation by the expected load-current step. The PDN must stay below that impedance across the bandwidth of the transient.
The highest-frequency capacitor must be closest to the supported pins with short, wide connections and adjacent power and ground vias. Bulk capacitance supports slower energy demand; local small-package capacitance supports fast edge current. Embedded package capacitance and tightly spaced power and ground planes extend the response to still higher frequency.
Figure 3-3 VIN Capacitor Placement Example Layout| Placement Choice | Electrical Effect | EMI Outcome |
|---|---|---|
| Capacitor beside pins with adjacent vias | Small loop and low mounted ESL | Local transient current; less plane excitation |
| Capacitor connected through a long neck | Added inductance isolates capacitor at high frequency | Rail bounce spreads and excites common-mode current |
| Shared narrow return for noisy and quiet loads | Shared impedance converts current to voltage | Reference modulation and cross-domain coupling |
| Bulk capacitor only | Good low-frequency energy, poor edge response | Fast current still travels across board |
Below is a summary of decoupling layout rules:
Bulk capacitors support lower-frequency load changes and prevent rail droop, but do not replace tight high-frequency bypassing. At the package and via scale, ESL often dominates. This is why a nearby 100nF capacitor can outperform a larger capacitor placed a few centimeters away for edge current.
Switching power stages deserve special treatment because they combine fast voltage edges, high current, magnetic fields, and nonlinear devices. Most regulator EMI problems originate from a few copper shapes: the switch node, the hot input loop, the gate-drive loop, and the diode/synchronous rectifier loop. Via stitching helps contain fields, but cannot compensate for a large switch loop or a switch node poured like a decorative logo.
Switching power stages combine high dv/dt, high di/dt, nonlinear devices, and magnetic components, so it is important to identify the commutation loops for each switch state. The highest priority loop is often the input ceramic capacitor through the switching devices and back. The switch node is a high-dv/dt electric-field source; the input and rectifier loops are high-di/dt magnetic-field sources.