PDN Resonance Analysis

PDN Resonance Analysis

Quality & Inspection Standards

PRODUCTS CENTER

Get Custom Quote

PRODUCTS CENTER

Supported formats: PDF, DWG, Gerber, Excel (Max 50MB)
Submit
UltroNiu Power Integrity PDN Resonance

PDN Resonance Analysis – Antiresonance, Target Impedance & Mitigation for High‑Speed Digital Systems

Power Distribution Network (PDN) resonance is a leading cause of simultaneous switching noise (SSN), voltage ripple amplification, and radiated EMI in high‑speed digital systems. When the PDN's impedance peaks at certain frequencies due to the interaction of board capacitance, package inductance, and bulk decoupling, the resulting voltage ripple can exceed noise margins and cause functional failures. This engineering framework from UltroNiu defines how to analyze, measure, and mitigate PDN resonances using target impedance methods, VNA measurement, simulation correlation, and design rules for decoupling capacitor placement, plane geometry, and stackup optimization.

📅 Published: May 19, 2026  |  ⏱️ 14 min read  |  🏷️ #PDNResonance #PowerIntegrity #TargetImpedance #SSN #Decoupling

1. Purpose – Why PDN Resonance Matters

PDN resonance occurs when the impedance of the power distribution network (PCB planes + package + decoupling capacitors) exhibits a peak at a specific frequency due to the parallel combination of inductance and capacitance. This causes:

  • Amplification of switching noise from digital ICs (SSN) → voltage ripple exceeding noise margin
  • Potential power supply rejection ratio (PSRR) violation for sensitive analog/RF circuits
  • Radiated EMI from cavity modes between power and ground planes
  • Reduced effectiveness of decoupling capacitors (capacitors appear inductive at resonance)
  • Oscillations in low‑dropout regulators (LDOs) if PDN presents negative impedance

This framework defines how to:

  • Measure PDN impedance profile (ZPDN) using a VNA in 2‑port shunt‑through configuration
  • Identify dominant resonant frequencies and corresponding impedance peaks
  • Set target impedance (Ztarget) based on transient current and noise margin
  • Mitigate resonances through decoupling capacitor selection/placement, plane stitching, and stackup design
  • Validate PDN performance through lab measurement and simulation correlation

2. Failure Physics – Antiresonance, Cavity Modes & Impedance Peaks

PDN impedance is a function of frequency. The PDN consists of three regions:

  • Low frequency (DC to ~1 MHz): Dominated by bulk capacitors (electrolytic / tantalum). Impedance is low, determined by ESR of bulk caps.
  • Mid frequency (1 MHz to ~100 MHz): Dominated by ceramic decoupling capacitors (MLCCs). Antiresonance peaks occur where the parallel combination of bulk inductance (ESL) and MLCC capacitance creates a high‑impedance spike.
  • High frequency (>100 MHz): Dominated by PCB plane capacitance and package inductance. Cavity resonance occurs when the dimensions of power‑ground plane pair form a quarter‑ or half‑wavelength resonator.
Resonance Type Typical Frequency Range Root Cause Mitigation
Antiresonance (MLCC + bulk) 500 kHz – 10 MHz Parallel combination of inductive bulk caps and capacitive MLCCs Use low‑ESL MLCCs; spread capacitance values; add damping resistance
Plane cavity resonance (quarter‑wave) 100 MHz – 2 GHz (depends on board size) Board dimensions create standing wave between power and ground planes Reduce plane size; add stitching capacitors; use thinner dielectric (higher plane capacitance)
Package resonance 50 MHz – 500 MHz Package lead inductance + on‑die capacitance Not user‑controllable (die‑level); use on‑package decoupling
VRM resonance 10 kHz – 1 MHz Output filter of switching regulator Adjust compensation network; add damping

3. Risk Drivers & Application Thresholds

The risk of PDN resonance‑induced failure increases with higher transient currents, lower core voltages (tighter noise margin), and faster edge rates.

Application / Device Transient Current (di/dt) Core Voltage (Vcore) Resonance Analysis Required?
Low‑power MCU (≤100 mA) Low 3.3 V No – standard decoupling sufficient
FPGA / processor (<1 A) Moderate 1.8 V – 2.5 V Recommended – avoid antiresonance peaks
High‑performance FPGA / CPU (≥5 A) High (10‑100 A/μs) 0.8 V – 1.2 V Mandatory – full ZPDN measurement & simulation
SerDes / PLL (analog sensitive) Low (but sensitive) 1.0 V – 3.3 V Mandatory – ripple <10 mV often required
Automotive ASIL‑D system Any Any Mandatory – resonance must be characterized

4. Mitigation Strategies – Decap Placement, Plane Geometry, Stackup

  • Decoupling capacitor selection: Use multiple capacitor values (e.g., 10μF, 1μF, 0.1μF, 0.01μF) to spread antiresonance frequencies. Prefer low‑ESL (reverse geometry) MLCCs for high‑frequency response.
  • Placement: Place smallest capacitors closest to IC power pins (lowest loop inductance). Use interleaved placement of different values to reduce resonance peaks.
  • Plane geometry: Reduce cavity resonance by adding stitching capacitors between power and ground planes at quarter‑wavelength intervals. Use many small (0.1μF) capacitors distributed across the board.
  • Stackup: Use thinner dielectric between power and ground planes to increase plane capacitance (shifts cavity resonances higher).
  • Damping: Add series resistance (e.g., 1Ω) in series with some decoupling capacitors to dampen antiresonance peaks.
  • Embedded capacitance: Use ultra‑thin dielectric (e.g., 2 mil) with high‑Dk material for extremely high plane capacitance.

5. Test Architecture & Measurement Setup

PDN impedance measurement requires a dedicated test coupon or direct probing on the actual PCB. The recommended method is 2‑port shunt‑through measurement using a VNA, which provides highest accuracy up to several GHz.

Test setup components:

  • Vector Network Analyzer (VNA) with DC bias tee (to allow voltage on the PDN during measurement)
  • Two high‑frequency probes (e.g., GSG or differential probe) or coaxial cables soldered to test points
  • Calibration substrate (SOLT or TRL) up to 3 GHz or higher
  • Test structure: Two test points placed on the same power net, typically 1‑2 cm apart, with ground references as close as possible
  • Measurement frequency range: 100 kHz to 1 GHz (or up to the maximum frequency of interest, typically 3× clock)

Alternative (lower frequency) method: Use a network analyzer with a shunt‑through current probe, but VNA with direct probing is preferred.

6. Measurement Methods – VNA (2‑Port Shunt‑Through), TDR, Near‑Field Scan

Method Measured Parameter Frequency Range Advantage Limitation
2‑port shunt‑through VNA ZPDN (S21 → Z) 100 kHz – 3+ GHz Wide bandwidth, high accuracy Requires DC bias tee, calibration, two probes
1‑port reflection VNA S11 → Z 100 kHz – 1 GHz Single probe, simpler Less accurate at low Z (<1Ω)
TDR step impedance Impedance vs time (converted to frequency via FFT) DC – 1 GHz (dependent on rise time) Also yields spatial information Lower frequency resolution
Near‑field scan (E‑field probe) Cavity resonance location 100 MHz – 6 GHz Identifies resonance hot spots Does not directly give impedance

The 2‑port shunt‑through method (VNA ports 1 and 2 connected to two test points on same net) is the industry standard for PDN impedance measurement per IPC‑2212 and IEEE 370.

7. Acceptance Criteria – Target Impedance Profile & Max Ripple

The PDN is acceptable if its impedance ZPDN(f) stays below the target impedance (Ztarget) across all frequencies relevant to the load transient. Ztarget is calculated as:

Ztarget = (Vsupply × ripple%) / Itransient

Typical ripple% is 3–5% for core rails, 1–2% for sensitive analog/PLL rails.

Application / Rail Type Typical Vsupply Max Ripple % Ztarget (example Itransient=5A)
FPGA core (0.8‑1.2V) 0.9 V ±5% → 45 mV 9 mΩ
FPGA core (high‑end) 0.8 V ±3% → 24 mV 4.8 mΩ
DDR memory VDDQ 1.2 V ±5% → 60 mV 12 mΩ
PLL / analog (3.3V) 3.3 V ±1% → 33 mV 6.6 mΩ (Itransient=5A)
Automotive ASIL‑D 1.2 V ±3% → 36 mV 7.2 mΩ

Pass/fail criteria: Measured ZPDN(f) SHALL be below Ztarget for all frequencies up to the maximum transient frequency (typically 3× clock speed). Any resonance peak exceeding Ztarget constitutes a failure requiring mitigation.

8. Decision Logic – When Resonance Analysis is Mandatory

Condition Analysis Scope Action
Itransient > 1A OR Vcore ≤ 1.2V Simulation + measurement recommended Perform initial ZPDN simulation; measure first prototype
Itransient > 5A OR Vcore ≤ 0.9V Mandatory full measurement on first article Include PDN test coupon; perform 2‑port VNA measurement
Contains PLL, ADC, or RF circuit Mandatory ripple measurement on clean analog rail Measure PDN impedance up to 1 GHz; ensure Z < Ztarget
Automotive / safety‑critical Mandatory simulation + measurement + margin Design Ztarget with 50% margin (Ztarget/2)
Board size > 100 mm × 100 mm with solid power‑ground plane Cavity resonance analysis required Simulate cavity resonances; add stitching caps if needed

9. Simulation Correlation & Model Validation

PDN simulation (e.g., in Ansys SIwave, Cadence Sigrity, Keysight ADS) must be correlated with measured data to ensure accuracy.

  • Build simulation model including PCB stackup, plane geometry, via locations, capacitor models (with ESL, ESR), and VRM output impedance.
  • Simulate ZPDN at the same probe locations as planned for measurement.
  • Compare measured vs simulated ZPDN up to 1 GHz. Acceptable correlation: within ±3 dB for resonant peaks and within ±20% for average impedance.
  • If correlation fails, refine capacitor models (ESL/ESR extracted from manufacturer's S‑parameters) or plane model (mesh refinement).
  • Once correlated, simulation can be used for design‑of‑experiments to optimize decoupling placement.

10. Sampling Plan for Qualification

  • First article (prototype): Measure ZPDN on 2 boards (to account for assembly variation) at all critical power rails (core, IO, PLL, DDR).
  • Production validation (automotive / medical / high‑reliability): Measure 1 board per lot (or per month) for each power rail with Ztarget < 20 mΩ.
  • For low‑risk designs: No production measurement required, but first article measurement mandatory.
  • Test points: Each power rail shall have two dedicated vias or test pads for VNA probing, placed within 1‑2 cm of the load IC power pin.

11. PCB Supplier Control – Dielectric Thickness, Plane Clearance

  • Supplier SHALL control dielectric thickness between power and ground planes to within ±15% of nominal to ensure plane capacitance consistency.
  • Anti‑pad clearance around via antipads SHALL be controlled to ±0.05 mm; excessive clearance increases plane inductance.
  • Supplier SHALL provide impedance test coupons for each panel; the same stackup applies to PDN planes.
  • Change in laminate material (Dk, thickness) or copper weight requires requalification of PDN (simulation + measurement).

12. Requalification Triggers

  • Any change in PCB stackup (layer count, dielectric material, thickness).
  • Change in decoupling capacitor type (different manufacturer, series, or voltage rating).
  • Relayout of power distribution (moving planes, adding/removing vias).
  • Change in load IC (different transient current profile).
  • Field failures attributed to power supply noise (determined by ripple measurement on failing units).
  • Annual requalification for high‑reliability systems.

13. Procurement Specification Notes

1. PCB supplier shall provide stackup documentation including dielectric thickness (nominal and tolerance) for each power‑ground plane pair.

2. Supplier shall control dielectric thickness to ±15% and copper etch to ±10%.

3. Customer may perform PDN impedance measurement (2‑port VNA) on first article boards. Acceptance criteria: ZPDN < Ztarget as defined in procurement specification.

4. Any change to laminate material or prepreg stackup requires 90‑day PCN and requalification of PDN performance.

5. For high‑reliability orders, supplier shall reserve two test points per power rail (0.5 mm via or test pad).

14. Position in Power Integrity Ecosystem

This PDN resonance framework is part of the UltroNiu Power Integrity Toolkit. Related resources:

15. FAQ – PDN Resonance Analysis

Q: How do I determine the target impedance (Ztarget) for my PDN?

A: Ztarget = (Vsupply × allowed ripple%) / Itransient. For example, 0.9V core, 5% ripple (45 mV), 5A transient → Ztarget = 0.045/5 = 9 mΩ. Use worst‑case di/dt from the load IC datasheet.

Q: What is the best way to measure PDN impedance above 1 GHz?

A: Use a 2‑port VNA with GSG probes calibrated with an impedance standard substrate (IS). For frequencies above 1 GHz, the test points must be very close (≤1 cm apart) and the PCB ground must be solid. Alternatively, use a dedicated PDN coupon with SMA launch.

Q: Why does adding more decoupling capacitors sometimes make resonance worse?

A: When capacitors of similar value are placed in parallel, their equivalent inductance combines, creating a higher‑Q resonance peak. Use distributed values (e.g., 10µF, 1µF, 0.1µF, 0.01µF) and place smallest capacitors closest to the load to spread resonance frequencies.

Q: What is a cavity resonance, and how do I detect it?

A: Cavity resonance occurs when the board’s power‑ground plane pair acts as a resonant cavity. It appears as a sharp impedance peak at frequencies where the board dimension is a multiple of half wavelength. Detect by near‑field scanning (E‑field probe) or by measuring ZPDN at multiple locations.

PDN Resonance Validation & Supplier Control Checklist

  1. 1. Itransient estimated from load IC datasheet or measurement.
  2. 2. Ztarget calculated with 20–50% margin.
  3. 3. PDN simulation performed (optional but recommended for high‑current rails).
  4. 4. Test points (two per rail) placed on PCB layout, within 2 cm of load.
  5. 5. First article PDN impedance measured using 2‑port VNA, 100 kHz–1 GHz.
  6. 6. Measured ZPDN below Ztarget across all frequencies → PASS.
  7. 7. If resonance peaks exceed Ztarget, mitigation applied (staggered caps, damping, stitching).
  8. 8. Cavity resonance analyzed for board size >100 mm.
  9. 9. Supplier stackup tolerances: dielectric thickness ±15% max.
  10. 10. Requalification plan defined (annual or after changes).

Need PDN resonance measurement and mitigation for your high‑current FPGA or processor? Our power integrity lab provides VNA measurement, simulation correlation, and decoupling optimization.

Request PDN Analysis Support →

References: IPC‑2212 (Power Distribution), IEEE 370 (Electrical Characterization), Intel PDN Design Guide, Keysight AN 5989‑8554EN (2‑Port Shunt‑Through Measurement), Murata capacitor simulation models.

Get Custom Quote

PRODUCTS CENTER

Supported formats: PDF, DWG, Gerber, Excel (Max 50MB)
Submit