RF PCB Project Services

VNA Test Service | S-Parameter Measurement to 40 GHz

A VNA test service measures the S-parameters of RF PCBs, cables, connectors and modules from DC to 40 GHz or beyond, giving engineers verified insertion loss, return loss and impedance data they can trust against simulation. This guide covers what…

Vector network analyzer measuring S-parameters of an RF PCB on a lab bench

What Is a VNA Test Service?

A VNA test service uses a vector network analyzer to measure the complex scattering parameters (S-parameters) of a device under test—typically an RF PCB, filter, coupler, antenna feed, cable assembly or complete module—across a specified frequency range, often DC to 40 GHz. The service delivers calibrated magnitude and phase data for every port combination (S11, S21, S12, S22 for a two-port), giving you the return loss, insertion loss, group delay and impedance information needed to validate a design against simulation or a procurement spec. If your measured S21 insertion loss on a 10 cm microstrip at 28 GHz is 0.4 dB worse than your EM model predicted, you know the problem before the board ships.

Why Outsource VNA Measurement?

GSG probe tips contacting a coplanar waveguide test pad on a high-frequency PCB

A quality 40 GHz two-port VNA (such as the Keysight PNA-L N5232B or Rohde & Schwarz ZNA40) costs upward of $80 k–$200 k depending on options. Add a calibration kit, phase-stable test cables, and a trained operator who understands connector torque, port extensions and uncertainty budgets, and the cost of ownership is significant. Outsourcing to a dedicated VNA test service makes sense when:

  • You need measurements only during prototype or qualification, not daily production.
  • Your in-house VNA does not cover the frequency range—e.g., you have a 20 GHz unit but need 5G FR2 data at 28–39 GHz.
  • You require an independent, third-party data set to satisfy a customer or a quality and test report deliverable.
  • Your design involves connectorized, on-wafer, or waveguide DUTs that need specialized fixturing.

For teams that also need the board built and tested end-to-end, a turnkey RF module service can fold VNA verification directly into the assembly flow, eliminating a separate shipping and handling step.

What Gets Measured: S-Parameters Explained

Two-Port Measurements

Most RF PCB traces, filters and transitions are characterized as two-port networks. The four S-parameters map directly to engineering quantities:

Parameter Physical Meaning Typical Spec Example
S11 (input return loss) Reflected power at port 1 ≤ −15 dB across 24.25–29.5 GHz
S21 (insertion loss / gain) Transmitted power from port 1 → 2 ≤ 0.15 dB/cm at 10 GHz on RO4350B microstrip
S12 (reverse isolation) Transmitted power from port 2 → 1 Reciprocal for passive networks (S12 = S21)
S22 (output return loss) Reflected power at port 2 ≤ −20 dB at center frequency

Multi-Port and Mixed-Mode

Differential pairs, couplers and antenna feed networks often require four-port or higher measurements. A 4-port VNA captures all 16 Sij combinations, from which mixed-mode parameters (Sdd11, Sdd21, Scd21, etc.) are derived. If you are comparing antenna simulation against measurement, the VNA provides the reference truth for input impedance and coupling between elements.

Calibration: The Make-or-Break Step

Every VNA test service lives or dies on calibration quality. Without proper calibration, systematic errors from cables, adapters and the VNA’s own directivity corrupt every data point.

Common Calibration Methods

  • SOLT (Short-Open-Load-Thru) — the workhorse method up to ~26.5 GHz with coaxial standards. Simple, well understood, but sensitive to connector repeatability above 30 GHz.
  • TRL (Thru-Reflect-Line) — preferred for on-board and waveguide measurements. Uses transmission-line standards whose impedance is set by geometry, not lumped models, making it more accurate at millimeter-wave frequencies.
  • eCal (Electronic Calibration) — a solid-state module that switches through multiple impedance states automatically. Fast and repeatable; most labs keep an eCal module for each connector family (2.4 mm, 2.92 mm, SMA, 3.5 mm).
  • Unknown Thru (SOLR) — relaxes the requirement for a fully characterized thru, useful when the DUT’s connectors differ on each port (e.g., SMA on one side, SMPM on the other).

A reputable service will document the calibration method, the standards used, the verification step (measuring a known device after cal to confirm residual error), and the measurement uncertainty budget. If they cannot tell you the expanded uncertainty of their S21 at 30 GHz, look elsewhere.

Specifying IPC Class for Boards Under Test

The IPC class of your RF PCB directly affects the consistency of the traces and vias the VNA will measure. A board built to IPC-6012 Class 3 has tighter copper-thickness tolerances, stricter annular-ring minimums and more rigorous cross-section inspection than Class 2—all of which reduce impedance variation and make your VNA data more representative of the design intent.

Use the selector below to determine which IPC class fits your application. Enter the end-use environment and reliability requirement; the tool returns the recommended class and highlights the key dimensional tolerances that change.

Loading IPC Class Selector…

If the tool returns Class 3, expect the fabricator to hold etch tolerances within ±0.5 mil (0.013 mm) on fine lines, versus ±1.0 mil for Class 2. On a 50 Ω microstrip at 28 GHz where a 0.5 mil width change shifts impedance by roughly 1.5 Ω, that tighter tolerance is the difference between a return loss of −22 dB and one that barely meets a −15 dB spec. Boards destined for test and measurement instruments almost always warrant Class 3.

Process Capability and Cpk Targets

Panel of RF filter boards with SMA connectors prepared for batch VNA testing

When a VNA test service measures dozens or hundreds of identical boards, the data set becomes a process-capability study. The Cpk index tells you how centered the measured parameter is within spec limits and how much margin exists. A Cpk of 1.33 means the process is comfortably within spec with roughly 63 defects per million opportunities; a Cpk of 1.67 pushes that below 1 ppm—essential for automotive radar and aerospace programs.

Plan your Cpk target before you request the VNA service. Enter your upper and lower spec limits plus your target defect rate below; the planner returns the Cpk you need to hit and the implied sigma level.

Loading Cpk Goal Planner…

For example, if your insertion-loss spec on a 77 GHz automotive radar feed network is 0.8 ±0.15 dB and you need Cpk ≥ 1.67, the planner shows you need a process standard deviation no larger than 0.030 dB. That is achievable only with tight laminate Dk tolerance (Rogers RO3003 specifies Dk = 3.00 ±0.04 at 10 GHz), Class 3 etch control, and a VNA measurement uncertainty well below the process spread—typically ±0.05 dB or better at that frequency.

How to Request a VNA Test Service: Checklist

  1. Define the DUT. Connector type, number of ports, board dimensions, any special handling (ESD-sensitive MMICs, flex tails, waveguide flanges).
  2. Specify frequency range and point count. Example: 100 MHz to 40 GHz, 1601 points, log or linear sweep.
  3. State the parameters needed. S-parameters (magnitude, phase, or both), group delay, impedance (Smith chart data), time-domain reflectometry gating.
  4. Declare pass/fail limits. Provide a limit-line file or a table of specs per frequency band.
  5. Calibration preference. SOLT, TRL, eCal; reference plane location (connector mate, board edge, probe tip).
  6. Output format. Touchstone (.s2p / .s4p), CSV, PDF report with plots, or all three.
  7. Sample size and Cpk requirement. How many boards, and whether statistical analysis is needed.
  8. Confidentiality. If the design is sensitive, confirm that an NDA is signed before any files are reviewed.

Common Pitfalls That Corrupt VNA Data

  • Over-torqued or under-torqued connectors. SMA connectors need 5 in-lb (0.56 N·m); 2.4 mm connectors need 8 in-lb. A calibrated torque wrench is non-negotiable above 18 GHz.
  • Cable movement after calibration. Phase-stable cables (e.g., Gore PHASEFLEX) drift less than 2° at 40 GHz over a 90° bend cycle; standard semi-rigid jumpers can shift 10° or more.
  • Missing port extension or de-embedding. If the reference plane is at the connector but you want data at the board edge, the extra connector-to-board transition adds loss and mismatch that must be removed mathematically.
  • Temperature drift. A 5 °C ambient shift can move the Dk of PTFE-based laminates by ~0.01, which shifts a resonant-filter center frequency by several MHz at X-band. Stabilize the lab or record the temperature.

Interpreting Your Results

When the Touchstone files arrive, load them into your EDA tool or a free viewer such as the Keysight ADS data display and overlay simulation versus measurement. Key things to check:

  • Return loss floor. If S11 never drops below −30 dB even at low frequencies, the calibration or fixturing may be limiting—not the DUT.
  • Insertion-loss slope. A steeper-than-expected roll-off with frequency usually points to higher conductor roughness or Df than the model assumed. Cross-check against the laminate datasheet values.
  • Resonances or spikes. Narrow dips in S21 at specific frequencies often indicate cavity resonances in a shielding enclosure or stub effects from unterminated vias.
  • Group-delay ripple. Ripple exceeding ±0.1 ns in a wideband channel can degrade EVM in digital modulation systems.

Frequently Asked Questions

What frequency range does a typical VNA test service cover?

Most services offer standard coaxial measurements from 10 MHz to 40 GHz. Extended ranges to 67 GHz or 110 GHz are available with 1.85 mm or 1.0 mm connectors, though fixturing cost and measurement uncertainty both increase.

Can I get VNA data on bare PCB traces without connectors?

Yes. Probe-based measurement using GSG (ground-signal-ground) RF probes on coplanar launch pads is common for characterizing bare-board transmission lines up to 67 GHz. TRL calibration on dedicated cal standards etched into the same panel gives the best accuracy.

How many samples should I send for a meaningful Cpk?

A minimum of 30 samples is the statistical baseline for a Cpk study. For high-reliability programs (automotive, aerospace), 50–100 samples across multiple production lots provide a more credible capability index.

What file format will I receive?

Touchstone (.s1p, .s2p, .s4p) is the industry standard and imports directly into every major EDA tool. Most services also provide a PDF report with annotated plots and pass/fail summary.

Does board material affect VNA results significantly?

Absolutely. A standard FR-4 laminate with a Df of 0.020 at 10 GHz will show roughly three to four times the insertion loss per unit length of a low-loss material like Rogers RO4350B (Df ≈ 0.0037). Material choice is the single largest variable in measured S21 at microwave frequencies.

Ready for a quotation?

Send your requirements for a material, stackup and manufacturing review. Price and schedule are confirmed for your project.

Upload Gerber