PCB manufacturing PCB manufacturing
Home > Blog

grounded coplanar waveguide

GCPW Solder Mask: Model the Gap, Not Just the Trace
Friday, November 6th, 2026

On a grounded coplanar waveguide (GCPW), the electric field extends into the gaps beside the signal trace. Changing solder mask there changes the dielectric environment, even if the copper artwork stays the same. Model the actual coverage and test a representative structure before using the result in an RF design.

Where the Electric Field Goes

A GCPW has a surface signal between nearby ground copper and a reference plane underneath. Some of the field lies near the surface gaps rather than entirely in the laminate. Rogers’ comparison of microstrip and GCPW explains the different structures.

Surface RF routing near a solder mask opening
AI-generated illustration, not an impedance-verified geometry.

What Solder Mask Changes

Mask replaces some air with a dielectric. That can alter the field distribution, characteristic impedance, phase and loss. Rogers’ millimetre-wave fabrication paper discusses insertion-loss and phase effects; the size of the effect depends on the construction.

Detailed Inputs for the Solver

Input Unit or record Why it matters
— — —
Finished signal width mm Sets conductor geometry
Lateral gap mm, both sides Defines coupling to surface ground
Reference spacing mm Describes coupling below the trace
Mask shape Thickness and lateral extent Defines dielectric coverage
Material properties Dk and Df with frequency Defines electrical inputs

Mask Over the Gap vs an Open Window

Construction Primary check Other trade-off
— — —
Continuous mask Model trace top and gaps May protect exposed copper
Open RF window Model air and actual finish Review handling and assembly
Local mask dam Model the transition if relevant Check registration and solder control

Do Not Apply One Thickness Everywhere

Mask thickness above a conductor may differ from its coverage beside the copper. Ask how the cross-section is represented in the solver and what the fabricator can substantiate. A single thickness copied from a catalogue may not describe the lateral gap.

Example: A Launch Opens but the Route Stays Covered

Consider a hypothetical RF launch whose pad is exposed, followed by a masked line. The mask boundary creates a local change in environment even though the straight line width is constant. Compare the continuous covered section, the opening and the boundary rather than modeling only one cross-section.

Use a Two-Dimensional Model for the Uniform Section

A cross-section solver can compare uniform masked and unmasked geometries with other inputs held constant. It is not a full model of the launch or a short mask island. Use an appropriate three-dimensional or measured comparison if the transition determines performance.

Review the Finish Alongside Mask Removal

Opening a window can expose a finished surface that differs from the original modeled conductor. Include the specified finish and assembly requirements in the comparison. Do not attribute every measured loss change to the absence of mask when other features changed too.

Check Mask Registration Against the Gap

Mark the intended opening in the artwork and compare the alignment tolerance with the clearance from the signal and adjacent ground. Check both sides. A boundary intended to be symmetric can move and create an asymmetric structure.

Use a Coupon That Represents the Actual Coverage

Specify the same relevant width, gap, stackup and mask condition on the test coupon. For loss or phase checks, agree on the test method, frequency span and how fixture effects are handled. A plain impedance coupon does not automatically represent the RF launch.

RF test coupon and coaxial cables on a bench
AI-generated illustration of a test setup, not measured evidence.

Check Impedance and Loss Separately

Impedance verification can reveal a mismatch but does not replace a frequency-domain loss assessment. Define the acceptance goal before deciding which measurement is sufficient. See the BestPCBs impedance control guide for the wider stackup and coupon requirements.

Put the Mask Decision in the Release Data

Record which sections are covered, where the opening begins and ends, and which dimensions or test structures confirm it. If a change affects assembly, involve the responsible engineer rather than removing mask as a fabrication-only correction.

Request an RF Stackup Review

Share the frequency range and marked mask geometry with Best Technology.

FAQ

Does solder mask always lower the impedance?

Replacing air with dielectric generally changes the effective electrical environment, but the practical result depends on the geometry and actual coverage. Compare the two constructions using appropriate inputs and a representative test. Do not use a fixed correction factor from a different board or transmission line.

Does the mask above the trace matter more than the gap?

That depends on the field distribution of the structure. A coplanar line has fields in the lateral gap as well as beneath and around the trace. Include the actual coverage in the model rather than assuming that only the coating directly over the signal matters.

Is bare copper the same as an open solder mask window?

No. An open window removes mask from an area, but the copper can still have a specified surface finish. Model or measure the construction as it will actually be delivered. Check assembly, handling and environmental needs before choosing how that area should be exposed.

Can I use the same coupon for a masked and unmasked route?

Only if the coupon arrangement actually represents both cases and the acceptance method distinguishes them. A single uniform structure cannot demonstrate the effect of two different coverages. Define the relevant dimensions, mask pattern and test objective before requesting the coupon from the fabricator.

Can I remove mask from an entire RF board?

Do not assume that is necessary or appropriate. Focus on the sections where the electrical performance requires a change, then review assembly, finish and handling needs. A targeted opening can be easier to verify than a blanket removal that creates new constraints elsewhere.

Does passing a 50-ohm check prove low loss?

No. Characteristic impedance and attenuation are related design considerations but are not the same result. A matched line can still have conductor or dielectric loss. If loss is a requirement, agree on a frequency-dependent measurement that represents the relevant route and construction.

How do I handle a small local solder mask dam?

Define its location and geometry, and check whether it changes the response at the frequencies of interest. A uniform section model may not capture a local discontinuity. Keep the soldering purpose of the dam visible while comparing electrical alternatives and verifying the selected arrangement.

What should I send to the RF fabricator?

Send the stackup, frequency range, finished line width and gap, mask artwork and relevant performance limits. Mark the connector launch and any mask boundaries separately. Ask which coupon or test demonstrates the uniform line and which evidence covers the transition or actual assembled route.

You may also like

RF Via Fence Guide: Check the Connector Launch First
Friday, November 6th, 2026

An RF via fence is a row of grounded vias that connects top ground copper to a reference plane. Along a grounded coplanar waveguide it helps keep the reference conductors tied together. At an edge connector, however, a regular pitch alone cannot repair a ground interruption or a poorly shaped transition.

Follow a hypothetical edge-launch coax connector feeding a grounded coplanar trace on a small RF board. The objective is to identify whether the first part of the transition is different from the uniform line used to calculate impedance.

Identify the Launch and the Uniform Line

The launch is the short transition from the connector’s pin and ground body into the PCB. The uniform line begins where trace width, side gaps and the reference geometry settle to their intended values. Review both regions; the geometry that describes one is not necessarily valid for the other.

Thin RF trace beside rows of vias on a PCB
AI-generated illustration of coplanar trace context, not a layout to copy.

Compare Three Grounding Structures

Choose the model that matches the conductors actually present. A row of vias somewhere near a signal is not enough to define a line type.

Structure Conductors near the signal Check
Microstrip Trace and lower plane Reference continuity and dielectric
Grounded coplanar line Trace, side grounds and lower plane Side gaps and ground connections
Connector launch Pin, ground body and changing PCB geometry Transition and return path

Specify Pitch, Offset and the First Ground Connection

Use an approved stackup and connector footprint before selecting geometry. Analog Devices’ RF layout guide illustrates grounded via rows beside a coplanar trace; the layout still needs to be evaluated for its own stackup and frequency band.

Parameter Definition Use
Fence pitch Via center to center along a row, mm Compare regular line geometry
Via offset Via center to a named trace or ground datum, mm Keep comparisons consistent
First-via distance Launch datum to first effective ground connection, mm Inspect the local discontinuity
Finished construction Drill, plating and connected layers Confirm the modeled return path

Trace the Return Path From the Connector Body

Start at the solderable ground contact or body defined by the connector drawing. Follow the available top ground to vias and the reference plane below. A via that lands on an isolated island does not create the intended continuous connection merely because it has the correct pitch.

Find the Longest Local Gap, Not Only the Repeated Pitch

In a hypothetical layout with 0.8 mm pitch along the line, the first ground via may be 1.8 mm from a chosen launch datum because the connector footprint blocks the normal row. Those example dimensions are not RF limits. They show why the first interval needs its own check instead of inheriting the repeated pitch.

Keep the Launch Pad and Side Gap in the Same Review

A wider signal pad or changed side gap can modify the transition even when the via row is unchanged. Overlay the manufacturer’s footprint, copper profile, board edge and reference plane. Evaluate the complete transition rather than making the via row responsible for every mismatch.

Edge-mounted RF connector with nearby PCB ground vias
AI-generated illustration of the connector region; use the actual connector drawings for assembly geometry.

Check Both Sides of the Transition

Inspect each side ground from the connector to the uniform line. A cutout, mounting screw or isolation gap may constrain one side more than the other. Preserve the actual asymmetry in the model rather than mirroring the cleaner side and missing the problem.

Include Fabrication Geometry That Changes the Ground

Confirm finished via construction, plane connection, copper gaps and the board edge location. A shifted edge, removed ground island or different dielectric can alter the modeled transition. Keep the reviewed fabrication revision with the electromagnetic model.

Use Wavelength Only to Start the Review

An electrical length estimate can help flag large intervals, but there is no single pitch fraction that qualifies every launch. The field distribution depends on geometry, dielectric and frequency. Use the upper relevant operating band and a model or measurement appropriate to the actual connector and board.

Simulate the Launch Separately From the Long Line

Give the model enough uniform line for defined ports, then compare variants that change one feature at a time. Hold the reference planes consistent when comparing results. Otherwise a change in reported reflection or loss may come from the measurement reference rather than the fence.

Check the Assembled Board Against the Model

Use an approved network measurement with known reference planes and inspect the physical launch. Compare boards or coupons that use the same connector, mounting and stackup. A discrepancy can come from solder geometry, edge fit or calibration as well as from the layout.

Release the Launch as a Controlled Geometry

Keep the connector part, stackup, ground connections, trace transition and measurement definition under revision control. See the Best Technology via stitching guide for broader stitching applications; this check is specific to the connector transition.

FAQ

Does making every via closer always improve the launch?

No. Moving a via changes the surrounding copper geometry and may interact with gaps, pads or keepouts. Examine the actual return path and compare the modified transition with the same ports. Avoid reducing a distance at the cost of an unreliable fabrication feature.

Is a ground via on only one side sufficient?

That depends on the connector and the complete field and return geometry. Do not assume one side is interchangeable with both. Compare the actual asymmetric layout with the intended performance and verify it in the relevant band before deciding that the second row is unnecessary.

Can I use the same pitch on every RF board?

No. Frequency band, dielectric, trace geometry and connection details can differ substantially. Start from the device and connector guidance, then validate the construction that will actually be built. A pitch that worked in one layout does not approve another connector or stackup.

Does a 50-ohm calculation verify the connector transition?

Not by itself. A uniform line calculator typically describes a cross section that is unchanged along the trace. The connector pin, pad, ground attachments and board edge create a local change. Evaluate that transition separately and keep its measurement reference planes clear.

Does a return loss peak prove the first via is too far away?

No. A peak may come from the launch, the line, the connector, another discontinuity or the measurement setup. Compare controlled variants and review the physical board before attributing the peak to one feature. The frequency and reference plane of the measurement both matter.

Can a solder mask change invalidate the comparison?

It can change the local dielectric environment when mask is near the RF fields. Compare the actual mask opening and construction with the model before assuming results transfer. Treat a change as meaningful when it alters the geometry or material assumptions used for the release.

Should the via row end where the connector body begins?

Only the approved launch layout can answer that. The body and pads need a defined ground connection, and the usable position of vias depends on the footprint and edge geometry. Review the entire transition rather than ending the fence at an arbitrary outline.

What files make a launch review easier?

Include the connector drawing and part number, stackup, artwork, drill and plane data, frequency band, and the measurement or model reference planes. Highlight any opening that interrupts the side ground. This allows reviewers to compare the same physical path and not just a screenshot.

Prepare an RF Fabrication Review

Send the connector drawing, ground-via construction, stackup and transition artwork to the Best Technology team.

You may also like