Impedance Coupon Design: 6 Rules for a Coupon You Can Trust

An impedance coupon is a set of test traces built into the panel border that lets the fabricator and the customer measure the controlled impedance the process actually produced. A well built test coupon is the only practical way to confirm that a stack-up calculation has been realised on the shop floor.

The quality of that evidence depends entirely on how the coupon is designed and measured. A coupon with the wrong launch, an unrepresentative reference plane or a length that is too short will report a number that looks precise and says very little about the boards beside it.

What an Impedance Coupon Has to Prove

The coupon demonstrates two things: that the geometry of the manufactured trace matches the design, and that the material properties assumed in the calculation are correct for the laminate batch in use. Both are needed, because a trace that is the right width in a material with the wrong dielectric constant will still miss its target.

It also provides a record. Where a customer later measures a board and finds an impedance outside tolerance, the coupon data taken from the same panel is the evidence that shows what the process delivered at the time of manufacture.

Coupon Design: Line Widths and Lengths

A coupon typically carries single ended and differential pairs at the widths used on the product, often with a set of lines etched a few micrometres wider and narrower to show the sensitivity of the result to width. That sensitivity data is what tells you whether a small process shift will move the impedance out of tolerance.

Length matters more than most designers expect. The line has to be long enough for the measurement to resolve the impedance plateau rather than the launch artefacts, but short enough to fit in the panel border and to avoid excessive loss at high frequency. A common working range is a few centimetres.

<img src="https://www.gopcba.com/wp-content/uploads/2026/08/Test-of-Box-Build-Assembly-1.jpg" alt="Time domain reflectometry trace from an impedance coupon measurement” />

Launch and Probing: Test Points and Connectors

The launch is where the probe or connector meets the line, and it is the largest source of measurement error. Options include ground-signal-ground pads, coaxial launches and solderable connectors, and each has a different effect on the measured waveform, especially above a few gigahertz.

Probing has to be consistent between measurements. The same probe, the same probe station and the same landing pattern should be used every time, because moving the reference point or changing the probe geometry shifts the measured value by more than the process tolerance being verified.

Reference Planes and the Coupon Stack

The coupon must be built with the same layer sequence, the same dielectric thicknesses and the same reference planes as the product. A coupon that shares the panel but uses a different copper distribution or a different prepreg will not behave the same way, and its data cannot be applied to the boards. Coupon and product should be manufactured in the same run, because the laminate batch and the plating conditions are what the measurement is describing.

Reference plane continuity is part of the design. A coupon trace that crosses a plane split, or that runs over a plane with a large opening beneath it, measures a discontinuity rather than an impedance, and that discontinuity is often mistaken for a manufacturing problem.

Measurement: Time Domain Reflectometry

Time domain reflectometry sends a fast edge down the line and records the reflection that returns. The reflected waveform shows impedance along the line, so it reveals not only the average value but also local variations, including the launch, the transition through a via and the effects of a change in width.

Reading the result takes care. The impedance plateau in the middle of the trace is the figure of interest, while the first part of the waveform reflects the launch and the last part the termination. Averaging over the plateau, and reporting the measurement conditions with the number, is what makes the result comparable with the next batch.

Correlating Coupon Data With the Panel

The coupon is only useful if it represents the panel. If the coupon is built in a border region where the plating and the etch behave differently from the centre, the data describes the border rather than the boards, and the impedance tolerance appears to be met while product traces are outside it.

Building the coupon with the same widths, the same spacing and the same layer build as the product minimises the difference. Where a panel carries more than one impedance family, each family needs its own coupon rather than one trace standing in for the rest.

Coupon Placement on the Panel

Placement affects what the coupon sees. Copper distribution changes plating thickness and etch rate across a panel, so a coupon placed in a sparse area will not reflect a trace that runs between dense copper. Placing coupons at more than one position shows the spread across the panel rather than a single value.

The coupon also has to survive the process in a measurable state. It needs to remain attached to the panel through assembly of the boards, so its tabs and its position have to be planned in the panelization scheme rather than squeezed into whatever space is left.

Accepting a Coupon Result

Acceptance compares the measured value with the specified target and tolerance, and it should also consider the measurement uncertainty. A result that sits exactly on the limit is not the same as one in the middle of the window, and the difference matters when the next batch drifts slightly.

Where a result fails, the coupon design should let you tell whether the cause is width, dielectric thickness or dielectric constant. A coupon that carries width variants and a differential pair alongside the single ended trace gives that separation, while a single test line leaves the cause unresolved. That separation is what allows the shop to adjust plating or etch rather than guessing at the laminate.

When Coupon and Board Disagree

Disagreement usually has a physical explanation. A large ground plane under the coupon that is absent on the product, a coupon trace with a different surface finish, or a board measured at a different temperature will all shift the result. Checking the coupon against the product geometry before questioning the process is the faster route.

Where the difference persists, the quality control plan should include destructive verification. A cross section of the product trace beside a coupon measurement resolves most disputes, because it shows the finished width and the dielectric thickness at the place the signal actually travels.

FAQ

How many test traces should a coupon have? Enough to cover every impedance family on the panel plus a width variant or two, so that a failure can be attributed to width or to material. A single trace makes the cause of a miss impossible to establish. Most shops also keep a record of the coupon results per laminate batch, so a shift can be traced to the material as well as to the process.

Can the coupon be measured after assembly? It can be measured at any time while it remains attached, and measuring before and after assembly is useful because the process adds thermal history. What matters is that both measurements use the same launch and the same conditions.

Does a coupon guarantee the product traces are within tolerance? It shows that the process held its geometry at the coupon location during that build. Because plating and etch vary across a panel, the coupon supports the conclusion rather than proving it, which is why placement and cross section data matter alongside it.