Impedance Coupon Design

An impedance coupon is a test structure built into the panel that allows the impedance of a specific geometry to be measured. It converts a claim about the stack into a number, and it is the only practical way to verify that the geometry a design calculated was actually produced. Designing the coupon well means thinking about what will be measured, how, and how the measurement relates to the boards around it.

What the Structure Should Represent

The coupon should carry a structure for every combination of layer, trace width and reference configuration that the product uses. A design with two impedance layers needs a structure on each, and a design with single ended and differential nets needs both.

The structure must be long enough for the measurement method. A very short trace cannot be measured without the probe and the launch dominating the result, so the coupon must provide the length even where the product traces are short.

The reference plane arrangement must match the product. A microstrip coupon cannot verify a stripline design, and a coupon with a different plane spacing measures a different impedance no matter how similar the trace looks.

Test Methods

A time domain reflectometer sends a fast edge along the trace and reads the reflection, which gives the impedance along the length. It is the most common method because it shows the whole trace rather than a single value. The rise time of the edge sets the resolution, so a slow edge averages over a longer section and hides a local variation.

A vector network analyser measures the scattering parameters and derives the impedance from the transmission line behaviour. It is more accurate for loss and for a wide band but it needs a calibration and a more careful launch.

Whichever is used, the launch is the difficult part. The transition from the probe or the connector to the coupon trace is a discontinuity, and a coupon that does not provide a controlled launch will report it as part of the impedance.

Launch and Termination

The coupon should provide a launch that is defined in the drawing, whether it is a set of pads for a probe, a coaxial connector footprint or an edge connector pattern.

The launch should be as short as possible and it should be compensated where the design allows. A long pad adds capacitance, which lowers the measured impedance near the launch and makes the first part of the trace look wrong.

The far end should be terminated or left open in a defined way. An open end produces a reflection that a time domain measurement will show, and the two way travel time sets the length of the coupon needed for a clean measurement window.

Length and the Measurement Window

The trace has to be long enough that the reflection from the far end arrives after the launch disturbance has passed, and short enough to fit in the available coupon area.

The length also affects the loss, and a long coupon on a lossy material will show a rise in impedance along its length as the edge is attenuated. That effect should be understood rather than read as a geometry variation.

A coupon that is too short, or one that is measured with a probe whose own transition is longer than the trace, produces a number that describes the probe rather than the board.

Where the Coupon Sits

The coupon is placed in the panel border or on a dedicated test panel. Its position should represent the process that the product sees, because the plating, the etch and the lamination vary across a panel.

Where the product occupies most of the panel, a coupon in the border is the only option. Where the process variation is significant, a coupon at the edge and one in the centre give a better picture than a single coupon.

The coupon should be separated from the product by a score line or a route so that it can be removed without touching a board, and it should not be placed where its removal would leave a burr on a board edge. A coupon that is left attached in handling can break and take a piece of the board edge with it.

Tolerances and the Relationship to the Product

The coupon measurement gives the impedance of the coupon. The product may differ because of a different copper density around the trace or a different position on the panel, and the difference should be known.

The specification is usually a tolerance band around the nominal. The coupon measurement is compared with that band, and a value inside it releases the panel.

Where the coupon and the product differ systematically, the correlation should be established once and recorded. A fabricator that knows the offset can adjust the artwork so that the product lands at the nominal value even when the coupon measurement does not, and the adjustment should be recorded with the stack.

Records and Traceability

The coupon should be identified so that the measurement can be linked to a panel and a work order. Without that link the measurement proves the process in general and nothing about the batch in question.

The record should include the structure, the layer, the measured value and the method, and for a time domain measurement the value should be taken from the same point on the trace each time.

The results should be tracked over time. A drift in the measured value over several batches indicates a change in the process before it appears as a customer complaint.

Design Rules for the Coupon Layout

Keep the structures separated so that the measurement of one does not see the others. The spacing should be enough that the probe and the ground return do not couple to a neighbouring structure.

Provide a ground reference close to the signal structure, in the same way as the product does. A coupon whose ground is far away measures a different environment from the board.

Keep the coupon geometry in step with the product. When the product geometry is revised, the coupon should be revised at the same time, because a coupon that describes an obsolete stack is worse than no coupon.

Practical Rules

Include a structure for every layer and geometry combination, provide a controlled launch, and place the coupon where it sees the production process. Record the value against the panel and track the trend.

Record the design and the measurements with the build records and the impedance stackup design, and review the test coupon practice and the coupon design when the coupon set is defined.

FAQ

Why is the launch critical? The transition from the probe or connector to the trace is a discontinuity. A poor launch adds capacitance and shows up in the measurement as a low impedance near the start.

How long should the coupon trace be? Long enough that the far end reflection arrives after the launch disturbance has passed, and short enough to fit the coupon area without excessive loss.

Does the coupon value equal the board value? Not exactly. Copper density around the trace and the panel position differ, so the correlation should be established and recorded rather than assumed.