Contact Us
  • Home
  • BLOG
  • What Does an Impedance Coupon Prove on a Panel?

What Does an Impedance Coupon Prove on a Panel?

An impedance coupon proves one thing: that the panel in the box was etched, laminated and plated to the width and dielectric thickness your stackup called for, and that someone measured it on the same laminate, at the same time, with a calibrated TDR. It does not prove the board works. It is the only evidence you get that a 50 ohm line is 50 ohms before it disappears under solder mask and finish.

That distinction decides how the report should be read. A statement that the process is capable talks about the factory. A coupon report talks about your panel.

What a coupon actually is

It is a strip of test structures on the panel border or in a waste area between boards: a single ended microstrip and a stripline at the width in the stackup, a differential pair at the gap in the stackup, and probe pads that fit the fixture. It is built in the same lamination cycle, from the same copper weight, through the same plating and the same finish, because none of those steps can be reproduced on a separate test panel. Dielectric thickness after lamination depends on copper coverage, so a coupon with a different copper density is a different dielectric thickness and measures a different number.

Placement matters as much as geometry. Coupons belong at least 10 mm in from the panel edge, ideally at two opposite corners, because plating current density and etch uniformity vary across a panel. A coupon at the extreme edge sees more copper in the hole and a slightly different line width than a board in the middle, and it will flatter the panel.

The structures are standardised. IPC-2221 Table 12-1 asks for an A/B or A coupon twice per panel at opposite corners for thermal stress and plating thickness, a D coupon once per panel for interconnect res Z coupon for controlled impedance. If a supplier cannot say which coupon is on the panel, the report is not traceable to anything.

How the measurement is made

Time domain reflectometry, method IPC-TM-650 2.5.5.7. A fast edge is launched into the coupon and the reflections map impedance along the line. Edge speed sets the resolution: an edge of 100 ps resolves features of roughly 7 mm, so a 25 mm coupon is measured across its length in one shot while a 3 mm neck inside the board is not the same measurement. Production systems average 64 to 256 acquisitions to pull the noise down.

Tolerance is where the argument usually starts. Ten percent is common on general high speed work. Five percent is what a tight SerDes or RF budget needs, and two percent is reserved for the critical nets. The number has to match the budget in the stackup, not the number the fabricator happens to publish, and it has to be written per layer rather than as one figure for the whole board.

Solder mask is part of the measurement. Mask with a Dk near 3.5 over a core at 4.2 pulls impedance down by a few percent, so a coupon measured bare and a trace measured under mask are not the same line. Measure the coupon in the same state as the product.

Three ways a coupon report lies

The coupon is from another panel. Sometimes it is a golden panel kept for the purpose, sometimes it is the panel the process engineer measured while the line was being set up. Either way it describes a different etch and a different lamination. The coupon has to carry the panel identification and the lot, or the report has no object.

The coupon does not match the product. Different copper weight, different mask, a differential pair at a gap nobody specified, only the outer layers represented when the controlled impedance is on layer 3 and 6. Any of these turns the report into a general statement about capability.

The tolerance was never agreed. A report showing 48.2 ohms on a 50 ohm target passes at ten percent and fails at two. Suppliers quote the tolerance they can hold, and customers assume the tolerance their simulator used. Write it into the fabrication drawing with the layer, the target and the acceptance, before the first panel is made.

What SUPERB puts in the report

A TDR coupon on every panel, measured on each layer that carries controlled impedance, against a tolerance of plus or minus 5 percent with plus or minus 2 percent on critical nets. The report lists the target, the measured value, the layer and the panel identification. Core Dk is 4.2 and mask Dk 3.5, so the numbers are computed for the masked condition. Layer to layer registration is held at 0.025 to 0.06 mm and back drilled stubs at 0.25 mm or less with depth control of 0.05 mm, both of which show up in the same measurement.

The window is described on PCB capability, the high speed build route on high speed digital PCB, the RF side on RF and microwave PCB, and what loss does to a link budget on 112G PAM4 insertion loss. SUPERB PCB and SUPERB Assembly work the same lot, so the coupon, the panel and the assembly share one record.

What to send

Send the stackup with the target impedance on every controlled layer, the tolerance per net class, whether the coupon is measured bare or masked, the coupon requirement per panel or per lot, and the acceptance class. If the budget in the stackup is tighter than the tolerance the drawing allows, that is worth resolving in the quote rather than in the first article meeting.

Need a quote or a review of your files against these criteria? pcba@superb-tech.com · WhatsApp +86 133 9608 1443