What Is PCB Impedance Recovery?
PCB impedance recovery is the process of reverse-engineering a board’s controlled-impedance parameters—trace width, dielectric thickness, copper weight, and material Dk—from a physical sample when the original design files are unavailable. Our engineers cross-section the board, measure geometry under a calibrated microscope, run 2D field-solver simulations, and produce a new stackup and Gerber set that reproduces the original characteristic impedance (Z₀) within ±7% or better. The result is a manufacturable file package that any qualified fabricator can build to the same impedance targets as the original design.
Why Controlled Impedance Matters in a Board Copy
Any board carrying signals above roughly 100 MHz—USB 2.0/3.0, HDMI, DDR3/4/5, PCIe, Ethernet, or RF front-ends—relies on trace geometry to maintain a target impedance, typically 50 Ω single-ended or 90–100 Ω differential. If you copy the copper artwork pixel-for-pixel but ignore the stackup, the fabricated clone can ring, overshoot, or fail eye-mask tests entirely.
Impedance is not just a trace-width problem. It depends on the interaction of four variables: trace width (W), dielectric height (H), copper thickness (T), and the laminate’s dielectric constant (Dk). Change any one—say the fab substitutes FR-4 with Dk 4.5 instead of the original’s 4.2—and a nominally 50 Ω line drifts to 47 Ω or worse. On a 10 Gbps differential pair, that 6% mismatch creates return-loss failures visible on a TDR plot.
How We Recover Impedance From a Physical Board
Step 1 — Cross-Section and Measurement
We pot a coupon (or a sacrificial area of the board) in epoxy, polish it, and photograph the cross-section under a metallurgical microscope at 200×–500× magnification. This gives us direct measurements of trace width (typically ±0.3 mil accuracy), copper thickness, prepreg and core heights, and solder-mask thickness over traces. For boards with 6+ layers, we repeat the process at multiple locations to capture any asymmetry in the stackup.
Step 2 — Material Identification
Color, weave pattern, and thickness of the glass cloth help narrow the laminate family: standard FR-4 (Dk ≈ 4.2–4.5 at 1 GHz), mid-loss Megtron 4 (Dk ≈ 3.8), or low-loss Rogers 4350B (Dk ≈ 3.48). When the board operates above 3 GHz, getting the Dk wrong by even 0.2 shifts Z₀ by 3–5%, so we confirm with a dielectric-constant measurement or by TDR-matching a known-length trace on the sample board.
Step 3 — Field-Solver Simulation
Measured dimensions feed into a 2D field solver (Polar Si9000 or equivalent). We model every impedance-critical net class: single-ended microstrip, embedded microstrip, edge-coupled stripline, and broadside-coupled stripline. The solver outputs the predicted Z₀, and we compare it against TDR readings taken on the original board. If the delta exceeds ±2 Ω, we iterate on Dk or geometry until the model converges.
Step 4 — Stackup Specification and Gerber Adjustment
Once the simulation locks in, we write a fabrication stackup note specifying target impedances, layer assignments, prepreg/core part numbers, and finished copper weights. If the original trace widths don’t match a standard fab capability (e.g., the original used 4.2 mil traces on a process that now holds ±0.5 mil), we adjust width and dielectric height together to maintain Z₀ while staying within the new fab’s design rules.
Step 5 — Coupon Verification
The fabricator builds impedance test coupons alongside the production panel. We specify coupon locations and net classes per IPC-2141A. Coupons are TDR-tested before panels ship. Acceptance criteria: ±7% of target for standard FR-4 builds, ±5% for low-loss materials.
Microstrip Impedance Calculator
Use the calculator below to estimate single-ended microstrip impedance from your measured or planned dimensions. Enter trace width, dielectric height, copper thickness, and Dk. The tool returns the approximate Z₀ so you can sanity-check cross-section measurements or evaluate stackup alternatives before committing to fabrication.
PCB design
Microstrip Impedance Calculator
Single-ended impedance of a surface trace above one reference plane. This is the geometry of most 50 Ω signals on the top layer of a four-layer board.
How this is calculated
Uses the IPC-2141 closed-form approximation Z0 = 87 / √(εr + 1.41) · ln(5.98h / (0.8w + t)).
It is accurate to a few percent only inside its validity window: 0.1 < w/h < 3.0 and 1 < εr < 15. Outside that range the tool flags the result rather than hiding it. For a board you are about to release, treat any closed-form number as a first pass and have the fabricator run a field solver against the stackup they will actually press, since their real Dk, glass weave and etch compensation move the answer by more than the formula error.
Dk for standard FR-4 is not a constant. It drifts from about 4.5 at 100 MHz to about 4.2 at 1 GHz, and varies with resin content, so the value your laminate datasheet quotes at one frequency is the one to enter.
How to read the result: The output assumes a lossless, infinite-ground-plane microstrip model. Real-world boards add solder mask (which lowers Z₀ by 1–3 Ω on outer layers) and finite ground-plane width. Treat the calculator output as a ±3% starting point; a 2D field solver with solder-mask modeling refines it further.
Key Variables That Affect Recovered Impedance
| Variable | Typical Range | Z₀ Sensitivity | Measurement Method |
|---|---|---|---|
| Trace width (W) | 3–15 mil | ~1 Ω per 0.5 mil change | Cross-section microscopy |
| Dielectric height (H) | 3–12 mil | ~2 Ω per 1 mil change | Cross-section microscopy |
| Copper thickness (T) | 0.6–2.8 mil (½–2 oz) | ~0.5 Ω per 0.3 mil change | Cross-section or etch-weight |
| Dielectric constant (Dk) | 3.2–4.8 | ~3 Ω per 0.2 Dk shift | Material ID + TDR correlation |
| Solder-mask thickness | 0.4–1.5 mil over trace | 1–3 Ω on outer layers | Cross-section or profilometry |
Common Failure Modes in Impedance Recovery
- Wrong Dk assumption: Using generic “FR-4 = 4.5” when the original used a mid-loss laminate at 4.0 creates a systematic 5–8% impedance error across every net.
- Ignoring etch factor: Inner-layer traces etch with a trapezoidal profile. Measuring only the top width overstates W and underestimates Z₀ by 2–4 Ω.
- Solder-mask omission: Outer-layer microstrips covered by solder mask behave as embedded microstrips. Omitting this from the model inflates predicted Z₀ by 1–3 Ω.
- Asymmetric stackup: Some legacy boards use an intentionally asymmetric stackup for cost reasons. Assuming symmetry misplaces the reference plane and skews differential impedance by up to 10%.
- Sample damage: Boards with inner-layer delamination introduce air gaps that locally raise Z₀. We identify delaminated zones during cross-section and exclude them from measurement.
Impedance Recovery vs. Standard Board Copy
| Aspect | Standard PCB Copy | Impedance-Recovered Copy |
|---|---|---|
| Deliverable | Gerber + BOM + schematic | Gerber + BOM + schematic + impedance stackup note |
| Stackup documentation | Generic layer order | Material callouts, Dk, prepreg/core PN, target Z₀ per net class |
| Trace-width accuracy | ±1 mil | ±0.3 mil (field-solver-matched) |
| Impedance guarantee | None | ±7% (standard) / ±5% (low-loss) |
| Verification | Visual DRC | TDR coupon test per IPC-2141A |
| Typical added time | — | 2–4 working days |
| Added cost factor | — | +15–30% over base copy price |
If your board runs only low-speed logic and power, a complete board copy project without impedance recovery is sufficient. For anything with signal integrity requirements, impedance recovery is not optional—it is the difference between a working clone and an expensive paperweight.
Which Board Types Need Impedance Recovery Most?
High-speed digital and RF boards are the obvious candidates, but we also encounter impedance-critical designs in industrial motor drives (controlled-impedance gate-drive traces), medical imaging (LVDS video buses), and telecom backplanes (25 Gbps+ channel speeds). Specific board constructions that raise the difficulty level include:
- HDI boards with build-up layers and microvias—dielectric heights under 3 mil amplify any measurement error. See our HDI boards with stacked microvias guide for details.
- Flex and rigid-flex where polyimide Dk (≈ 3.2–3.4) differs significantly from FR-4 and adhesive layers add thickness uncertainty. Our flex PCB recovery workflow covers the added steps.
- Boards with blind or buried vias that constrain reference-plane continuity and create impedance discontinuities at via transitions.
What We Deliver
- Full Gerber RS-274X or ODB++ file set with impedance-matched trace geometry.
- Fabrication stackup drawing: layer order, material callouts, Dk at target frequency, copper weights, finished dielectric heights.
- Impedance control table listing every controlled net class, target Z₀, tolerance band, and coupon reference.
- Cross-section photographs with annotated measurements.
- BOM and schematic (if part of the full reverse engineering engagement).
Turnaround and Pricing
Impedance recovery adds 2–4 working days on top of the base copy timeline. For a 4-layer board with one or two impedance classes, expect the impedance-recovery portion to start around $300–$600 depending on complexity. 8+ layer boards with multiple impedance classes (single-ended, edge-coupled differential, broadside-coupled) run higher. Cross-section costs are included. TDR verification on the original board, if requested, adds one day.
Check feasibility for your project
Send the chip marking or board details and tell us what you need to recover. Add clear photos if available.
Request a feasibility reviewFrequently Asked Questions
Can you recover impedance without destroying the original board?
Partial recovery is possible using TDR measurements and educated Dk assumptions, but accuracy drops to roughly ±10–12%. For ±7% or better, we need at least one cross-section coupon. If the board has a test coupon strip in the panel border, we can section that and leave the functional board intact. We discuss options during the feasibility review.
What if the original laminate is discontinued?
We identify a current-production substitute with matching Dk and loss tangent at the operating frequency, then adjust trace width and dielectric height in the field solver to hit the same Z₀. This is routine—laminate product lines change every few years, and the physics doesn’t care about the brand name, only the electrical properties.
Do you support differential impedance recovery?
Yes. We recover both odd-mode (single-ended) and differential impedance for edge-coupled and broadside-coupled pairs. The impedance-controlled stackup reconstruction guide walks through the differential-pair workflow in detail. Typical differential targets are 85 Ω (USB 2.0), 90 Ω (USB 3.x, HDMI), and 100 Ω (Ethernet, PCIe).
Is impedance recovery needed for boards under 100 MHz?
Generally no. Below 100 MHz, signal wavelengths are long enough that reflections from impedance mismatches are negligible. Exceptions include long bus traces (>15 cm), fast edge-rate logic (rise times under 1 ns regardless of clock frequency), and any analog RF path. When in doubt, send us the board for a free feasibility assessment.
Check feasibility for your project
Send the chip marking or board details and tell us what you need to recover. Add clear photos if available.
Request a feasibility reviewWorking on a board like this?
Send the chip marking or two photos. You get feasibility, lead time and price within 24 hours, and the check costs nothing.
Get a free quote