ECParts Toolkit LogoECParts Toolkit

Engineering Guide · PCB

PCB Controlled Impedance: Microstrip, Stripline, and Differential Pairs

Design controlled-impedance PCB traces using microstrip, stripline, differential-pair geometry, stackup data, dielectric assumptions, and fabrication limits.

Updated September 24, 2026 · Intermediate · 18 min read

From PCB Geometry to Controlled Impedance

Controlled-impedance design connects an electrical target to a manufacturable PCB stackup. The target comes from an interface or system requirement; the board implementation comes from trace width, copper thickness, dielectric height, relative permittivity, reference planes, and, for differential pairs, spacing and coupling.

A PCB trace needs transmission-line treatment when its propagation delay is significant relative to signal transition time and the resulting reflections matter. That decision cannot be reduced to one universal clock-frequency threshold. Once distributed behavior matters, characteristic impedance Z0 becomes distinct from DC trace resistance.

This guide follows the exact closed-form approximations used by the ECParts PCB impedance calculators. It focuses on geometry, stackup, design iteration, and fabrication handoff. Reflection coefficient, VSWR, return loss, matching networks, quarter-wave sections, and general electrical-length theory remain with the RF Impedance Matching Guide.

Controlled-Impedance Workflow

StepDecisionEngineering Action
1Obtain the targetUse the interface, device, or system specification. Do not assume every single-ended trace is 50 Ω or every differential pair is 100 Ω.
2Choose the structureIdentify an outer-layer microstrip or an embedded stripline and its continuous reference plane or planes.
3Use an actual stackupCollect finished trace width, finished copper thickness, dielectric height, and a defensible relative-permittivity assumption.
4Estimate single-ended Z0Apply the calculator model that matches the selected geometry; do not exchange microstrip and stripline equations.
5Add differential couplingFor a pair, include the calculator spacing input and the coupling term rather than assuming Zdiff = 2Z0.
6Explore sensitivityChange width, height, spacing, copper thickness, and Er one at a time while remaining inside the approximation's useful range.
7Review fabricationConfirm material, finished copper, etch compensation, minimum geometry, and controlled-impedance capability with the fabricator.
8Verify the final stackupRecalculate with the released stackup and use fabricator modeling or a field solver when tolerance or geometry demands it.

Geometry and Output Definitions

SymbolMeaningPractical Interpretation
WFinished trace widthNarrower or wider copper changes characteristic impedance and must reflect etching, not only nominal CAD width.
HDielectric height used by the selected modelIt is the trace-to-reference geometry represented by the calculator, not total board thickness.
TFinished copper thicknessThe current ECParts approximations include T in the logarithm denominator.
ErRelative dielectric constantUse a value appropriate to material, construction, frequency, resin content, and the source test method.
SDifferential-pair spacing inputThe current calculator consumes S directly in S/H. Its UI does not declare edge-to-edge versus center-to-center, so do not transfer geometry from another tool without confirming the convention.
Z0Single-ended characteristic impedanceThis is a wave quantity of the modeled structure, not the trace's DC resistance.
ZdiffDifferential characteristic impedance estimateThe current calculator derives this from single-ended Z0 and an empirical spacing-dependent coupling factor.

ECParts Microstrip and Stripline Models

The calculators use natural logarithms and SI dimensions internally. Because every geometric term appears as a ratio, consistent dimensions such as mm, mil, or µm give the same result after conversion. The logarithm argument must exceed one, and the calculator rejects geometry outside the useful range of its approximation.

Formula reference

Microstrip approximation used by ECParts

Z0 = [87 / √(Er + 1.41)] × ln[5.98H / (0.8W + T)]εeff = (Er + 1)/2 + [(Er - 1)/2] / √(1 + 12H/W)v = c / √εeff

Variable definitions

W
trace width
H
dielectric height
T
copper thickness
Er
relative dielectric constant
εeff
effective dielectric constant
v
propagation velocity

Formula reference

Stripline approximation used by ECParts

Z0 = (60 / √Er) × ln[4H / (0.67π(W + T))]εeff = Erv = c / √Er

Variable definitions

The adopted implementation is a symmetric stripline-style closed-form estimate.
Do not reuse microstrip H or effective-permittivity assumptions without matching the physical stackup.
PropertyMicrostripStripline
Conductor locationOuter layer above a primary reference planeEmbedded between reference planes in the adopted symmetric approximation
Field distributionPartly in dielectric and partly in surrounding mediumMore strongly confined within dielectric
Dielectric treatmentUses calculated effective permittivityUses Er directly in the current model
ECParts Z0 coefficient87 / √(Er + 1.41)60 / √Er
ECParts logarithmln[5.98H / (0.8W + T)]ln[4H / (0.67π(W + T))]
Implementation warningSolder mask and external environment can shift the resultPlane spacing and conductor centering must match the model

Differential-Pair Coupling

A coupled pair does not generally have differential impedance equal to exactly twice the isolated single-ended impedance. The ECParts calculator first evaluates the appropriate microstrip or stripline Z0, then applies a structure-specific coupling factor based on S/H.

Formula reference

Differential-pair approximation used by ECParts

Microstrip: K = 0.48e^(-0.96S/H)Stripline: K = 0.347e^(-2.9S/H)Zdiff = 2Z0(1 - K)

Variable definitions

K
empirical coupling factor
S
calculator trace-spacing input
H
dielectric height input
Z0
single-ended baseline from the selected structure
Zdiff
differential impedance estimate

As S/H grows, K approaches zero and Zdiff approaches 2Z0. Smaller S/H increases the modeled coupling and moves Zdiff farther below 2Z0. The current calculator does not expose odd-mode, even-mode, or common-mode impedance as separate outputs.

Spacing convention boundary: the implementation labels S as “Trace spacing S” and uses it directly, but does not encode whether an external drawing should interpret it as edge-to-edge or center-to-center. Treat it as a model-specific input, document the convention in the design, and confirm the geometry with the fabricator or field-solver workflow.

Stackup and Fabrication Reality

A calculator can explore geometry, but the fabricator controls available laminates, pressed dielectric thickness, finished copper, etch compensation, and process capability. Controlled impedance is therefore a specification-and-feedback process rather than a one-time equation.

VariableWhy It MovesDesign Action
Trace widthEtch removes copper laterally and the finished width may differ from the design value.Ask whether the fabricator adjusts artwork for impedance.
Copper thicknessBase copper plus plating can change finished thickness and sidewall shape.Use finished copper assumptions rather than copper-weight shorthand alone.
Dielectric heightPressed thickness and laminate construction vary by stackup and process.Use the released stackup, not generic board thickness.
Dk / ErNominal datasheet values depend on method, frequency, resin content, and glass construction.Use the fabricator's modeling value where available.
Pair spacingEtch variation changes coupling as well as individual trace impedance.State the spacing convention and tolerance explicitly.
DiscontinuitiesVias, pads, connectors, neck-downs, plane changes, and reference gaps are outside the uniform-line equations.Review transitions separately and preserve return paths.

Keep a continuous reference plane beneath or around the line. Plane splits, return-path gaps, vias, connectors, and layer transitions create discontinuities that these uniform-line formulas do not calculate. A line can meet its nominal cross-section target and still perform poorly at a discontinuity.

Worked Examples

Example 1: Microstrip Geometry Estimate

  • Given: W = 0.15 mm, H = 0.18 mm, T = 0.035 mm, and Er = 4.2.
  • Model argument: 5.98 × 0.18 / (0.8 × 0.15 + 0.035) = 6.944516.
  • Z0 = 87 / √(4.2 + 1.41) × ln(6.944516) = 71.184 Ω.
  • εeff = (4.2 + 1)/2 + [(4.2 - 1)/2]/√(1 + 12 × 0.18/0.15) = 3.00772.
  • Propagation velocity = c/√εeff = 1.72863 × 10^8 m/s.
  • Interpretation: this is a first-pass geometry estimate, not a fabrication guarantee or a DC-resistance result.

Example 2: The Same Numbers in the Stripline Model

  • Given: W = 0.15 mm, H = 0.18 mm, T = 0.035 mm, and Er = 4.2, now selecting stripline.
  • Model argument: 4 × 0.18 / [0.67π(0.15 + 0.035)] = 1.848997.
  • Z0 = 60/√4.2 × ln(1.848997) = 17.995 Ω.
  • The current stripline model uses εeff = Er = 4.2, giving v = 1.46284 × 10^8 m/s.
  • Interpretation: identical numeric W, H, T, and Er do not describe electrically interchangeable structures. H and reference-plane geometry must match the selected model.

Example 3: Coupled Microstrip Differential Pair

  • Given: the Example 1 microstrip plus S = 0.15 mm using the calculator's spacing input.
  • Single-ended baseline: Z0 = 71.184 Ω.
  • Coupling factor: K = 0.48e^(-0.96S/H) = 0.215678, or 21.568%.
  • Zdiff = 2Z0(1 - K) = 111.662 Ω.
  • This differs from 2Z0 = 142.368 Ω because the pair is coupled in the adopted approximation.
  • Interpretation: spacing materially changes the result. Confirm whether a fabrication or field-solver workflow defines spacing the same way before transferring this geometry.

Example 4: Unit and Geometry Iteration

  • The Example 1 dimensions are 5.9055 mil, 7.0866 mil, and 1.3780 mil because 1 mil = 0.0254 mm.
  • Entering either unit set returns the same estimate because the calculator converts all dimensions to meters before applying ratios.
  • Within this model's valid region, increasing W while holding H, T, and Er fixed lowers the calculated Z0.
  • Use this directional behavior to approach a target, then rerun with the fabricator's actual stackup and finished dimensions.

Practical Limitations

  • The formulas are closed-form approximations for early stackup exploration and sanity checking.
  • The microstrip and stripline equations are the equations implemented by ECParts; this guide does not assign an IPC, Wheeler, or Hammerstad label that the source code does not establish.
  • The model includes finite copper thickness through T but does not model trapezoidal etch profiles.
  • Microstrip effective permittivity is estimated; the model does not calculate broadband dielectric dispersion.
  • The calculators do not model dielectric loss, conductor loss, skin effect, copper roughness, or total insertion loss.
  • Solder mask, glass weave, material anisotropy, resin content, and local copper environment are not explicitly modeled.
  • The differential formula is an empirical coupling approximation, not an odd/even-mode field solution.
  • The current UI calls S trace spacing but does not formally state edge-to-edge or center-to-center; treat S as model-specific until the convention is documented.
  • Uniform-line formulas do not model vias, pads, connectors, neck-downs, plane splits, reference changes, or return-path discontinuities.
  • Use a 2D/3D field solver or fabricator impedance service for tight tolerances, complex stackups, unusual geometries, or very high-speed channels.

Common Mistakes

  • Confusing characteristic impedance with DC trace resistance.
  • Using the microstrip equation for an embedded stripline, or vice versa.
  • Using total PCB thickness where the calculator expects dielectric height to the reference structure.
  • Assuming every single-ended trace should be 50 Ω or every differential pair should be 100 Ω.
  • Assuming differential impedance always equals twice the isolated single-ended impedance.
  • Moving spacing values between tools without checking edge-to-edge versus center-to-center definitions.
  • Using a nominal laminate Dk as an exact broadband design value.
  • Ignoring finished copper, etch compensation, dielectric tolerance, and the fabricator's available stackups.
  • Treating controlled-impedance spacing as electrical-safety clearance or creepage.
  • Treating a closed-form calculator result as a fabrication or compliance guarantee.

Engineering Boundaries

The calculators do not use frequency as an input and do not predict dielectric dispersion, loss tangent, skin effect, copper roughness, or total channel insertion loss. They estimate characteristic impedance and, for the single-ended tool, idealized propagation velocity and delay.

Controlled-impedance spacing is not a substitute for creepage or clearance. Safety spacing depends on working voltage, insulation type, pollution degree, material group, altitude, and the applicable requirements. Those topics remain reserved for a dedicated PCB safety-spacing reference.

For tight impedance tolerance, unusual geometry, asymmetric stackups, broad frequency ranges, or critical high-speed channels, verify with a 2D/3D field solver, fabricator impedance modeling, test coupons, and measurements on the finished construction.

Support reference

FAQ

What is controlled impedance on a PCB?

Controlled impedance is a design and fabrication process in which trace geometry, reference planes, dielectric construction, and manufacturing controls are selected so a transmission line meets a specified characteristic-impedance range.

What is the difference between microstrip and stripline?

A microstrip is generally an outer-layer conductor referenced primarily to a plane, with fields in both dielectric and the surrounding medium. A stripline is embedded between reference planes, so its field confinement and impedance equation differ.

Is PCB trace impedance the same as DC resistance?

No. DC resistance describes conductor loss from resistivity, length, and cross-sectional area. Characteristic impedance describes the voltage-to-current relationship of a traveling wave in a transmission-line geometry.

What determines microstrip impedance?

In the ECParts approximation, trace width W, dielectric height H, copper thickness T, and relative dielectric constant Er determine Z0. Real boards also involve solder mask, copper profile, roughness, frequency-dependent material behavior, and fabrication tolerances.

What determines differential pair impedance?

The current calculator starts with single-ended Z0 and applies a spacing-to-height coupling factor. Width, dielectric height, copper thickness, Er, structure type, and spacing input all affect the estimate.

Is differential impedance always twice single-ended impedance?

No. It approaches 2Z0 only when coupling is weak in the adopted model. Closer coupling changes the differential impedance, so spacing cannot be ignored.

Does pair spacing affect differential impedance?

Yes. The ECParts model uses S/H in an exponential coupling term. Because spacing conventions differ among tools, confirm whether the source geometry is expressed using the same definition before comparing results.

Should every PCB trace be 50 ohms?

No. The target comes from the interface, device, connector, cable, or system specification. Many digital and differential interfaces use targets other than 50 Ω.

When does a PCB trace need transmission-line treatment?

It depends on signal transition time, interconnect delay, trace length, return path, and acceptable reflection. Clock frequency alone is not a universal threshold; edge rate and physical implementation matter.

Why can the PCB fabricator change the final trace width?

A fabricator may compensate artwork for etching, finished copper, dielectric construction, and the impedance target. The released stackup and process model can therefore require a different width from an early calculator estimate.

When should I use a field solver?

Use a field solver or fabricator impedance service for tight tolerances, complex or asymmetric stackups, solder-mask effects, unusual copper profiles, high-speed transitions, anisotropic materials, or geometries outside the closed-form approximation's useful range.

Are controlled-impedance spacing and safety clearance the same?

No. Pair spacing controls electromagnetic coupling in a signal structure. Clearance and creepage address insulation and safety constraints involving voltage, environment, material, altitude, and applicable standards.

Disclaimer

These closed-form equations provide first-pass engineering estimates. Confirm critical controlled-impedance designs with the released fabrication stackup, material data, field solving or fabricator modeling, impedance coupons where appropriate, and measurements on finished hardware.