Calculating Z Axis Permittivity Anisotropy for Stripline Controlled Impedance Routing
Calculate stripline impedance by applying the geometric mean of in-plane and z-axis permittivity to sidewall fringing fields to eliminate 2-ohm routing offsets.
Grain
Laminate data sheets routinely quote a single dielectric constant measured perpendicular to the copper foil at 10 GHz using IPC-TM-650 Method 2.5.5.5 clamp stripline testing or split-post dielectric resonators. That single value misrepresents the electromagnetic environment of a stripline conductor. Standard printed circuit substrate materials combine woven E-glass yarn, possessing a relative permittivity between 6.3 and 6.8 at 10 GHz, with hydrocarbon, polyphenylene ether, or epoxy resin systems exhibiting relative permittivities between 2.6 and 3.2.
Mechanical lamination settles these continuous filaments into alternating strata parallel to the board surfaces. The resulting composite behaves as a uniaxial anisotropic medium. The dielectric tensor resolves into two distinct orthogonal components: an out-of-plane permittivity along the z-axis and an in-plane permittivity across the x-y plane.
Woven glass styles directly dictate the magnitude of this tensor split. Lightweight weaves like 1035 or 1067 carry high resin fractions above 65 percent by weight, keeping the composite bulk permittivity low while suppressing directional variation. Heavy commercial fabrics such as 7628 contain thick yarn bundles with resin fractions dropping below 45 percent.
The in-plane permittivity of a 7628 prepreg layer routinely exceeds its out-of-plane permittivity by 12 to 18 percent because horizontal electric field lines travel along continuous glass filaments without interruption. Vertical field lines cross alternating barriers of low-permittivity resin pockets and high-permittivity glass yarns, acting as capacitors in series.
Woven glass fabrics with high yarn densities concentrate electric displacement along continuous horizontal glass bundles.
When routing single-ended or tightly coupled differential striplines, treating the substrate as an isotropic bulk medium introduces systematic impedance errors. Stripline field distributions do not orient purely along the z-axis. While the broad top and bottom faces of the copper conductor launch vertical displacement fields straight into the ground planes, the sidewalls launch fringing fields that project horizontally across the x-y plane.
The relative contribution of horizontal fringing grows as line width shrinks relative to ground-to-ground dielectric spacing. An impedance calculation that assumes a single z-axis permittivity overstates characteristic impedance by ignoring the faster capacitive loading produced by the elevated in-plane permittivity.
Heavier glass weaves elevate horizontal field loading while light glass styles suppress directional divergence.
Fringe
Total capacitance per unit length of a symmetric stripline trace decomposes into two parallel plate capacitances and four sidewall fringe capacitances. Let the ground plane spacing equal b, the conductor width equal w, and the copper foil thickness equal t. The vertical displacement field between the wide conductor faces and the reference planes aligns strictly with the z-axis.
The parallel plate capacitance term depends directly on the out-of-plane relative permittivity component:
Cp = 2 ε0 εrz
The sidewall fringing capacitance occupies a mixed vector field. Electric field lines leave the vertical copper sidewall horizontally, bending through ninety degrees before terminating on the ground planes. Conformal mapping techniques convert this anisotropic region into an equivalent isotropic domain through a coordinate transformation.
Scaling the horizontal spatial axis by the square root of the anisotropy ratio transforms the elliptical field contours into circular arcs. The effective fringing permittivity governing the sidewall capacitance term becomes the geometric mean of the tensor components:
εrf = (εrz × εrxy)0.5
Is Out-of-Plane Permittivity Always Lower than In-Plane?
Planar lamination aligns the higher-permittivity reinforcement fibers parallel to the foil boundaries across every standard glass-reinforced grade. The in-plane permittivity remains strictly higher than the out-of-plane value for all woven E-glass, NE-glass, and quartz-reinforced prepregs. Ceramic-filled hydrocarbon laminates exhibit reduced directional variation because isotropic ceramic particles disperse uniformly through the matrix, though the woven glass carrier still forces a measurable delta between the two principal axes.
| Laminate Grade and Weave | Resin Content (%) | Out-of-Plane Dk (εrz) | In-Plane Dk (εrxy) | Anisotropy Ratio (εrxy / εrz) |
|---|---|---|---|---|
| Standard FR-4 (7628 Glass) | 43 | 4.25 | 4.82 | 1.134 |
| High-Tg Mid-Loss (2116 Glass) | 54 | 3.80 | 4.21 | 1.108 |
| Low-Loss PPE/PPO (1080 Glass) | 62 | 3.45 | 3.72 | 1.078 |
| PTFE / Microfiber Ceramic Fill | N/A | 3.00 | 3.06 | 1.020 |
| High-Speed Low-Loss (1035 Glass) | 68 | 3.15 | 3.32 | 1.054 |
Trace sidewall geometry alters the balance between vertical and horizontal capacitance components. Chemical subtractive etching produces a trapezoidal cross-section with an etch factor typically between 2.0 and 3.0. The top width of the conductor is narrower than the base width by an amount equal to roughly half the copper foil thickness.
The sloping sidewalls increase the physical surface area launching horizontal fringing fields into the surrounding prepreg. For narrow traces where the line width is less than the ground-to-ground dielectric height, fringing capacitance contributes up to 45 percent of total trace capacitance.
Conformal transformation scales the horizontal coordinate space by the square root of the dielectric anisotropy ratio.
Ignoring the in-plane permittivity component during stackup synthesis pulls the fabricated line impedance below target, triggering layout revisions or high-frequency link attenuation.
Arithmetic
Extracting the true characteristic impedance of an anisotropic stripline requires transforming the permittivity tensor into an effective bulk dielectric value for the specific cross-section. The analytical procedure treats the conductor as an equivalent system containing two distinct dielectric domains operating in parallel.
- Tensor Component Extraction evaluates the raw resin volume fraction and woven glass yarn density to establish the base out-of-plane permittivity and in-plane permittivity at the operational Nyquist frequency.
- Coordinate Space Normalization applies a geometric transformation factor k = (εrz / εrxy)0.5 to the trace line width and ground plane clearance dimensions.
- Capacitance Segmentation computes the parallel vertical capacitance using the out-of-plane constant and the fringing capacitance using the geometric mean permittivity.
- Impedance Synthesis sums the segmented capacitance components to extract total transmission line capacitance and propagation velocity.
Why Stripline Geometries Magnify Out-of-Plane Dispersion?
Phase velocity along a single-ended stripline equals the speed of light in vacuum divided by the square root of the total effective permittivity. Unlike microstrip, where part of the wave propagates through ambient air, stripline fields remain fully contained within the solid composite. Because the horizontal field vector interacts with the higher in-plane permittivity while the vertical vector interacts with the lower out-of-plane permittivity, the effective dielectric constant changes as line width narrows.
The trace geometry itself shifts the relative weight of the two tensor components.
The closed-form analytical expression for effective stripline permittivity with anisotropic dielectrics combines these factors into a single formulation:
εeff = εrz ×
In this equation, Cf0 represents the fringe capacitance calculated for an isotropic medium with relative permittivity equal to unity. The characteristic impedance Z0 follows directly from the total line capacitance Ct and effective permittivity:
Z0 = (μ0 ε0 εeff)0.5 / Ct = 1 / (vp Ct)
| Design Parameter | Isotropic Assumption | Anisotropic Calculation | Observed Delta |
|---|---|---|---|
| Substrate Out-of-Plane Dk (εrz) | 3.60 | 3.60 | 0.00 |
| Substrate In-Plane Dk (εrxy) | 3.60 (Assumed) | 4.05 (Extracted) | +0.45 |
| Conductor Base Width (w) | 0.145 mm | 0.145 mm | 0.000 mm |
| Parallel Capacitance (Cp) | 42.1 pF/m | 42.1 pF/m | 0.0 pF/m |
| Fringe Capacitance (Cf) | 24.8 pF/m | 26.3 pF/m | +1.5 pF/m |
| Total Line Capacitance (Ct) | 66.9 pF/m | 68.4 pF/m | +1.5 pF/m |
| Calculated Line Impedance (Z0) | 49.8 Ω | 47.4 Ω | -2.4 Ω |
Assume a high-density interconnect stackup using 2116-type prepreg cores. The out-of-plane permittivity measured by clamped resonance equals 3.60 at 10 GHz, while the in-plane tensor component sits at 4.05. A standard 2D field solver ignoring anisotropy calculates a nominal line width of 0.145 mm to hit 50.0 ohms.
The actual anisotropic line capacitance reaches 68.4 pF/m due to the added fringe displacement along the horizontal glass weave. Fabricated traces yield a true characteristic impedance of 47.4 ohms. The 2.4-ohm error consumes nearly half of the standard plus-or-minus five percent manufacturing tolerance window before etching variations or dielectric thickness tolerances occur.
Fixture
Verifying laminate anisotropy requires distinct test fixtures capable of isolating orthogonal field orientations. Single-cavity perturbation methods and split-post dielectric resonators measure the out-of-plane permittivity because their TE01δ resonant modes establish electric fields oriented parallel to the circular specimen face. To measure in-plane permittivity, laboratories utilize Fabry-Perot open resonators, stripline resonators with balanced balanced-line excitation, or multi-line Thru-Reflect-Line test coupons configured with varying line widths.
- Split-Cylinder Resonator Fixtures clamp unclad or etched substrate disks inside a cylindrical metallic cavity, exciting transverse electric modes at discrete frequencies between 10 GHz and 28 GHz to extract out-of-plane permittivity with repeatability within 0.5 percent.
- In-Plane Waveguide Cavities orient rectangular substrate coupons parallel to the dominant waveguide E-field vector, capturing the x-y dielectric response across 8.2 GHz to 12.4 GHz X-band ranges.
- Balanced Stripline Coupons place narrow etched conductors between identical prepreg packages, isolating sidewall fringe capacitance through differential line width extractions defined under IPC-TM-650 Method 2.5.5.5.1.
- Interdigital Surface Capacitors deposit fine interdigitated comb structures onto bare laminate surfaces, directing electric flux lines across the surface resin layer and top glass weave bundles to extract in-plane tensor parameters.
IPC-TM-650 Method 2.5.5.5 measures out-of-plane permittivity under clamped mechanical pressure without accounting for horizontal prepreg yarn alignment.
Coupon extraction data reveals systematic shifts between raw laminate datasheets and finished multilayer boards. High lamination pressures squeeze resin into copper clearance cavities, increasing the local glass-to-resin ratio directly beneath narrow stripline conductors. The local out-of-plane permittivity climbs by 0.05 to 0.15 relative to nominal raw material sheets.
Fabricators frequently state that their impedance models incorporate empirical offset factors to absorb these material shifts without explicitly breaking out the directional permittivity tensor.
Ledger
Failing to account for dielectric anisotropy directly impacts production cost and panel yields. When pre-production impedance coupons fail a narrow five-percent specification window due to unaccounted fringe capacitance, fabricators adjust line widths on subsequent artwork iterations. Modifying photo-tooling adds tooling charges and extends production cycles by several days.
If line width reductions push conductor geometries below standard chemical etching process thresholds, manufacturing yields decline across production lots.
High-frequency routing demanding tight impedance tolerances forces a careful trade-off between laminate raw material pricing and geometric tolerance management.
- Glass Fabric Selection determines the anisotropy spread, with spread-glass weaves like 1078 or 1067 reducing tensor deltas at a price premium of fifteen to thirty percent over standard 2116 or 7628 fabrics.
- Resin System Formulation dictates base loss tangent and permittivity stability, where polyphenylene oxide and fluoropolymer matrices reduce raw anisotropy relative to standard multifunctional epoxy systems.
- Etch Compensation Allowances require modification to account for anisotropic fringe growth on trapezoidal sidewalls, preventing over-etching on high-density routing layers.
Engineering drawings specifying controlled impedance stripline layers protect yield by defining target geometries based on multi-axis permittivity parameters. IPC-6012 Class 3 fabricators evaluate impedance coupons using time-domain reflectometry with specified receiver rise times matched to operational signal edges. Specifying stackup requirements under IPC-4101 slash sheet parameters with explicit anisotropy limits ensures the chosen manufacturing facility delivers boards meeting physical impedance targets without costly re-tooling delays.