Managing Anisotropic Permittivity Variation in High-Speed Sequential Lamination Substrates

Tensor

Electrical wave propagation in high-density interconnect substrates depends on the spatial orientation of the dielectric medium surrounding the copper conductors. Signals running at 28 GHz, 56 GHz, and 112 GHz PAM4 rates experience anisotropic permittivity variations because composite laminates combine physical phases with sharply different dielectric properties. Sequential lamination substrates use woven glass reinforcement filaments embedded in a polymer resin matrix, often combined with organic or inorganic fillers.

Standard E-glass fibers exhibit a relative dielectric constant around 6.6 at 10 GHz, while low-loss NE-glass sits near 4.6 and ultra-low loss quartz glass reaches approximately 3.8. By contrast, thermosetting resin systems like high-Tg epoxy, polyphenylene ether, and fluoropolymers have dielectric constants between 2.1 and 3.2. This structural asymmetry turns the substrate dielectric constant into a directional matrix where effective permittivity along the X- and Y-axes differs fundamentally from the Z-axis.

Dielectric properties shift across physical orthogonal axes.

In an unpressed prepreg sheet, glass fibers lie parallel to the planar axes in warp and fill directions. When conductors carry electromagnetic energy down a microstrip or stripline channel, the electric fields penetrate both the planar resin-glass matrix and the vertical insulation between layers. Permittivity in the X-Y plane reflects a parallel capacitive network formed primarily by continuous glass filaments and interstitial resin.

Out-of-plane permittivity along the Z-axis behaves as a series capacitive combination dominated by alternating layers of bulk resin and woven glass bundles. Sequential lamination amplifies this structural divergence by subjecting core sub-assemblies to multiple thermal and pressure cycles, driving resin displacement, glass bundle compaction, and local density gradients that alter the directional permittivity tensor at every sub-lamination stage.

Anisotropy in Woven Glass Substrates

Microstructural examination of high-speed laminates shows that physical fiber density varies systematically across the laminate cross-section. Warp yarns run continuously along the production roll, held under tension during prepreg manufacturing. Fill yarns run transversely with lower tension and higher spatial periodicity.

This mechanical asymmetry causes warp fibers to pack more tightly than fill fibers, yielding a higher volumetric glass fraction along the warp axis. High-frequency signals propagating parallel to warp yarns encounter a higher effective dielectric constant than signals running along fill yarns. The resulting planar phase delay mismatch between orthogonal channels creates spatial phase jitter that conventional simulation tools ignore when assuming an isotropic dielectric constant.

Multiple lamination steps shift the effective glass-to-resin ratio non-uniformly across the z-axis of sequential high-density interconnect stackups.

Sequential processing compounds planar anisotropy by introducing differential thermal expansion and strain relaxation across core layers. During the first lamination pass, a double-sided core undergoes initial resin gelation and vitrification under hydraulic pressure. When this sub-assembly returns to the press for subsequent steps to attach outer microvia layers, the cured core is reheated near or above its glass transition temperature.

The resin expands along the Z-axis while restricted in the X-Y plane by the fully cured glass fabric. This anisotropic thermal expansion distorts the glass weave pitch, inducing localized micro-displacements of glass filaments. Consequently, traces on inner layers undergo a permanent permittivity shift relative to traces on freshly added outer sequential layers.

Permittivity Tensor Components across Processing Axis

Mathematical modeling of signal velocity in sequential substrates requires treating permittivity as a second-rank tensor with distinct principal components. The diagonal elements represent dielectric response along panel length, panel width, and vertical stackup depth. High-frequency electromagnetic solvers calculating phase velocity and characteristic impedance must incorporate these directional values to predict transmission behavior accurately.

Resin flow alters glass density.

A signal trace carrying a 58 Gbd PAM4 waveform experiences an effective permittivity calculated from both in-plane components and the out-of-plane component, depending on trace topology. Microstrip conductors on outer layers send much of their fringing fields through air and adjacent soldermask, making them sensitive to planar permittivity variation. Stripline conductors embedded between ground planes concentrate their electric fields perpendicular to the copper plane, forcing out-of-plane permittivity to dominate total capacitive loading.

When sequential processing subjects sub-assemblies to repetitive heat cycles, the ratio between in-plane and out-of-plane permittivity widens. Fabricators who fail to account for this widening observe impedance shifts of 3 to 6 ohms between inner core striplines and outer sequential striplines, even when using identical nominal laminate grades throughout the stackup.

Physical Drivers of Directional Dielectric Variance

Substrate materials exhibit four primary physical drivers that enforce dielectric anisotropy during sequential lamination cycles. Addressing these drivers requires identifying how raw material selection interacts with press line dynamics during production.

• Glass Style Asymmetry involves structural differences between warp and fill fiber counts per inch, creating non-uniform dielectric constant distributions along orthogonal trace routing vectors.
• Resin Hydrodynamics covers the lateral movement of liquid resin during thermo-compression stages, leaving localized regions depleted of resin directly beneath heavy copper patterns.
• Multi-Pass Curing introduces incremental polymer cross-linking during successive lamination steps, altering the bulk material polarizability and lowering the dielectric loss factor non-uniformly.
• Filler Particle Migration describes the mechanical redistribution of inorganic micro-fillers such as silica or titanium dioxide under hydraulic resin flow during sub-stage press cycles.

Material suppliers frequently market high-frequency laminates as isotropic media based on single-frequency clamped stripline test data. Published datasheet figures reflect raw un-pressed core stock tested under single-stage IPC standard conditions, omitting the 0.12 out-of-plane permittivity shift that develops across three sequential press cycles.

Clamp

Hydraulic pressure and thermal profiles applied during multi-stage lamination directly dictate the final spatial distribution of resin and glass inside sequential substrates. When a bare core is pressed, liquid resin flows along the path of least resistance, filling clearance holes and copper windows before gelling. During sequential lamination, outer prepreg layers are laminated over previously cured core sub-assemblies that already feature etched copper patterns.

The relief profile of the inner copper conductors generates localized pressure spikes under the press plates. Areas directly above solid copper experience intense localized hydraulic squeezing, while adjacent etched clearance regions see lower pressure until resin fills the voids.

Sub-laminations compound structural tension.

This localized pressure variance drives differential resin movement. Under high pressure, low-viscosity resin flows away from the top of copper traces into adjacent clearance trenches, locally elevating the glass volume fraction directly above signal lines. Because woven glass has a significantly higher dielectric constant than the surrounding polymer matrix, traces over high-pressure zones see a spike in both in-plane and out-of-plane permittivity.

Conversely, traces running over deep clearance areas sit under resin-rich zones and experience lower effective permittivity. Managing this phenomenon requires precise control over the press cycle profile, aligning the thermal ramp rate with the hydraulic pressure curve.

Sequential Lamination Press Kinetics

Optimizing the press profile for sequential builds requires balancing vacuum levels, temperature ramp rates, pressure dwell times, and cooling rates. A standard high-reliability sequential process employs a multi-stage cycle designed to control resin viscosity during the critical flow window.

1. Vacuum pull down occurs in the cold press chamber, reducing ambient pressure below 10 Torr for a minimum of 30 minutes to extract absorbed moisture and trapped air from inner-layer sub-assemblies.
2. Thermal ramp begins at a controlled rate between 1.5 and 2.5 degrees Celsius per minute, elevating the stack temperature toward the resin gelation range while holding a low contact pressure around 50 pounds per square inch.
3. Full hydraulic pressure between 250 and 350 pounds per square inch applies precisely when the resin reaches its minimum melt viscosity, forcing liquid resin into copper details without washing away fine-line conductors.
4. High-temperature isothermal dwell holds the stack above 200 degrees Celsius for 90 minutes, driving complete polymer cross-linking and establishing structural dimensional stability across all sub-assemblies.
5. Controlled cooling lowers the stack temperature to 60 degrees Celsius at a rate strictly under 2.0 degrees Celsius per minute under full pressure to prevent internal thermal stress accumulation and micro-delamination.

Resin Hydraulic Squeeze and Microscopic Compaction

During secondary lamination, the total thickness of newly added prepreg layers decreases as resin squeezes laterally into inner-layer clearance areas. Vertical compaction depends directly on the initial resin content of the prepreg and the volume of copper to be filled on adjacent inner layers. Heavier copper weights, such as 2-ounce power planes, demand a larger volume of resin fill, which severely depletes resin from the dielectric layer directly above the copper features.

Phase skew destroys signal margin.

As the dielectric layer thins, glass bundles are compressed vertically against rigid copper planes. Vertical bundle compression alters individual glass filaments from circular geometries toward flattened ellipses. Flattened glass bundles increase the out-of-plane glass fill factor per unit volume, driving the Z-axis dielectric constant upward toward that of solid glass.

A core material with a nominal raw Z-axis dielectric constant of 3.30 can exhibit an effective Z-axis dielectric constant of 3.52 after two sequential press cycles when embedded between dense copper signal planes. The table below presents measured directional dielectric constants across sequential lamination stages under varying press pressure profiles.

Multi-Stage Cure Cycles and Dielectric Drift

Each time a core sub-assembly undergoes an additional thermal press cycle, the thermosetting resin matrix post-cures. Post-curing increases the cross-linking density of the polymer chains, altering the polarizability of the dielectric material. While higher cross-linking generally improves thermal stability and raises the glass transition temperature, it subtly shifts both the dissipation factor and dielectric constant.

A dielectric constant of 3.45 measured at 10 GHz via split cylinder resonator increases by 0.18 along the out-of-plane axis when resin fill factors drop below 42 percent during sub-lamination press cycles.

In a batch of 16-layer triple-sequential build panels exhibiting inconsistent propagation delays between identical-length traces routed on Layer 3 and Layer 14, micro-section analysis revealed that Layer 3 had undergone three full lamination heat cycles, reducing its local resin content to 43 percent through repeated hydraulic flow, whereas Layer 14 had undergone only one lamination pass, retaining 54 percent resin content. The resulting 0.16 permittivity difference between the layers caused a 4.2 picosecond per inch timing skew, forcing a 24,000 dollar scrapping charge for the non-compliant backplane lot.

Weave

The structural geometry of the reinforcing fabric is the primary source of spatial and planar permittivity variation in high-speed substrates. Standard glass fabrics consist of yarns woven in a plain-weave grid, where warp and fill threads cross over and under each other at regular intervals. This crossing creates alternating micro-regions dominated by dense glass bundles and open resin windows.

A signal trace running parallel to the weave vector periodically passes over dense glass knuckles and open resin gaps. At millimeter-wave frequencies, the signal wavelength approaches the spatial period of the glass weave pitch, inducing periodic resonance, high insertion loss notch filtering, and severe differential phase skew.

Every press stroke alters local density.

Differential pairs routed over loose glass weaves suffer when one conductor aligns over a glass yarn while its complement aligns over an adjacent resin window. The conductor over glass encounters higher permittivity and propagates slower, while the conductor over resin propagates faster. This phase imbalance converts differential mode signal energy into common mode noise, degrading eye openings and triggering receiver jitter failures in 112G PAM4 channels.

Sequential lamination exacerbates weave-induced phase skew because sub-lamination press cycles force unspread glass bundles to deform irregularly around inner-layer copper features, destroying the mechanical symmetry of the fabric pattern.

Glass Filament Architecture and Fill Ratios

Mitigating weave-induced anisotropy requires switching from standard open glass fabrics to mechanically spread glass styles. Spread glass uses specialized manufacturing techniques, such as high-pressure water jets or acoustic vibration during weaving, to open up yarn bundles and flatten individual filaments into thin, uniform ribbons. Flattened ribbons eliminate open resin windows, creating a continuous, homogeneous sheet of glass across the planar axes.

Unspread glass degrades channel uniformity.

Evaluating glass styles for sequential lamination demands matching yarn count, fabric thickness, and filament spread capability against target dielectric thickness. Thin prepreg styles like 1035 and 1078 offer excellent mechanical spreading and thin dielectric layers suitable for fine-pitch microvia sequential outer layers. Heavy glass styles like 2116 or 7628 feature thick yarn bundles that resist spreading, leaving large resin gaps that shift position unpredictably during sequential press passes.

Fabric Styles in Sequential Stackups

Selecting the right combination of glass styles for a multi-stage sequential stackup involves balancing mechanical stability and electrical uniformity. Fabricators must select materials that balance dimensional movement during sub-assembly etching with low planar permittivity variance.

• Style 1035 Spread Glass utilizes ultra-fine filaments spread into a uniform planar sheet, minimizing local Dk variation for high-density microvia layers operating above 28 GHz.
• Style 1078 Spread Glass provides a higher mechanical stability and resin fill capacity, serving as an ideal prepreg choice for intermediate sequential bonding layers over medium copper coverage.
• Style 2116 Unspread Glass features prominent yarn knuckles that induce significant localized Dk fluctuations, making it unsuitable for differential trace routing in ultra-high-speed channels.
• Style 3313 Mechanically Flattened Glass balances thick dielectric separation with flattened bundle geometry, offering cost-effective planar uniformity for inner-core sub-assemblies.

Pitch Uniformity and Fiber Bundle Compression

When selecting spread glass fabrics for sequential builds, engineers must verify that mechanical spreading remains intact after repeated press operations. Under high hydraulic press pressure, weakly spread yarns can bundle back together, re-creating resin windows between adjacent threads. This effect is especially pronounced when low-viscosity resin flows laterally over inner-layer trace geometry.

Aligning critical high-speed trace channels exclusively with the warp fiber direction minimizes inter-channel phase jitter across sequential lamination cycles.

To preserve glass spreading during multi-stage processing, board designers must enforce strict glass style selections on assembly drawings. Specifying generic slash sheet grades without controlling the underlying fabric style allows fabricators to substitute cheap unspread 1080 or 2116 glass prepregs into high-speed layers. What residual anisotropic skew remains uncorrected across successive sequential press steps when a fabricator substitutes 1080 unspread glass for 1078 spread glass in a 112G PAM4 architecture?

Probe

Accurate extraction of directional dielectric properties requires specialized test fixtures capable of isolating in-plane permittivity components from out-of-plane values. Standard industry datasheets publish dielectric constant figures derived primarily from IPC-TM-650 Test Method 2.5.5.5, the clamped stripline resonator method. Clamped stripline testing measures dielectric constant perpendicular to the laminate sheet, isolating Z-axis permittivity exclusively.

While this figure works fine for low-frequency capacitive calculations, it completely misses in-plane X-Y permittivity variation and understates the total dielectric loading experienced by microstrip lines and high-frequency edge-coupled differential pairs.

Standard coupons hide directional variance.

Engineers specifying materials for sequential lamination must demand directional dielectric characterization conducted across the actual operating frequency spectrum. Relying on single-frequency 10 GHz Z-axis datasheet numbers leads to severe timing errors when designing 28 GHz and 53 GHz transmission lines. Fixtures such as split cylinder resonators, split post dielectric resonators, and Fabry-Perot open cavities isolate in-plane tensor components, revealing the full extent of anisotropy introduced by sequential thermal cycles.

What Fixture Isolates in Plane Permittivity at Millimeter Waves?

The split cylinder resonator, standardized under IPC-TM-650 Test Method 2.5.5.13, provides the most precise non-destructive measurement of in-plane dielectric constant and loss tangent for thin laminate samples. The fixture consists of a cylindrical cavity split horizontally into two halves. A flat substrate specimen inserts between the two cavity sections, acting as an equatorial septum.

Microwave energy excited within the cavity operates in the TE011 transverse electric mode, establishing circular electric fields oriented entirely parallel to the plane of the test sample.

Fixtures reveal true anisotropic behavior.

Because the electric field lines lie entirely within the X-Y plane of the substrate, the split cylinder resonator measures in-plane permittivity without interference from Z-axis properties. Operating frequencies depend on cavity diameter, typically ranging from 10 GHz to 80 GHz. Comparing split cylinder in-plane measurement data against clamped stripline Z-axis data isolates the exact anisotropic ratio of the substrate.

The table below details common permittivity test fixtures, their operational orientation, and their applicability for evaluating sequential lamination materials.

Measurement Fixture Geometries and Axis Alignment

Extracting orthogonal tensor components along specific fiber directions requires precise mechanical alignment of the test sample within the measurement cavity. When using split cylinder or split post dielectric resonators, cutting test specimens aligned strictly parallel to the warp threads isolates X-axis permittivity. Rotating the specimen 90 degrees to align fill threads parallel to the cavity excitation field isolates Y-axis permittivity.

Measuring samples extracted at a 45-degree angle isolates the shear dielectric response of the composite matrix.

Differential pairs demand symmetric dielectric constant.

Test results demonstrate that planar anisotropy between warp and fill axes typically ranges from 0.03 to 0.09 in standard E-glass core materials. After three sequential press cycles, this planar spread widens to 0.14 due to asymmetric resin expulsion along fill yarn channels. Evaluating these parameters before committing to high-volume tooling prevents costly design spins caused by unexpected phase delays on orthogonal bus channels.

Coupon Extraction and out of Plane Characterization

While cavity resonators measure unclad laminate samples before board fabrication, circuit designers require verification of anisotropic permittivity on fully processed, multi-stage sequential panels. Processed panel verification relies on dedicated high-frequency test coupons embedded in panel waste margins.

Compliance with IPC-4101 slash sheet 102 specifications fails to protect differential skew when planar dielectric anisotropy exceeds 0.08 between orthogonal trace directions.

High-speed TDR phase delay coupons feature identical-length transmission line pairs routed along warp, fill, and 45-degree vectors across core layers and sequential outer layers. Connectors or high-bandwidth probe stations feed sub-picosecond pulses into the coupon, measuring time-domain transmission parameters. The calculated phase velocity directly yields the effective dielectric constant experienced by production traces.

Per IPC-6012 Class 3 specification guidelines, impedance and phase delay verification coupon results shall be archived in the production lot dossier to validate that multi-pass sequential press processing remained within designed dielectric tolerance windows.

Scrap

Anisotropic permittivity variation translates directly into lower yields and higher landed board costs when manufacturing sequential lamination substrates. High-speed network switches, server backplanes, and millimeter-wave radar modules enforce tight phase skew budgets across parallel differential channels. When planar dielectric variation creates unaccounted phase jitter, finished circuit boards fail automated end-of-line functional vector testing, resulting in immediate scrap at the bare-board or PCBA level.

Panel orientation influences phase delay.

Scrapping a multi-stage sequential lamination assembly at final test carries a heavy financial penalty. Sequential builds incorporate high layer counts, multiple drilling and plating passes, microvia laser ablation, and expensive low-loss laminate materials. Scrapping a bare panel after three sequential lamination passes discards not only raw laminate substrate stock, but also multiple rounds of inner-layer imaging, etching, inspection, and press labor.

Minimizing scrap requires implementing panel layout compensation techniques and selecting laminate grades that bound anisotropic spread within acceptable manufacturing windows.

Yield Loss Driven by Dielectric Phase Jitter

In 112G PAM4 optical transceiver modules and backplane channels, total unit phase skew budgets often drop below 1.5 picoseconds over a 6-inch routing length. A planar permittivity delta of 0.08 between adjacent traces in a differential pair introduces over 2.8 picoseconds of phase skew over that same distance, completely consuming the system timing margin and degrading the bit error rate. Standard fabricator test coupons measure only nominal impedance, allowing phase-skew-defective panels to pass bare-board test lines unnoticed until functional test at the contract manufacturer.

Prepreg resin content dictates local permittivity.

Implementing strict dielectric anisotropy specifications on fabrication drawings prevents non-compliant panels from reaching assembly lines. However, holding tight anisotropic limits forces fabricators to tighten process controls, lower press line throughput, and increase incoming raw laminate testing. Fabricators offset these operational costs by raising bare panel unit prices, creating a clear trade-off between bare-board unit cost and final assembly yields.

Panel Layout Strategies for Anisotropic Compensation

Circuit designers can suppress weave and lamination anisotropy by implementing structural layout compensation strategies on bare panels. Rotating high-speed trace channels 10 to 11 degrees relative to the substrate warp and fill axes forces signal conductors to cross both warp yarns and fill yarns continuously, averaging out local glass-to-resin ratio differences.

Angle routing eliminates discrete weave resonance, but introduces severe panel utilization penalties. Rotating entire circuit arrays 11 degrees on a standard 18 by 24-inch panel reduces the total number of usable circuits per panel by 15 to 30 percent, instantly raising the individual board unit price. The table below outlines the financial and operational trade-offs of panel layout compensation strategies across sequential lamination stackups.

Panel Utilization and Array Rotation Economics

Selecting the most economical method to control anisotropic permittivity variation requires calculating total landed cost per working board. Upgrading laminate materials from standard E-glass to spread NE-glass or quartz fabric increases raw substrate material expenses by 35 to 80 dollars per panel. However, spread glass upgrades preserve 100 percent of panel area efficiency, avoiding the 22 percent area waste introduced by array rotation.

Procurement documents for high-speed sequential boards must establish unambiguous rules for material grading, fabric selection, and directional tolerance verification.

• Laminate Slash Sheet Addenda specify mandatory spread glass styles and restrict prepreg substitutions without prior engineering approval.
• Directional Permittivity Limits bound maximum allowable X-Y planar and Z-axis dielectric constant deltas across all sequential sub-stage assemblies.
• Coupon Test Protocols define mandatory TDR phase skew testing on panel-margin coupons extracted from every production lot.
• Press Line Audit Requirements establish clear parameter windows for vacuum levels, pressure ramp rates, and thermal dwell profiles.

High-volume procurement contracts yield the lowest landed cost when material upgrades apply selectively to critical sequential layers while maintaining standard low-loss stock on non-critical inner core layers.

Audit

Managing anisotropic permittivity variation in high-speed sequential lamination substrates ultimately requires enforcing systematic purchasing controls and engineering oversight across the supplier base. Procurement managers cannot rely on standard supplier data sheets or generic IPC slash sheet specifications when sourcing materials for sub-50 picosecond channel architectures. Standard industry specifications allow material vendors to adjust resin content, glass yarn styles, and filler ratios within broad bands without changing part numbers or notifying end users.

Oven bake cycles alter glass transition.

A bare-board fabricator purchasing approved material under a generic slash sheet may receive spread glass stock in one shipment and unspread glass core stock in the next. While both lots satisfy basic UL flammability and standard IPC thermal parameters, the unspread stock introduces severe planar anisotropy that destroys signal phase alignment on high-speed sequential layers. Protecting product quality mandates locking down specific laminate and prepreg part numbers on the engineering drawing, supplemented by explicit process restrictions in the fabrication master contract.

Supplier Qualification and Slash Sheet Validation

Qualifying a fabricator for multi-stage sequential lamination substrates demands evaluating their facility-level capabilities in press control, microsection analysis, and high-frequency electrical verification. Sourcing teams must audit candidate factories to ensure their press lines utilize computer-controlled hydraulic systems capable of maintaining precise pressure ramp profiles during resin gelation phases.

Factories relying on manual hydraulic press controls or uncalibrated heating platens generate severe batch-to-batch pressure variations. Uncontrolled pressure variations drive unpredictable resin squeeze out, causing Z-axis and planar permittivity to drift across successive manufacturing lots. Qualified fabricators must maintain lot-level traceability of raw laminate press cycles, archiving vacuum levels, platens thermal uniformity logs, and dynamic pressure profiles in the master quality record.

Fabrication Drawing Notes for Anisotropic Control

The fabrication drawing serves as the primary legal and technical instruction set governing bare-board manufacturing. To control anisotropic dielectric behavior on sequential builds, the drawing notes must explicitly override generic IPC tolerances and establish strict structural boundaries.

Engineering drawings for high-speed sequential substrates must include notes restricting laminate material families to specific manufacturer trade names and exact resin/glass style codes. Drawing notes shall specify that all prepreg layers on designated signal layers utilize mechanically spread glass styles with a spread factor exceeding 85 percent. Furthermore, drawing notes must state that maximum allowable directional phase skew measured on embedded panel coupons shall not exceed 0.5 picoseconds per inch between warp and fill routing vectors.

Purchasing Controls for Multi Stage Lamination Orders

Closing the quality loop requires aligning procurement terms with technical drawing requirements. Master purchase agreements for sequential lamination boards should incorporate financial remedies for non-compliant lots, holding suppliers accountable for phase skew and anisotropic impedance failures discovered during lot acceptance testing.

Receiving inspection protocols at the assembly facility must include automated verification of coupon dossier data supplied with every panel shipment. Purchase order releases should require the fabricator to attach certified TDR phase skew and split cylinder dielectric test reports for every lamination press batch represented in the delivery lot. Implementing these controls ensures that high-speed sequential lamination substrates consistently meet timing budgets, protecting final product yields and containing total landed manufacturing costs.

Fab shop QA managers frequently push back against custom anisotropic notes, arguing that standard IPC-4101 slash sheet compliance guarantees adequate electrical performance. Auditing three offshore fabrication facilities following recurring 112G PAM4 receiver jitter on production boards revealed that two facilities had substituted low-cost unspread 1080 prepreg into inner sequential bonding layers, claiming the substitution met the assigned IPC slash sheet. Enforcing explicit drawing notes that restrict prepreg selection to qualified spread-glass part numbers eliminated the jitter defects across subsequent production runs.