Fiber Weave Effect
What Is the Fiber Weave Effect?
The fiber weave effect is the variation in signal delay and impedance caused by the woven glass inside a PCB laminate. Glass bundles have a higher dielectric constant (Dk) than the resin between them, so the effective Dk depends on how much glass is under the trace. On a differential pair, a Dk difference between the two traces creates intra-pair skew. The skew grows along any stretch where the imbalance holds and closes the eye at the receiver, even on a pair you length-matched exactly.
How Glass and Resin Create Skew
The laminate in your stackup is woven fiberglass cloth set in resin, in both FR-4 cores and prepreg. Signal velocity depends on the Dk of the material the field travels through:
\[v = \frac{c}{\sqrt{D_k}}\]
where c is the speed of light.
Part 1 of Quilter's fiber weave series uses illustrative values of Dk 6 for glass and Dk 3 for resin. With those values, a signal in pure glass travels about 29% slower than in pure resin. A real trace's field passes through both materials, so its effective Dk is between the two, set by the ratio of glass to resin under and around it.
A differential pair needs both signals to arrive at the receiver together. When one trace has more glass under it than its partner, that trace is slower, and the pair develops skew. Length matching equalizes copper length and leaves the material under each trace unchanged, so two equal-length traces can still arrive at different times.
How Skew Accumulates Along a Trace
Skew grows roughly linearly with length only along a section where one trace has more glass under it than the other. When the traces cross the weave at a different offset, the imbalance changes and can partly cancel. A trace running parallel to a glass bundle for its full length is the worst case and sets the upper bound on skew.
Why Fiber Weave Skew Varies From Board to Board
Neither you nor your fab house controls a trace's position relative to the glass. Registration tolerance puts a small, random offset between the copper artwork and the weave, and the glass bundles themselves bunch, stretch, and twist. Two boards built from the same Gerber files on the same laminate can measure different skew. Across 46 test boards, Bogatin et al. found that 79% had more than 0.5° of rotation between the traces and the weave, and that a rotation as small as 0.3° was enough to average out the skew on their 4-inch test lines. Treat fiber weave skew as a distribution with a tolerance, the same way you treat impedance tolerance on a controlled-impedance stackup. Part 2 of the series covers two ways to model it:
- Analytical bounds: Lambert Simonovich's method takes the Dk of the resin-rich 106 glass style and the glass-rich 7628 style from the same laminate family as the lowest and highest Dk to expect. In his worked example, the bounds are Dk 3.30 and Dk 4.02. Assign one value to each leg of the pair and you get a worst-case skew for simulation.
- Statistical prediction: Manukovsky, Shlepnev, and Mordooch (2023) use 3D electromagnetic simulation to compute skew for a differential pair at offsets across one period of the weave, then calculate the probability that skew exceeds a limit, which they call differential skew exceedance (DSE). For their geometry (4 mil traces with 4 mil spacing over a 4 mil laminate) at a 3 ps/in limit, the exceedance probability was 77% for 3313 glass, 67% for 1080, 47% for 1078, and 0% for 1035. The authors note that wider traces show less variation, so your stackup and trace geometry will give different numbers.
Effects on Signal Quality
Where It Shows Up | What Happens |
Eye Diagram | Skew subtracts almost directly from the horizontal eye opening. In the Part 1 model, 0.4 unit interval (UI) of skew shrinks a 1 UI eye to about 0.6 UI. |
Differential Insertion Loss (SDD21) | Transmission collapses where skew equals half the signal period. The first null falls at \(f = 1/(2 \times \text{skew})\), so 100 ps of skew puts it at 5 GHz, with more nulls at odd multiples. |
Impedance | Local Dk also sets impedance, so impedance varies along the trace and between traces. |
Fiber weave becomes a design constraint when accumulated skew uses a meaningful share of the unit interval, which depends on your data rate and trace length. Long, fast channels such as backplanes, PCI Express (PCIe) links, and other serializer/deserializer (SerDes) links are the first to hit it. See High-Speed PCB Design and Signal Integrity.
How to Reduce the Fiber Weave Effect
Most fixes are material and fabrication choices, so settle them with your fabricator when you agree on the stackup, before routing starts:
- Glass style: In the Manukovsky, Shlepnev, and Mordooch simulations above, exceedance probability fell from 77% for 3313 glass to 0% for 1035. In measurements by Bogatin et al. (2017), single-ply constructions ranged from 2.67 ps/in peak to peak for spread 2116 glass to 7.6 ps/in for 1067.
- Spread glass and ply count: Bogatin et al. had one direct comparison of spread and non-spread glass, single-ply 1078, and found almost no difference: 7.0 ps/in spread and 6.8 ps/in non-spread. The authors say one comparison is too little to generalize from. Ply count made a larger difference. Two plies of 1067 measured 1.23 ps/in, against 6.5 and 7.6 ps/in for single-ply 1067, and the lowest result in the study, 0.81 ps/in, came from two plies of mechanically spread 1078 L-glass. Rogers Corporation, a laminate manufacturer, reports its own measurements showing lower skew with spread glass.
- Angled routing: Routing at an angle to the weave averages the glass and resin under each trace. Jeff Loyer, an author of Intel's 2007 fiber weave study, summarized its results: rotating the design about 13° cut the maximum skew on a 10-inch trace from 51 ps to 6 ps, and a 45° rotation cut it to 11 ps. He suggests 7° relative to the board edge, which covers about 2° for 1080 glass plus up to 5° of misalignment between the glass and the board edge. Bogatin et al. note that traces at 12° to the weave have shown resonances near 35 GHz, which affects insertion loss.
- Panel rotation: Your fab can rotate the artwork on the panel instead, which keeps your layout orthogonal and costs panel space. In Rogers' example, two 12 × 18 in. boards fit on an 18 × 24 in. panel, and a 12° rotation leaves room for only one, doubling the material cost per board.
- Skew budget and test coupons: Model the skew range for your chosen laminate, set length budgets from it, and have your fab measure impedance and skew on test coupons.
Frequently Asked Questions
What Is the Fiber Weave Effect in a PCB?
It is the variation in signal delay caused by the glass and resin pattern inside a PCB laminate. Glass has a higher Dk than resin, so a trace with more glass under it carries its signal more slowly. On a differential pair, that difference shows up as intra-pair skew.
Is the Fiber Weave Effect the Same as the Glass Weave Effect?
Yes. Fiber weave effect, glass weave effect, and glass weave skew all refer to the same thing.
Does Length Matching Fix Fiber Weave Skew?
No. Length matching makes the copper lengths equal, and the two traces can still have different amounts of glass and resin under them. The fixes are laminate choice, routing angle or panel rotation, and a skew budget.
Does Fiber Weave Skew Always Increase Linearly With Length?
Only along stretches where one trace has more glass under it than the other. Where the traces cross the weave at changing offsets, the imbalance shifts and can partly cancel. The fully aligned case is the upper bound.
Can Fiber Weave Skew Be Predicted Exactly?
No. Each trace's position relative to the weave is random, so skew is a distribution. You can bound it analytically, or calculate the probability of exceeding a skew limit and use that as a yield estimate.
Should I Talk to My Fabricator About Fiber Weave?
Yes, if your channel is long and fast enough for skew to use a meaningful share of the unit interval. Glass style, spread glass, ply count, and panel rotation are all fabrication choices, so settle them with your fab house when you agree on the stackup, and ask about test coupons for measuring impedance and skew.
Related Terms
Prepreg · Differential Pairs · Length Matching · Dielectric Constant · FR-4 · Signal Integrity · High-Speed PCB Design
Relevant Resources
- Understanding Fiber Weave Effect: From Glass Weave to Differential Skew (Part 1)
- Fiber Weave Skew Modeling: Analytical Bounds and Statistical Skew Prediction (Part 2)
Sources
Bogatin, Eric, Bill Hargin, Venkata Satya Sai Teja Paladugu, Don DeGroot, Amendra Koul, Seungyong Baek, and Mike Sapozhnikov. "New Characterization Technique for Glass-Weave Skew." Signal Integrity Journal, March 2, 2017. Presented at DesignCon 2017, Santa Clara, CA. https://www.signalintegrityjournal.com/ext/resources/article-images-2017/4103/4103_SIJ_GlassWeave.pdf.
Loyer, Jeff. "Re: Resedn:Mitigating PCB Fiber Weave Effect." SI-LIST mailing list, October 26, 2012. https://www.freelists.org/post/si-list/ResednMitigating-PCB-fiber-weave-effect,6.
Manukovsky, Alex, Yuriy Shlepnev, and Shimon Mordooch. "Quantification of Delay and Skew Uncertainty due to Fiber Weave Effect in PCB Interconnects." Paper presented at the 2023 IEEE 32nd Conference on Electrical Performance of Electronic Packaging and Systems (EPEPS), Milpitas, CA, October 15 to 18, 2023. https://www.simberian.com/AppNotes/FWE_Quantification_EPEPS_2023_final.pdf.
Nurmi, Bob, John Coonrod, and Vitali Judin. "Material-Induced Skew in High-Speed Multilayer PCBs: Influences and Mitigation Strategies." Signal Integrity Journal, January 17, 2026. https://www.signalintegrityjournal.com/articles/4105-material-induced-skew-in-high-speed-multilayer-pcbs-influences-and-mitigation-strategies.
Quilter. "Understanding Fiber Weave Effect: From Glass Weave to Differential Skew (Part 1)." August 27, 2026. https://www.quilter.ai/blog/fiber-weave-effect-part-1.
Quilter. "Fiber Weave Skew Modeling: Analytical Bounds and Statistical Skew Prediction (Part 2)." September 16, 2026. https://www.quilter.ai/blog/fiber-weave-effect-part-2.
Simonovich, Lambert. "Practical Fiber Weave Effect Modeling." White paper, issue 3. LAMSIM Enterprises, March 2, 2012. http://lamsimenterprises.com/Practical_Fiber_Weave_Modeling_Iss3_Mar2_12.pdf.
