Dielectric Constant of PCB Materials: Measurement and Use

Dielectric constant is the property that links a laminate datasheet to the width of a trace. It describes how much the material slows an electric field and how much capacitance a given geometry produces, and it is the input that most impedance calculations are most sensitive to. Getting the value wrong, or using the wrong version of it, shifts every controlled impedance net on the board.

What the Number Means

The dielectric constant, also called relative permittivity, is the ratio of a material’s permittivity to that of free space. Vacuum is 1 by definition, air is very close to 1, and most PCB materials fall between about 2 and 5.

A standard FR-4 sits near 4.2 to 4.6 when measured at 1 MHz, a mid-loss laminate between 3.6 and 4.0, and a PTFE based material near 2.2. Ceramic filled materials can be tuned across a wide range, which is why they are used where a specific value is needed.

Effective Dielectric Constant

The value that matters to a trace is not the material figure but the effective dielectric constant, which accounts for the fact that part of the field travels in air. A microstrip on the outer layer has copper below and air above, so its effective dielectric constant is lower than the laminate value, often by 15 to 25 percent.

A stripline buried between two planes has its field entirely inside the dielectric, so its effective value is close to the material figure. That is why the same 50 Ω trace is narrower as a stripline than as a microstrip on the same stack, and why the two must be calculated separately. The field structures are described in the overview of microstrip and stripline routing.

Frequency Dependence

Dielectric constant falls slowly as frequency rises, while loss tangent generally increases. For a digital design running at a few hundred megahertz, a value quoted at 1 GHz is a reasonable approximation; for millimetre wave work the curve has to be read at the actual operating frequency.

Loss tangent, often written Df, is the second number on the datasheet and it governs attenuation. Standard FR-4 sits near 0.02, mid-loss materials near 0.008 and low loss materials between 0.002 and 0.004. At low frequency, conductor loss dominates and the laminate choice matters little; at high frequency, dielectric loss becomes the larger term and the material decides the channel budget. The routing consequences are covered in the notes on high frequency trace and data bus routing.

Moisture and Temperature Effects

Absorbed moisture raises the dielectric constant, because water has a value near 80. A laminate that takes up a fraction of a percent of water by weight can shift its dielectric constant by a few percent, and that is enough to move a controlled impedance trace outside its tolerance. This is one reason boards are baked and packed before assembly, and one reason the impedance coupon is measured on a dry sample.

Temperature has a smaller but still measurable effect, and the coefficient is stated on good datasheets. Where a product operates over a wide temperature range, the coefficient should be included in the tolerance analysis rather than ignored.

How the Value Is Measured

The classical method is a clamped stripline resonator, described in IPC test methods for permittivity and loss tangent. A split post dielectric resonator gives a quick measurement on a coupon, and resonant cavity methods are used for accurate work at microwave frequencies.

For a production board the practical check is an impedance coupon: a test trace with the same geometry as the signal nets, measured with a time domain reflectometer or an impedance analyser. The measured impedance is compared with the calculated value, and the dielectric constant is back-calculated to see whether the material matched its datasheet. This closes the loop on the design and is far more useful than an argument about catalogue numbers.

Using the Value in Design

Choose the material first, then take the dielectric constant from the datasheet at the frequency and test method that match the application. Use the pressed dielectric thickness rather than the nominal figure, because the press reduces thickness and that changes impedance more than a small error in dielectric constant.

Then verify. Fabricate an impedance coupon on the panel, measure it, and adjust the trace widths in the next revision if the result is outside tolerance. Because dielectric thickness tolerance is usually the dominant contributor to impedance variation, a stackup with tight thickness control is often more valuable than a material with a marginally better datasheet value. Stackup arrangements that keep impedance predictable are discussed in the notes on stackup and layout for EMI reduction.

Why the Datasheet Value Is Not the Whole Story

Datasheet figures come from a specific test method, frequency and sample preparation. Two suppliers quoting the same headline number may produce materials that behave differently in a real stack because of resin content, glass style and fillers.

That is why data preparation for a controlled impedance board should include the supplier, the material part number and the target impedance, so the fabricator can select the ply combination that actually achieves it. Treating dielectric constant as a single fixed number is the most common source of a first article that misses its impedance target.

From Dielectric Constant to Trace Width

The calculation chain is straightforward. Given a target impedance, a dielectric thickness and a dielectric constant, the trace width follows from the characteristic impedance formula for the chosen structure. A 50 Ω microstrip on a 0.2 mm dielectric with an effective dielectric constant of 3.6 works out to a trace width of roughly 0.36 mm, and the same trace as a stripline in the same dielectric is noticeably narrower.

Sensitivity is what makes the value important. A ten percent increase in dielectric constant lowers the impedance of a fixed geometry by roughly five percent, which is enough to push a 50 Ω net outside a ±10 percent tolerance once other variations are added. Dielectric thickness has an even stronger effect, which is why the stackup and the material have to be considered together rather than one after the other.

Differential pairs add a second dimension, because the coupling between the two traces depends on the spacing as well as on the dielectric. Changing the material changes both the single ended impedance and the differential impedance, and the pair geometry has to be re-optimised rather than scaled. Working through the geometry with the field structure in mind is the approach described in the guidance on differential trace routing.

FAQ

Which value should be used for a microstrip calculation? Use the effective dielectric constant, not the bulk material figure. The field partly travels in air above the trace, so the effective value is lower and the trace is correspondingly wider.

Does moisture really change impedance? Yes, by a measurable amount. Absorbed water raises the dielectric constant and lowers the impedance, which is why test coupons are measured dry and boards are packed with desiccant.

How often should an impedance coupon be checked? On every new stackup and material lot, and then periodically during production. A change of laminate supplier or a stack revision is enough to justify a fresh measurement.