How to Measure FR4 PCB Dielectric Constant

Something that comes up often in signal integrity at high frequencies is determining whether a reported dielectric constant value is accurate. This is not a trivial issue; it matters when correlating simulation to lab data, setting impedance targets, and predicting propagation delay in a proposed design. In some cases, it becomes necessary to infer dielectric constant from measurements on a test coupon rather than relying only on a datasheet value.

If you are questioning the dielectric constant value in a datasheet, it helps to understand how PCB material dielectric constants typically vary with frequency. It also helps to understand which measurement methods give a process-specific value and which methods are better for extracting a more fundamental bulk material property.

Typical Variation of FR-4 Dielectric Constant With Frequency

Because FR-4 laminates are built from glass and resin systems, different commercial products tend to show similar trends in dielectric constant as frequency increases. The same general trend also appears in other material systems, including PTFE-based materials, newer thermoset resins, ceramic-reinforced materials, and other specialty glass-reinforced laminates. There is extensive published data showing this frequency dependence, and materials suppliers have documented it for many years.

Materials companies try to engineer laminates so that dielectric constant variation with frequency remains small. Even so, datasheets and application notes still show measurable changes across frequency. Some graphs and dielectric constant tables are listed below.

The first example is reproduced from another article written by Kella Knack, which shows variation in dielectric constant over frequency for several materials.

The next set of data shows measurements provided by John Coonrod for microstrips on Rogers RO3003 materials. The data was captured for two commercially available copper foils available for RO3003 laminates.

From these graphs, we can see how dielectric constant varies over frequency for two different copper roughness values. The same general trend of decreasing dielectric constant at higher frequencies appears again, along with the relatively stable behavior at higher frequencies that is characteristic of this material system. We also see that higher copper roughness, represented by the ED copper, shifts the dielectric constant curve upward. This effect is discussed in another article linked here.

Finally, consider tabular data from a commercial core and prepreg laminate set. Isola does an excellent job of providing dielectric constant data at multiple frequency values. These values are often broken down by material thickness, glass weave style, and core versus prepreg options. The example below shows data for Isola 370HR, a mid-range Dk laminate suitable for many moderate-speed digital designs that require controlled impedance.

Sample of core and prepreg data for Isola 370HR materials [Source: Isola Group, access a copy of this data at this link]

These graphs and tables all show the same basic trend: in FR-4 grade PCB materials, dielectric constant generally decreases somewhat as frequency increases. In practical terms, higher-frequency components of a signal propagate slightly faster than lower-frequency components. That effect, together with insertion loss along the interconnect, contributes to signal distortion and pulse spreading during propagation.

Determination of Dielectric Constant From Measurements

There are multiple methods for measuring the dielectric constant of a PCB material. In practice, PCB designers do not perform these measurements very often unless they need to verify values for signal integrity work. Materials manufacturers are much more specialized in this area and typically have access to equipment and fixtures that most designers do not. Even so, there are several methods that designers can use, and these are also the same general methods used by materials suppliers to characterize laminates.

Before looking at specific methods, there are a few important points to keep in mind:

• There is no direct measurement of dielectric constant. It is always calculated from some other measured quantity.
• Some measurement methods are specified under IPC standards, but designers can also use simpler methods for their own analysis.
• The goal is usually to determine dielectric constant over a range of frequencies, not just at one frequency.
• The measured value depends on the structure used for the measurement.
• Rough conductor surfaces and rough dielectric interfaces can increase the extracted dielectric constant relative to an ideal smooth structure.

With that in mind, let's look at the main dielectric constant measurement methods.

Reflectometry Measurements

The simplest technique for measuring propagation delay and determining dielectric constant is reflectometry. This is typically done with a time-domain reflectometer, which measures the round-trip delay through an interconnect. Half of the round-trip time gives the one-way propagation delay, which then gives the propagation velocity in that structure. This method is most naturally applied to microstrip structures, where the measured result is really an effective dielectric constant that can then be related back to the substrate Dk.

TDR measurements give the round-trip delay between the input port and the load. 

To perform the measurement, the interconnect under test is terminated with a very large impedance. The reflected time-domain waveform can be interpreted as an average result over a broad frequency range, so the extracted dielectric constant is effectively a broadband average determined by the rise time of the TDR stimulus. This makes the method useful for timing and delay extraction, but it is not the right method when frequency-dependent dielectric behavior is needed.

Resonator Methods

Resonator methods determine dielectric constant by fabricating a structure with known geometry, measuring its resonant frequencies, and then solving for dielectric constant from those resonance points. Because each resonance occurs at a different frequency, these methods can be used to build a frequency-dependent dielectric constant curve rather than a single average value. This is the main advantage of resonator methods: they provide a relatively direct way to map Dk versus frequency using printed test structures that are practical to build on laminate coupons.

• Ring resonator: Uses a closed-loop transmission line that resonates at discrete frequencies determined by its electrical length, allowing Dk to be extracted from the resonance locations.
• Stripline resonator: Uses a stripline-based resonant structure to extract Dk in a more enclosed field environment, which is often useful when trying to reduce air-field effects seen in microstrip structures.

Ring resonators and stripline resonators still suffer from the same limitations as other printed transmission line structures. The extracted dielectric constant is influenced by copper roughness, conductor geometry, and dielectric anisotropy, so the result is not purely a bulk material property. Instead, it is the value seen by that specific resonant structure built with that specific process.

S-Parameter Measurements

Another common approach is to measure S-parameters of a transmission line over a broad frequency range and extract propagation delay from the phase behavior of the measured response. In practice, this is done on well-controlled microstrip or stripline coupons and is especially useful when broadband dielectric behavior is needed for simulation.

For this method, the transmission line needs to be very closely impedance matched over the intended measurement bandwidth so that reflections do not corrupt the result. Among the measured quantities, insertion loss is often the preferred response to analyze because it provides the most useful broadband information for extracting propagation characteristics and relating them back to dielectric constant.

The phase of the insertion loss curve can be used to determine the propagation delay and dielectric constant, but only after de-embeddiing any test fixture; use a 2x-thru structure for accurate transmission line measurements.

The limitation is that transmission line structures always include process effects, so the extracted dielectric constant is influenced by copper roughness, dielectric anisotropy, and field distribution in the test geometry. Another challenge is that the measurement must isolate the actual dielectric-sensitive section of the structure, which requires de-embedding the launches and fixtures. On PCB coupons, this is commonly done with a 2x-thru structure so the connector and transition effects can be removed before extracting material-related behavior.

Parallel Plate Resonance

This method is one of the few options that can be used to determine bulk dielectric constant without the added influence of etched copper roughness in PCB transmission lines. For that reason, it is often preferred when the goal is to obtain a more process-independent material parameter rather than the effective value seen in a specific interconnect.

Parallel plate resonance measurements involve measuring resonant frequencies over a broad frequency range and using each resonant frequency to calculate dielectric constant. The underlying formula is straightforward and is the same basic relationship used to analyze power-plane resonances in a PDN.

Which Measurement Technique Should You Use?

Assuming the required equipment is available and you can choose any of these methods, the best option depends on what kind of dielectric constant you actually need.

In practice, the real choice is between a process-specific effective dielectric constant and a bulk dielectric constant. Transmission line methods, including printed resonators and S-parameter measurements, are process-specific. They tell you the dielectric constant experienced by a signal in your actual interconnect geometry, including the effects of copper roughness and the fabricated conductor profile.

That is often the most useful number when you are trying to correlate measurements to an existing design or build an accurate channel model for simulation. If, instead, you want a material parameter that is less tied to a specific PCB fabrication process and more useful for material-level comparison, then a bulk measurement is the better option. In that case, the parallel plate resonator method is usually the better choice.

Whether you need to build reliable power electronics or advanced digital systems, use Altium’s complete set of PCB design features and world-class CAD tools. Altium provides the world’s premier electronic product development platform, complete with the industry’s best PCB design tools and cross-disciplinary collaboration features for advanced design teams. Contact an expert at Altium today!