Based on practical project experience, an exploration of high-frequency PCB material selection guidelines from a systems perspective.

I have seen far too many people dive headfirst into technical parameters when selecting high-frequency PCB materials, only to inadvertently overlook critical issues that arise in real-world applications. Material selection is never a simple game of data comparison; rather, it is akin to finding the perfect partner for a specific scenario. Take PTFE, for instance: many people’s eyes light up the moment they hear the term “low loss.” However, if your product is destined to undergo drastic temperature fluctuations, fixating solely on the Df (dissipation factor) value could lead you straight into a pitfall—the coefficient of thermal expansion for this material differs significantly from that of standard FR4.

I recall an instance where I assisted a client in debugging an automotive radar module. The PTFE substrate they had initially selected performed flawlessly in the laboratory environment. Yet, once the project reached the road-testing phase, problems began to emerge—the temperature differential between day and night caused the connector solder joints to crack on three separate occasions. The issue was ultimately resolved only after switching to a modified epoxy resin laminate combined with specialized fillers.

Nowadays, when designing, I place a much greater emphasis on the synergistic effects of materials—for instance, by utilizing a combination of materials with varying dielectric constants across different layers within a multilayer PCB stackup.

The aspect of processing compatibility is also frequently underestimated. I once observed an engineer select a PTFE composite material boasting top-tier performance specifications, only to find that the manufacturing facility was utterly unable to process it consistently; issues arose during drilling (such as fiber delamination) and during lamination (resulting in excessive fluctuations in dielectric thickness).

Ultimately, the selection of high-frequency PCB laminates demands a systematic, holistic mindset.

I typically break down the application scenario into several key dimensions: signal integrity requirements, mechanical strength demands, environmental tolerance, and—perhaps most critically—the collaborative capabilities of the entire supply chain. While recently working with millimeter-wave sensors, I stumbled upon an intriguing phenomenon: certain medium-loss materials—through structural optimization—actually outperformed ultra-low-loss materials. The key lay in achieving a smoother phase response within a specific frequency band.

These experiences taught me that high-frequency design is never about simply selecting the most expensive materials, but rather the most suitable ones. Sometimes, taking a step back to view the overall solution can actually lead to a more optimal outcome. After all, the final product must function reliably in the real world, not merely reign supreme on a datasheet.

Whenever I see discussions regarding material selection for high-frequency PCBs, people tend to focus their attention almost exclusively on the dielectric material—as if dielectric parameters were the be-all and end-all. In reality, I believe the details that truly impact performance are often hidden in inconspicuous places—such as the choice of copper foil. Many people may not realize that for the exact same PCB laminate, simply changing the copper foil treatment method can result in a night-and-day difference in signal performance.

I, too, fell into this trap when I first started working in high-frequency design. On one occasion, while developing a microwave module, I was meticulous in my selection of the dielectric laminate but overlooked the issue of copper foil roughness. Consequently, testing revealed that the insertion loss was consistently—and significantly—higher than anticipated. The issue was ultimately resolved only after switching from standard electrolytic copper to ultra-low-profile copper foil. At high frequencies, current flows only within a thin layer near the conductor’s surface; a rough surface is akin to forcing a signal to run a long-distance race over a gravel road—naturally, it requires far more effort.

Nowadays, when reviewing manufacturer datasheets, I pay particular attention to the descriptions of the surface treatment processes. Some materials are touted as “low-loss,” yet they are paired with copper foil that has not been optimized for roughness at all. This is like buying a sports car only to equip it with worn-out tires—no matter how excellent the vehicle’s fundamental performance may be, it simply cannot be fully realized. This is especially critical in the millimeter-wave band, where the skin depth can be less than one micron; in this regime, the microscopic morphology of the copper foil’s surface directly determines whether the signal glides along smoothly or stumbles along unevenly.

I recall a particularly striking comparative test: using the exact same dielectric material—but swapping standard copper foil for specially treated, smooth copper foil—resulted in a 15% difference in loss at 28 GHz. This disparity is not something that can be compensated for simply by tweaking the circuit topology. Consequently, whenever I encounter a high-frequency design project now, I treat the copper foil parameters with the exact same level of importance as the dielectric parameters.

Ultimately, there is no single, absolute “correct answer” when it comes to material selection; the key lies in carefully considering the specific application scenario. For instance, in certain cost-sensitive applications where signal frequencies are not exceptionally high, using copper foil with moderate surface roughness might prove more cost-effective. Conversely, applications such as millimeter-wave radar demand a significantly larger budget allocation to ensure superior copper foil surface quality. Such trade-offs must be evaluated in the context of specific design objectives, rather than through a blind pursuit of theoretically optimal parameters.

At times, I feel that material selection is akin to blending coffee beans—the most expensive option is not necessarily the best fit for your particular cup. One must clearly understand the specific “road conditions” required by the designed signal to select a combination of materials that truly provides the perfect match.

I always find it rather amusing when I see people overcomplicating the process of high-frequency PCB design. In reality, selecting the right materials need not be a mysterious endeavor; the key lies in understanding the signal’s actual requirements, rather than blindly chasing the numbers listed in a datasheet.

I have encountered numerous engineers who, right from the outset, fixate on the dielectric constant for extended periods while overlooking the demands of the actual application environment. For example, when designing millimeter-wave radar systems, material stability often outweighs initial parameter values; since temperature fluctuations can induce signal drift, one must prioritize substrate materials with a low temperature coefficient.

Impedance matching is another critical area requiring particular attention. Many assume that simply calculating the values ​​based on standard formulas is sufficient; however, batch-to-batch variations in the PCB laminate itself can introduce significant deviations. We once conducted a test revealing that the dielectric constant within a single roll of material could vary by as much as 3% between the edges and the center. This discrepancy directly caused the impedance of the microstrip lines to fluctuate beyond the acceptable design tolerance, ultimately necessitating adjustments to the line width to compensate.

While the market is currently flooded with a dizzying array of high-frequency PCB laminates, the materials that prove truly practical and effective are rarely those boasting the most dazzling specifications on paper. For instance, certain hydrocarbon-based substrates—despite exhibiting slightly higher loss factors—offer excellent processing characteristics and are compatible with standard FR-4 production lines. This compatibility makes them particularly well-suited for small-to-medium batch manufacturing environments. After all, the time and cost associated with re-calibrating an entire production line can often exceed the marginal cost difference between the materials themselves.

Here is a valuable lesson worth sharing: when developing a 24 GHz product, we evaluated materials from three different suppliers. We ultimately selected a model that fell in the mid-range price bracket but offered consistent supply reliability. We discovered that the performance advantages of the “top-tier” materials were negligible in the context of our actual application, whereas their excessively long lead times posed a significant threat to our project schedule.

Ultimately, selecting materials for high-frequency PCBs is much like getting fitted for eyeglasses: a higher prescription is not inherently better; what matters is finding the right fit for your specific usage scenario. Sometimes, opting for materials of excessive quality can actually result in unnecessary costs—particularly when dealing with large-format PCBs, where material uniformity may be a more critical factor than absolute performance metrics. Having worked with high-frequency PCBs for several years, I’ve gradually noticed a rather interesting phenomenon: many people immediately get bogged down in poring over material data sheets, yet they overlook the most fundamental element—one that is right under their noses: the choice of copper foil. I recall making this very mistake myself when I first started working on millimeter-wave projects; I focused exclusively on parameters like dielectric constant and loss tangent, only to find that—once the boards were fabricated—I simply couldn’t get the insertion loss down to acceptable levels.

It wasn’t until an experienced senior engineer suggested I consider the impact of copper foil surface roughness that I had my “aha!” moment. When high-frequency signals propagate along a conductor’s surface, the skin effect comes into play; the shallower the signal penetration depth, the more critical the requirement for surface smoothness becomes.

This is precisely where the value of ultra-low-profile copper foils—such as the HVLP type—becomes evident. However, there is a common misconception here: not every high-frequency application requires blindly opting for the smoothest copper foil available. I once worked on a Ku-band satellite receiver module where, in pursuit of ultimate performance, I initially selected a copper foil with a roughness of 0.5 μm; however, this resulted in delamination issues during the lamination process. I subsequently switched to a medium-profile specification of 1.2 μm, which not only ensured that insertion loss at 28 GHz remained within a reasonable range but also guaranteed a high yield rate for mass production.

The compatibility between the copper foil and the substrate is often far more critical than any single isolated parameter. For instance, during recent tests on RO4500 substrates, we discovered that even when using the exact same model of LoPro copper foil, variations in adhesion strength between different production batches could actually impact the final product’s thermal cycling reliability.

Regarding the categorization of frequency bands, I hold a somewhat unconventional view. Many people are accustomed to rigidly classifying material types based solely on frequency ranges; however, I’ve found that in practical applications, it is often more important to consider the modulation complexity of the signal. For example, a 6 GHz Wi-Fi 6E signal—due to the high demands for phase coherence imposed by its OFDM modulation scheme—may actually require lower-loss materials than certain simpler modulated signals operating at 10 GHz. We frequently encounter such scenarios in our clients’ real-world projects.

Recently, while assisting a client specializing in automotive radar with material selection, we deliberately stepped outside the traditional framework of frequency-band classification. Given that their 77 GHz radar modules were structurally required to withstand severe vibration, we actually relaxed the surface roughness requirements for the copper foil, thereby ensuring sufficient peel strength to meet the structural integrity demands. Ultimately, selecting high-frequency PCB materials is never merely a simple comparison of parameters; it is more akin to a process of tailoring a solution for a specific application scenario. Every project possesses a unique equilibrium point that must be carefully sought out—and therein, perhaps, lies the true joy of an engineer’s work.

The task of selecting high-frequency board materials cannot truly be decided by simply glancing at a few numbers on a datasheet. I have seen far too many engineers get bogged down in the minutiae of parameters like Dk and Df, only to overlook the inherent complexities of the actual application environment. A material’s performance can vary drastically across different frequency ranges.

I recall working on a millimeter-wave project where the data measured in our lab failed to match the specifications provided by the supplier. We later discovered that the humidity in the testing environment was skewing the results. This experience made me realize that a material’s so-called “stable” Dk value is, in reality, dynamic; it must be evaluated in the context of specific operating conditions.

The impact of temperature fluctuations on material characteristics is often underestimated. Parameters measured in the summer can differ significantly from those measured in the winter. On one occasion, while working on a board for an automotive radar system, a client complained of unstable performance; we eventually traced the issue back to an incorrect choice of material based on its coefficient of thermal expansion. In high-frequency applications, even such subtle temperature variations can amplify technical issues.

Nowadays, many manufacturers like to boast about how incredibly low their materials’ Df values ​​are; however, in practical applications, an excessively low dissipation factor can sometimes introduce other complications. For instance, certain ultra-low-loss materials are notoriously difficult to process, placing extremely stringent demands on the PCB fabrication facility’s manufacturing capabilities. Boards produced by standard factories using such exotic materials may, in fact, turn out to be less reliable than those made using more moderately lossy materials.

I have come to feel increasingly that selecting a material is much like choosing a romantic partner: one cannot simply judge based on superficial attributes, but must instead assess true compatibility. High-speed digital circuits and high-frequency RF circuits have vastly different material requirements. The former prioritizes impedance matching, while the latter places greater emphasis on phase coherence. Attempting to select materials for both using a single, uniform set of criteria is almost guaranteed to lead to problems.

Sometimes, it pays to take a step back and ask: why must we relentlessly chase after extreme, boundary-pushing parameters? Designing with reasonable margins is often a far more practical approach than obsessing over minute parameter values. After all, the end product is intended for mass production—

not to be a museum piece confined to a laboratory.

Fundamentally, material selection is an exercise in balancing performance, cost, and manufacturability. I have witnessed too many projects “crash and burn” due to an excessive fixation on optimizing a single specific parameter; conversely, it is often the seemingly ordinary, pragmatic choices that yield the best overall results.

Ultimately, when it comes to high-frequency board materials, there is no such thing as “absolutely good” or “absolutely bad”—there is only “suitable” or “unsuitable.”

Rather than agonizing over minute differences in the third decimal place, one’s energy is far better spent focusing on system-level optimization. After all, what the end user ultimately cares about is the product’s overall performance, not the value of a single, isolated parameter.