Dielectric Constant & Loss Tangent: Frequency Selection

A datasheet once made me overconfident.

Bad start.

The material looked fine, the dielectric constant looked “close enough,” and the buyer only cared whether the price could stay inside the old FR-4 budget; then the prototype came back, the RF line didn’t tune cleanly, the lab time got ugly, and suddenly that tiny Df number on page two mattered more than everyone’s polite quotation emails.

Why does this keep happening?

Because dielectric constant and loss tangent are not decoration. They’re not two nice values you paste into a stack-up note so the drawing looks professional. They decide how fast the signal travels, how much energy disappears into the substrate, and whether your “same as before” PCB behaves the same at 500 MHz, 2.4 GHz, 6 GHz, or 10 GHz.

Dk Isn’t a Personality Trait

Dielectric constant, or Dk, tells you how much the PCB material slows an electromagnetic wave compared with air or vacuum. Simple enough. But the number shifts with frequency, resin content, glass weave, test method, humidity, laminate lot, and sometimes the unpleasant little gap between datasheet theory and press-room reality.

That’s the part people skip.

I frankly believe most bad PCB material decisions begin when someone says, “FR-4 is around 4.2.” Around? At what frequency? Measured how? Which laminate? Which glass style? Which resin percentage after lamination?

Not a detail. A trap.

In low-speed control electronics, this usually won’t kill the project. You can build plenty of industrial control PCBs, LED drivers, power boards, and ordinary digital products without turning every laminate decision into a PhD meeting. For early builds, a normal custom PCB prototype manufacturing service may be exactly enough.

But once the design starts caring about impedance, antenna behavior, signal delay, or phase stability, “close enough Dk” becomes dangerous. Quietly dangerous, which is worse.

Loss Tangent: The Small Decimal That Eats Margin

Loss tangent is the ugly one.

Df, dissipation factor, loss tangent — different names, same problem. It tells you how much signal energy gets converted into heat inside the dielectric. At low frequency, maybe nobody notices. At GHz-level work, the PCB starts charging rent for every millimeter of trace.

Here’s the ugly truth: buyers often argue hard over surface finish, lead time, tooling cost, and whether the price can drop another 3%. Fine. That’s business. But then they ignore Df, copper roughness, and dielectric thickness tolerance — the exact things that can wreck insertion loss and force another prototype spin.

It happens a lot.

A “cheap” laminate can become very expensive after two failed samples, three late-night VNA sessions, and one engineer muttering that the old supplier’s PCB somehow worked better. The reason may not be magic. It may just be Dk/Df mismatch, rougher copper, or a stack-up nobody actually controlled.

For RF builds, especially microwave and antenna work, I’d rather discuss material honestly through a Rogers 4003 RF PCB supplier for high-frequency applications than pretend standard FR-4 can do every job because the purchasing file says “same material as usual.”

Frequency Selection Is Not Just the Operating Frequency

People say “the PCB works at 2.4 GHz.”

Maybe.

But what about harmonics? Edge rate? Trace length? Rise time? Return path? Via transitions? Coplanar clearance? Solder mask on or off? That little RF feedline doesn’t care what the project manager wrote in the summary line. It only sees geometry, dielectric, copper, and fields.

That’s why frequency selection gets messy.

For digital boards, the clock frequency may be much lower than the real signal content if the edge is fast. For RF circuits, the carrier may be obvious, but the tuning window, bandwidth, antenna match, and insertion loss budget still matter. For mixed boards — MCU plus wireless plus power plus sensor front end — one “FR-4 multilayer PCB” description can hide three very different electrical risk zones.

And that’s where quoting gets weird.

A buyer sends: 4 layers, 1.6 mm, FR-4, ENIG.

The engineer means: controlled impedance, stable Dk, low loss path, no silly stack-up changes, and please don’t swap the laminate without telling me.

Those aren’t the same request.

The Datasheet Problem Nobody Likes Discussing

But datasheets aren’t lying. Usually.

They’re just selective.

One material may list Dk at 1 MHz. Another at 1 GHz. Another gives multiple test methods, which is better, but now you need to know which number is relevant to your actual structure. A laminate can look stable in the chart and still behave differently in a finished PCB because final Dk depends on resin percentage, glass bundle distribution, copper pattern density, and lamination thickness.

Annoying? Yes.

Real? Also yes.

This is why I’m suspicious when someone says “equivalent material” too quickly. Equivalent for drilling? Equivalent for Tg? Equivalent for CTI? Equivalent for Dk at 10 GHz? Equivalent after production tolerance? These are different questions.

For a standard control product, a halogen-free FR-4 multilayer control PCB can be the smart, clean, cost-effective choice. But if the product has RF, fast differential lines, antenna tuning, radar, or repeatable impedance needs, the material conversation must be tighter.

Not fancy. Tighter.

FR-4 Can Work. Until It Doesn’t.

I don’t hate FR-4.

Actually, FR-4 is underrated when people use it correctly. It’s cheap, available, manufacturable, mechanically familiar, and good enough for a huge range of products. The mistake is not using FR-4. The mistake is using FR-4 while pretending it’s a low-loss RF laminate.

There’s a difference.

At 100 MHz, ordinary FR-4 may behave perfectly well. At 1 GHz, the answer depends. At 2.4 GHz, trace length and loss budget start talking. At 6 GHz and above, Dk spread, Df, copper roughness, and weave effect can become the annoying characters in the room. By 10 GHz, if you’re still waving a generic FR-4 datasheet like a flag, I’m going to ask questions.

Probably too many questions.

Comparison Table: Dk, Df, and Frequency Risk

Frequency Range	Typical Material Direction	Dk Concern	Loss Tangent Concern	Common Risk	Buyer’s Practical Move
Below 100 MHz	Standard FR-4 often acceptable	Usually manageable	Usually low impact	Over-specifying material	Don’t pay for RF material without reason
100 MHz to 1 GHz	Better FR-4 or controlled stack-up	Starts to affect impedance	Depends on trace length	Treating fast edges as low-speed signals	Review impedance and rise time, not only clock rate
1 GHz to 6 GHz	Low-loss FR-4 or RF laminate	Dk frequency stability matters	Df can affect insertion loss	Using datasheet Dk blindly	Ask for Dk/Df test frequency and laminate brand
6 GHz to 20 GHz	Rogers-type or RF-grade laminate	Strong effect on tuning and phase	Serious loss factor	Cheap substitute material	Lock material, stack-up, copper type, and tolerance
Above 20 GHz	Advanced RF/microwave laminate	Very sensitive	Very sensitive	Prototype works, production drifts	Use strict material control and RF-focused review

What I Ask Before Material Selection

Sometimes I ask for the Gerber first. Sometimes I don’t.

If the project smells like RF, high-speed, radar, telecom, automotive sensing, or medical electronics, I want the electrical context before the quote becomes too “clean” to be useful. Layer count alone is weak information. Thickness alone is weak information. Even impedance alone is weak if nobody gives stack-up preference or frequency range.

I usually ask:

What’s the highest meaningful frequency?

Is this RF, high-speed digital, antenna, or mixed signal?

Any controlled impedance? 50Ω, 90Ω, 100Ω?

How long are the sensitive traces?

Is loss budget defined, or are we guessing?

Any specific laminate brand requested?

Will future production use the same material?

Will the customer accept an equivalent material?

Basic questions. Expensive answers.

For medical electronics, I’m more conservative because the cost of unstable performance is not just another PCB spin. It can mean documentation issues, validation delays, and uncomfortable customer conversations. That’s why a medical HDI PCB supplier for high-reliability devices should treat material, stack-up, inspection, and repeatability as one package — not separate checkbox items.

Automotive Makes the Mistake Louder

Automotive electronics don’t forgive sloppy thinking.

Heat. Vibration. Long service life. Humidity. Voltage spikes. Production pressure. And now add radar, sensing, battery systems, communication lines, infotainment, and compact control modules into the same general category people casually call “automotive PCB.”

Too broad.

From my experience, the dangerous sentence is “It’s just a control PCB.” Maybe it is. Maybe it’s a simple relay board. Or maybe it has a frequency-sensitive path, strict thermal exposure, and a customer who expects stable production for years.

A practical automotive electronics control PCB prototype should be reviewed around actual use: temperature range, copper weight, dielectric thickness, impedance, current load, coating, reliability class, and whether Dk/Df actually affect performance.

Don’t guess. Ask.

Flex PCB Has Its Own Tricks

Flex is sneaky.

People think of it as a shape problem — bend radius, outline, stiffener, connector area, gold fingers, adhesive, coverlay. Fair. But if the flex carries high-speed or RF signals, polyimide Dk, adhesive loss, copper type, and impedance geometry enter the conversation fast.

And because flex moves, folds, and fits into cramped products, the routing often gets compromised before anyone admits it.

A gold-plated FPC panel supply option can be right for compact electronics, sensors, wearables, camera modules, and medical devices. But please don’t treat every flex as a dumb cable. Some are transmission lines wearing a soft jacket.

That’s the better way to think about it.

How to Select PCB Material by Frequency

Start with the signal, not the laminate name.

Sounds obvious. Isn’t.

If the design is low speed, standard FR-4 may be enough. If it’s a short 2.4 GHz path, maybe low-loss FR-4 is acceptable. If it’s 6 GHz, 10 GHz, radar, antenna, microwave, or a long controlled channel, then Df, Dk tolerance, copper roughness, and stack-up repeatability deserve real attention.

My rough rule: don’t buy the most expensive material just to feel safe. That’s lazy in the other direction. But don’t force cheap material into a low-loss job and call it “cost control.” That’s also lazy.

Good material selection is a trade-off between:

frequency

loss budget

impedance tolerance

trace length

laminate availability

copper roughness

finished thickness

thermal environment

prototype speed

future production stability

final product risk

No drama. Just physics.

FAQ

What is dielectric constant in PCB material?

Dielectric constant in PCB material is the relative permittivity that shows how much the substrate slows electrical signal propagation compared with air or vacuum, directly affecting impedance, trace geometry, phase delay, antenna tuning, and controlled high-frequency performance in real PCB stack-ups.

In PCB work, it’s usually written as Dk or εr. Lower Dk often means wider traces for the same impedance. Higher Dk can shrink RF structures, but it can also make the design more sensitive to tolerance. The number is useful only when the test frequency and material condition make sense.

What is loss tangent in PCB material?

Loss tangent is the dissipation factor of a PCB dielectric material, showing how much electromagnetic energy is converted into heat as signals travel through the substrate, making it one of the most important values for RF, microwave, and high-speed digital designs.

It’s usually written as Df. Small number, big attitude. At higher frequency, poor Df can show up as insertion loss, weaker signal margin, tuning trouble, or unstable real-world performance. This is why low loss PCB material exists in the first place.

Why does dielectric constant change with frequency?

Dielectric constant changes with frequency because dielectric polarization mechanisms inside the material respond differently as signal frequency rises, which means the same laminate may show different Dk values at 1 MHz, 1 GHz, 10 GHz, or higher test conditions.

That’s why copying one Dk value from a datasheet without checking the test condition can mislead the stack-up calculation. For high-frequency PCB design, the wrong Dk reference may shift impedance, phase, antenna resonance, and signal timing.

How do I select PCB material by frequency?

PCB material should be selected by matching the operating frequency, loss budget, impedance tolerance, trace length, environment, laminate stability, and production repeatability instead of choosing only by layer count, color, thickness, or the lowest quoted PCB price.

For lower-speed products, FR-4 may be fine. For RF, microwave, antenna, radar, or high-speed digital designs, low loss PCB material or Rogers-type laminate may be safer. The correct choice depends on actual electrical risk, not just the product name.

Is FR-4 suitable for high-frequency PCB design?

FR-4 can be suitable for some lower-frequency or short-trace high-speed PCB designs, but it becomes risky when insertion loss, phase stability, impedance control, antenna tuning, or repeatable RF performance are important at GHz-level frequencies.

I’d never reject FR-4 automatically. That’s too simplistic. But I also wouldn’t push generic FR-4 into a serious RF path just because it keeps the quote low. Low-loss FR-4 or RF laminate may be needed when performance margin is tight.

What is the difference between Dk and Df values?

Dk value describes how much a PCB material affects signal speed and impedance, while Df value describes how much signal energy is lost inside the dielectric as heat, so both values must be reviewed together for high-frequency PCB material selection.

Dk affects geometry, delay, phase, and impedance. Df affects loss. A material can have a usable Dk and still be a poor choice if the Df is too high for the frequency and trace length. That’s why dielectric constant and loss tangent belong in the same conversation.

Conclusion

If your PCB is low-frequency and not impedance-sensitive, don’t overcomplicate it. Use the right FR-4, control the basics, and move on.

But if your design involves RF, high-speed digital signals, controlled impedance, medical electronics, automotive electronics, flex interconnects, antenna paths, or long-term production repeatability, don’t leave dielectric constant and loss tangent until the last quotation round.

Send the Gerber files, stack-up notes, frequency range, impedance targets, copper requirement, material preference, and expected volume. We’ll help check whether standard FR-4, low-loss FR-4, Rogers-type material, HDI, flex PCB, or another build makes sense before the material choice becomes an expensive surprise.