RF Communication PCB: How to Avoid Expensive Material Traps and Achieve Reliable High‑Frequency Performance

I have seen many engineers think that the design of high-frequency PCBs is too complicated, always thinking that they need to use the most expensive materials and the most top-notch craftsmanship to ensure performance. Actually that’s not the case. Once we were working on a millimeter-wave project. The customer initially insisted on using some kind of imported high-end substrate, which was prohibitively expensive. Later, we tried to use a more common material and just made some optimization adjustments to the wiring and grounding structure. As a result, the measured S parameters, especially the insertion loss, were actually better than their original plan. This incident inspired me a lot. In many cases, the performance bottleneck does not necessarily lie in the material itself but in the design ideas and understanding of the signal path. For example, we optimized the gradual curvature of key transmission line corners and increased the density of ground vias to better control electromagnetic field distribution and return paths. These small adjustments often have a more significant impact on high-frequency performance than small differences in substrate dielectric constants.

When it comes to testing, I particularly dislike the atmosphere of testing for the sake of testing. You put a board into the incubator and cycle it dozens of times and then take it out to measure the parameter changes. This is of course necessary, but you can’t just stop at recording data. You have to wonder why this parameter drifts at this temperature point? Is it a problem with the characteristics of the material, or is there something wrong with the solder joints or vias due to thermal expansion and contraction? This kind of thinking is much more valuable than simply recording a “pass” or “fail” result. For example, the observation that insertion loss deteriorates at low temperatures may indicate that the polarization characteristics of the dielectric material change with temperature, or that the thermal expansion coefficient mismatch between the copper foil and the substrate causes microcracks.

Nowadays, when it comes to the quality of RF communication PCBs, many companies like to brag about how low their defective rate is. Numbers such as a few parts per million sound bluffing, but I always feel that this is putting the cart before the horse. The real quality commitment should not just be a ppm value calculated in the final statistics, but should be something that runs through the entire design, material selection, processing and even testing concepts.
For example, in order to pursue extremely low insertion loss, you design the line width to be extremely fine. As a result, the processing yield cannot be improved, and you have to rely on post-screening to ensure shipment quality. This is actually a kind of internal friction. This is as unsustainable as dieting to the point of harming your health in order to lose weight and then relying on drugs to maintain it. A good design should find the best balance between performance, manufacturability and cost.

I think the reliability of high-frequency PCBs, especially those used in communication systems, is an overall concept that cannot be taken apart. You can’t say that the Dk/Df value of my board is optimal. If the signal integrity simulation result is perfect, then this board must be good. The actual assembly process welding quality and even the metal shielding of the final product casing will affect these factors that you cannot fully see in a simple board-level S-parameter test. For example, a filter that tests perfect on the board may have its response completely changed when installed in a metal cavity due to cavity resonances or improper mounting pressure. Therefore, system-level joint debugging and environmental testing are crucial.

So my current attitude towards testing is more inclined to regard it as a tool for verification and learning rather than a cold quality inspection level. Every time I get a new high-frequency PCB design, I will first take the time to understand its application scenario: how far the signal must travel, how powerful the power is, and how sensitive it is to noise, and then decide which parameters to focus on for testing, such as whether I need to pay special attention to the return loss or group delay characteristics of a certain frequency band. In this way, the test itself becomes a part of the design iteration and not just the end point.

I want to laugh every time I see someone talking about those industry standards as if they were magical. Yes, I am in the radio frequency communication PCB industry, but I have to tell you not to be fooled by a bunch of certifications. Getting a piece of paper to prove that you meet a certain standard, such as that IPC or some Class rating, sounds awesome, right, but it really doesn’t guarantee that your board will run smoothly at high frequencies.

I have seen too many factories. Their walls are covered with various certificates, from IPC to national military standards. The things they can make are almost meaningless. High-frequency PCB requires too much experience. It is not an art that can be completely limited by a standard process. If you think about how fast the signal runs on such a thin trace, any slight impedance mismatch or dielectric loss will greatly reduce the performance. At this time, it is far from enough to rely on machine detection data to meet the standards. You have to have a master to adjust the feeling based on experience.

Let’s take materials as an example. Now when it comes to low loss, many people only think of those big foreign brands, as if using Rogers boards is half the battle. This is actually a misunderstanding. Of course, the material is important, but more importantly, whether your processing technology can match this material. Some manufacturers use good materials in order to save costs or do not improve the process, but the advantages of the materials are ruined because of poor lamination temperature control or rough surface treatment. High-frequency performance is the result of the synergy between design, processing and materials. Being too superstitious about any one of them will lead to overturning.

So my view may be a bit counter-intuitive. Don’t think of certification as the first level or even the most important level for you to screen suppliers.
You should be more concerned about the projects they have actually done, especially the successful cases that have similar needs to yours. Ask them about the specific problems they encountered in the projects and how they solved them. A truly capable team’s eyes will light up when they talk about these details. They can tell you pitfalls that are not in the book, such as the resonance problem of a certain laminated structure at a specific frequency or the actual impact of a certain surface treatment on the insertion loss.

As for the traceability systems required by IATF or aerospace in automotive electronics, they are of course very important, especially in fields where you need to clearly divide responsibilities and have extreme requirements for reliability. But these systems essentially guarantee the standardization and consistency of the process to prevent human oversights. They cannot directly give you a high-frequency circuit board with excellent performance.

In the final analysis, making RF communication PCB, especially those involving cutting-edge applications such as 5G millimeter wave, is more like walking a tightrope. It requires finding an extremely delicate balance between theoretical design, actual process and cost. This process is full of compromises and trade-offs and there is no standard answer. The best suppliers are often not those who are best at taking exams and certifications, but those who truly understand what the signal is thinking and can implement this understanding with their hands.

So next time when you evaluate a supplier, you might as well ask less about what certifications you have and more about what technical difficulties caused you last time and how you solved them. The answer to this question is far more telling than a certificate.

Of course, I am not saying that standards are completely useless. Specifications like IPC provide the industry with a basic dialogue platform and quality bottom line. But for high-frequency applications that pursue top performance, it is not enough to just stay at the bottom line. You must find partners who are willing and able to go beyond standards because they understand that the real challenges often lie in the gray areas outside the standards.

I recently discovered an interesting phenomenon when I was helping the company find a new high-frequency PCB supplier: many manufacturers claim that they can make RF communication boards and high-frequency circuit boards with complete product lines. But when you really talk about it in depth, you will find that their so-called “capabilities” may be different from what you think.

In the past, I always felt that as long as the equipment looked advanced enough, the factory was reliable. Now I don’t think so. Of course, the equipment is important, but more importantly, do the people who operate these equipment really understand what high-frequency signals are about? For example, if you ask them about the impact of dielectric constant stability or loss tangent in practical applications, some people can talk to you in depth, while others can only memorize the parameters in the data sheet. The difference between them is huge. For example, a truly knowledgeable engineer will combine specific frequency bands and plate characteristics to explain how fluctuations in dielectric constant with temperature affect phase consistency, and can even share that they have improved the Dk stability of a certain high-speed material by adjusting the resin system and curing process. This kind of understanding extracted from actual cases is far more convincing than listing instrument parameters.
I met a factory. They had very good imported testing equipment. The workshop looked clean and tidy, but the samples they produced were barely satisfactory in signal integrity testing. Later I found out that their process engineers’ understanding of impedance matching was still in the theoretical stage and they did not have enough practical experience in multi-layer boards. In this case, even if the equipment is high-end, it is difficult to guarantee the stability of mass production. Specifically, they may know the impedance formula of microstrip lines, but they are not familiar with the local impedance deviation caused by the glass fiber braiding effect in actual production, and they are unable to predict and compensate for the slight change in dielectric thickness caused by the lamination process for different stacking schemes. These experiences often require many trials and errors and test iterations to accumulate.

Another thing that impresses me deeply is whether suppliers are willing to discuss their own limitations candidly. Some manufacturers are willing to do everything and dare to pick up any frequency band from the beginning, which makes people feel unsure. Those manufacturers who will proactively tell you “We have done something similar in this frequency band with confidence” or “We have not come into contact with this material before and need to do process verification first” are often more trustworthy. After all, radio frequency cannot be fake at all. For example, when faced with the demand for a 77GHz automotive radar board, a responsible supplier will inquire in detail about the layout of the antenna array, specific requirements for isolation, and frankly admit the boundaries of the existing capabilities in the PTFE material mixing process. This cautious attitude exactly reflects their professionalism.

Now I will pay more attention to the complexity of their past projects rather than simply looking at the equipment list. A team that has dealt with complex laminated structures and done rigorous reliability testing is more valuable than a team with only new equipment. It is like the difference between an old chef and a new chef. Everyone has the same tools, but the understanding of heat is the key. For example, a team that has successfully mass-produced high-frequency modules containing embedded resistors and capacitors, multiple laminations, and dense blind-buried via structures has experience in material selection, expansion and contraction control, and heat dissipation design that cannot be quickly obtained by purchasing new equipment.

As for those certification systems, I think they are more like admission tickets than quality guarantees. I have seen companies that have a full set of certifications but still have problems with quality control. I have also seen small factories that are not large but pay close attention to every link. So now I am more inclined to see their production processes and chat with engineers to feel their attitude towards details. For example, observe whether they strictly implement baking before processing high TG plates, or ask about their control points and monitoring frequency for electroplating uniformity. The rigor in these daily operations often reflects the true level better than the certificates on the wall.

Of course, cost control needs to be considered, but I would not put it first because the performance fluctuation of the RF board will bring greater hidden costs. Sometimes, in order to save a little money on board materials or processing fees, the final product performance is not up to standard and the cost of rework will be higher. Finding that balance point does require experience and honest communication from suppliers. For example, choosing a board with slightly greater loss but low price may result in insufficient budget margin for the system link in the millimeter wave band, and ultimately have to be redesigned. The time and material costs far exceed the original savings.
After all, choosing suppliers is a two-way process. While we are evaluating them, they are also observing whether we are reliable customers. Only when both parties respect technology and pursue quality can long-term cooperation be possible. A simple price war or technical bragging will not go far. Real good partners in this industry are all developed slowly.

After making radio frequency boards for many years, I have a very different experience. Sometimes people are too obsessed with chasing the top materials, such as having to use a specific type of PTFE or pursuing some kind of imported high-frequency board. Of course, materials are the basis, and that’s true, but I found that what really determines the upper limit of a board’s performance is often not how “magical” the material itself is, but how you use it.

I have seen many designs that used good high-frequency boards, but ended up with a lot of signal integrity problems because the wiring layout or stacked structure was not done right. It’s like giving you a top-notch steak, but grilling it over high heat until it becomes mushy. For RF, PCB design rules are almost part of the laws of physics. For example, impedance matching may cause the efficiency of the entire link to drop. There is also the treatment of the copper foil. The surface roughness has a great impact on the loss of high-frequency signals. No matter how good the substrate is, if the surface of the copper foil is bumpy, the high-frequency signal will still suffer heavy losses when it passes over it.

So now when I look at whether a supplier is reliable, I don’t usually ask him first if he has a certain “Internet celebrity” material in stock. What I would rather talk about is how deeply their engineering team understands the design. For example, if I send a preliminary RF Communication PCB design, can they quickly identify potential risks? For example, will the dielectric constant of a multilayer board suddenly change when the voltage is mixed? How to deal with the fringe field effect of the microstrip line so as not to affect the sensitive lines next to it.

Speaking of this, I think there is a trend that is underestimated, which is that the collaboration between design and manufacturing must be advanced and advanced. In the past, you might have drawn the board and then went to the factory for DFM inspection. Now it is best to involve experienced manufacturers when drawing the schematic diagram and planning the stack-up. They are exposed to various practical production problems every day, know which high-frequency PCB structure is easier to control tolerances, and know which type of via design has a higher yield in mass production. Feedback of this experience to the design end can save countless troubles in the later stage.

Another point is the old issue about heat dissipation. It is inevitable that the high-power RF part will heat up, and everyone is thinking about adding more thermal vias or using thicker copper foil to conduct heat. But sometimes the idea can be changed: Can we consider treating the PCB area under the chip that generates the most heat separately? For example, locally using materials with higher thermal conductivity to inlay them, or optimizing the array distribution of thermal vias instead of simply covering them. This requires close coordination between design and process to achieve this.

As for future applications of higher frequency bands such as millimeter waves, the challenges will definitely be greater. But I think instead of worrying that the materials are not new enough, it is better to first understand the limits of the existing technology.
Controlling the inter-layer alignment accuracy better and making the side walls of the circuit smoother. These improvements in basic skills may contribute more to signal integrity than changing to a new material. Of course, things like ultra-low profile copper foil will definitely become more and more important because as the frequency increases, the skin effect becomes more and more obvious. The smoothness of the conductor surface directly determines the loss.

Generally speaking, I think making radio frequency boards is a bit like traditional Chinese medicine, which emphasizes systemic balance and harmony. Materials, design, and technology must form a closed loop to continuously feedback and optimize each other. Pursuing a single extreme while ignoring the whole will often not yield the best results. This is what I have learned over the years.

Many people think that making high-frequency circuit boards is just a matter of following the drawings. In fact, this is not the case at all. I’ve seen too many projects get stuck on signal integrity issues. At first, everyone thought that the design software did not adjust the parameters properly or the components were selected incorrectly. After much tossing and tossing, I discovered that the problem lies in the most basic board.

Those invisible and intangible signal losses are the biggest headache. Do you think you can get through it by connecting the line to the signal? This is not the case at all in the high frequency world. When the signal walks on it, it will attenuate, reflect, and even fight with itself.

I have a friend who suffered a setback in his previous automotive radar project. Their team’s design was very beautiful, and the simulation passed the results. A sample test found that the detection distance was always not up to standard. After checking for a long time, I finally found that the wrong plate material was selected, which caused unstable dielectric constant and affected the performance of the antenna array.

This made me understand a truth: making RF PCB cannot only depend on how high the processing accuracy is or how advanced the equipment is.

A truly knowledgeable manufacturer will focus on material selection and process control. For example, they know how to process the surface roughness of copper foil to reduce transmission loss and how to control impedance continuity through precise stacking design.

These details often determine the success or failure of the entire system. For example, in the millimeter-wave frequency band, tiny bumps on the surface of copper foil can significantly increase the skin effect loss of the signal, while subtle fluctuations in the resin content between laminates can cause impedance mutations and trigger signal reflections. An excellent process team will optimize the browning or copper deposition process to obtain a smoother conductor surface, and use high-precision dielectric thickness control to ensure the uniformity of impedance of each layer.

Many factories now advertise that they can make high-frequency boards, but their actual capabilities vary greatly. Some just bought a few good machines, while others accumulated more than ten years of experience in basic craftsmanship.

The latter are usually more trustworthy because they understand how electromagnetic waves “walk” in circuit boards rather than just know how to “make” a board. They could foresee that a poorly positioned via could act as a resonant cavity, or that a tiny gap in the ground could radiate unwanted electromagnetic energy.

So when you need to find a supplier, don’t just look at their brochures or quotations. Ask them about the specific problems they have encountered and how they solved them.
A truly experienced team will talk to you about plate characteristics, phase consistency, and how to maintain stable performance in mass production instead of just talking about delivery time and price. They may share how to customize the lamination parameters for different customers to stabilize the dielectric constant, or how to control the accuracy during solder mask window opening to avoid causing additional parasitic capacitance to high-frequency signal paths.

After all, the core of radio frequency communication lies in how to keep signals pure and reliable in complex physical structures. This requires in-depth collaboration between design and manufacturing.

A good manufacturer should be able to become an extension of the design team and predict potential risk points in advance instead of passively waiting for drawings to be issued. For example, during the review phase, they may recommend moving a critical transmission line from the inner layer to the surface layer to reduce discontinuities introduced by vias, or recommend using a bond pad material with a lower loss factor.

In this field, experience is often more valuable than equipment because many problems cannot be detected by machines and need to be solved by human judgment. For example, subtle differences in performance between batches of the same plate, or the impact of ambient temperature and humidity on the processing process, all require engineers to control and compensate with their long-term accumulated “feel” and “eyesight” to ensure that each batch of products can meet stringent high-frequency performance requirements.