Stackup Choices: Low Loss Laminate or FR4
Every board begins with a stackup, and that decision is hard to reverse. An FR4 PCB stackup fixes the dielectric constant, the loss, the layer count and the impedance targets before a single trace is routed, while a high frequency PCB stackup changes all of those numbers for the same physical geometry. Knowing what separates the two lets a design team keep the cheaper option wherever the electrical margin allows and spend on material only where the loss budget truly demands it. This guide compares the two families across materials, electrical behaviour, impedance control, cost and fabrication, and shows where a hybrid build makes sense.
What a Stackup Decides Before Routing Starts
A PCB stackup is the ordered arrangement of copper layers, cores and prepregs that make up the finished board. It fixes the dielectric thickness between each pair of layers, and with it the characteristic impedance of every controlled trace, the coupling between adjacent signal layers, and the amount of copper available for power and ground. Those numbers are set long before the first route is drawn, and the layout team inherits them.
The effects propagate through the whole programme. The distance from a signal layer to its reference plane sets the trace width needed for fifty ohms; the dielectric material sets how much of the signal is lost per unit length; the copper weights set how much current the planes can carry. A stackup defined late, or copied from an unrelated project, forces compromises that reappear as impedance discontinuities, crosstalk and thermal problems after fabrication, when changes are expensive.
How an FR4 Stackup Is Built
FR4 is a woven glass fabric bonded with an epoxy resin, and it remains the default substrate for the great majority of boards. A typical four layer FR4 PCB stackup places the two outer copper foils on either side of a core, with prepreg bonding the inner layers. Layer one carries components and signals, layer two is a solid ground plane, layer three is a power plane split as required, and layer four carries the remaining routing. The dielectric constant of standard FR4 sits between roughly four and four point eight depending on resin content and glass style, and the value moves with frequency and temperature.
Variants refine the base material without changing the process. High glass transition grades raise the temperature limit for lead free assembly and thick boards, low loss FR4 reduces dissipation for moderate speed designs, and halogen free grades address environmental requirements. None of these variants turns FR4 into a microwave material, but a well chosen low loss grade can carry multi gigabit links over short distances at a fraction of the cost of a specialised laminate.
How a High Frequency Stackup Differs
High frequency materials replace the epoxy glass system with chemistries that lose far less energy. PTFE based laminates, ceramic filled composites and modified resin systems such as the RO4000 series and the Megtron family dominate this group. Their dielectric constants range from about two point two for PTFE up to around three point five for ceramic filled grades, and the loss tangent, the number that governs insertion loss, is typically an order of magnitude lower than standard FR4.
The stackup consequences are practical rather than cosmetic. A lower dielectric constant means wider traces for the same impedance, which improves conductor loss and manufacturing tolerance. A lower loss tangent means the dielectric contribution to attenuation stays small even at ten gigahertz and beyond. The trade is mechanical and thermal: PTFE is softer, expands more with temperature, and needs different drilling, plating and lamination parameters, so the fabrication window is narrower and the price is higher.
Dielectric Constant and Insertion Loss in Practice
Dielectric constant controls two things at once. It sets the propagation velocity, so a high value slows signals and changes the electrical length of a route, and it sets the impedance for a given trace geometry. When a supplier batch drifts, the impedance of every controlled trace drifts with it, which is why impedance tolerance and material consistency are quoted together on high speed projects.
Insertion loss is where the gap becomes dramatic. In FR4 the loss rises quickly above a few gigahertz, dominated by the dielectric, and a long channel can consume most of the link budget. A low loss laminate flattens that curve. For a ten gigabit channel spanning a backplane the difference is the margin between a working link and one that needs heavy equalisation, and for radar and satellite hardware it is the difference between a usable front end and an unusable one.
Impedance Control Across the Two Families
FR4 stackups are normally held to plus or minus ten percent on controlled impedance, a figure that mature processes achieve comfortably. High frequency stackups target plus or minus five percent or tighter, because the applications that justify the material rarely tolerate more. Meeting the tighter number requires well characterised dielectric thickness, tightly specified copper weights, and test coupons measured on the same panel as the product.
Simulation supports both cases. Field solvers and full wave tools derive geometry from the stackup, and the results are verified against coupons after fabrication. What differs is sensitivity: on a low loss material a small thickness error still leaves the impedance close to target, while on FR4 the same error moves it further, so process control carries more of the burden and the fabricator matters more.
Cost, Availability and Lead Time
FR4 is cheap because the supply chain is enormous and every fabricator runs it daily. A four layer prototype is a commodity purchase, and volume pricing reflects little more than copper and processing. High frequency laminates are made in fewer plants, in smaller panels, and often with longer lead times. A specialised prototype can cost several times an equivalent FR4 board, and a complex radio frequency multilayer can cost far more again.
The decision is therefore not which material is better but where the electrical margin actually runs out. Below a few gigahertz, with short routes and a generous timing budget, FR4 is the rational choice. Above that, or where loss, phase stability or impedance precision decide whether the product works, the material premium buys performance that no amount of layout effort can recover from a lossy substrate.
Hybrid Stackups: Both Materials in One Board
The most common answer on modern hardware is a hybrid. One stackup can place a high frequency laminate only on the layers that carry the radio or the fastest serial links, with ordinary FR4 forming the rest of the build. The radio layer sits close to the surface, referenced to a ground plane, while the digital and power layers stay on standard material.
Hybrid builds cut cost sharply compared with an all high frequency board, but they add fabrication complexity. The two materials expand differently, so lamination cycles, panel size and registration tolerances must be planned around them, and the prepreg chosen to bond them must suit both. Discussing the mix with the fabricator before the stackup is frozen removes most of the risk. Our notes on high frequency laminate selection criteria and PTFE versus ceramic laminates cover the material side of that decision.
Choosing a Stackup for a Programme
Start from the highest frequency that must travel a meaningful distance, and the loss budget that link can absorb. If the answer fits inside FR4 with margin, stay there and spend the money on layer count and grounding instead. If it does not, identify which nets need the better dielectric and build the hybrid around them rather than converting the whole board.
Then confirm the mechanical side: board thickness and layer count follow from the connector, the enclosure and the assembly process, while the impedance targets follow from the interface specifications. A stackup that satisfies all three, reviewed by the fabricator before layout starts, is worth more than any later correction. Related reading covers power plane design and the wider set of PCB manufacturing processes the choice of material affects.
FAQ
Can FR4 be used for high frequency designs? For moderate frequencies and short routes, yes. Low loss grades push the usable range higher, but above a few gigahertz the dielectric loss and the variation in dielectric constant become the limiting factors, and a dedicated low loss laminate is normally required.
Is a high frequency stackup always more expensive? Per square metre, yes, often several times more. In a hybrid build only the radio layers use the expensive material, which brings the premium down to a fraction of the board while keeping the electrical performance where it matters.
What dielectric constant should be used in simulation? The value at the operating frequency, not the one megahertz figure on the datasheet. Both dielectric constant and loss tangent vary with frequency, so the supplier curve for the specific laminate and resin content should drive the model.
