Six-Layer Stackup Options: Signal, Ground and Power Layers

A six-layer board is the first stack that gives a designer a dedicated ground plane and a dedicated power plane without giving up two routing layers. That is why it has become the default for consumer products, industrial controllers and anything with a microcontroller and a few interfaces. The structure is simple once it is laid out, but the choices inside it matter more than the layer count suggests.

What the Six Layers Are

A six-layer board is built from one core with copper on both faces, plus prepreg and foil added to each side. The core contributes two copper layers, and each outer addition contributes one more, which is where the even number comes from. Copper layers are what the layer count refers to; the dielectric between them is not counted.

The six layers are normally assigned to signal, ground and power in some combination, and the outer layers are almost always signal. That leaves four inner layers to divide between routing and planes.

The Four Common Configurations

The usual arrangements are two signal layers and two planes, three signal layers and one plane, or one signal layer and two planes plus a second ground. A stack with two inner signal layers and a ground and power pair is the most common because it gives predictable impedance on the outer microstrip layers and good return paths throughout.

A stack with two ground planes and one power plane is the choice for a design with several supply rails and little inner routing, and it has the lowest impedance between the planes. A stack with three inner signal layers is used when the routing does not fit, but it costs the low-impedance plane pair that makes the six-layer board attractive in the first place.

Why the Signal Layers Sit on the Outside

Outer signal layers are microstrip: a trace on the surface with a plane below it, which is easy to fabricate and easy to route because components are on that side. Inner signal layers between two planes are stripline, which is better shielded but needs a via every time the signal has to reach the surface.

The mixed arrangement used by most six-layer boards puts the impedance-critical high-speed signals on the inner layers as stripline, and the general routing on the outer layers as microstrip.

Reference Planes and Return Paths

Every signal layer should have a solid reference plane adjacent to it. In a six-layer stack that is straightforward: route the outer layers against the ground plane below, and route the inner signal layers between the ground and power planes. Then keep the reference intact. A ground plane that is split under a high-speed trace forces the return current to detour, which increases loop area and radiation.

The power plane can serve as a reference for signals that are not sensitive, because the return current will take the path of least impedance, which at high frequency is the nearest plane regardless of its net. It is not a substitute for a ground plane on critical nets, and the layer assignment should not be made as though it were.

Prepreg, Core and Dielectric Thickness

The dielectric thickness between a signal layer and its reference sets the impedance, so the prepreg specification is an electrical parameter. Standard prepreg comes in a limited set of thicknesses, and the fabricator will build the stack from what is available, which is why the impedance calculation should be done using real prepreg values rather than an idealised dielectric thickness.

Where a specific impedance is required, state the target and the tolerance, and ask for the stackup that will be used to be confirmed before production. A stackup that is adjusted at the shop without an impedance re-calculation is a stackup that no longer matches the design.

How the Board Is Built

The inner layers are imaged, etched, inspected and oxide-treated, then laid up with prepreg between them and pressed. After pressing, the panel is drilled, plated, imaged on the outside, etched, coated with solder mask and finished. Each step has a tolerance, and the registration budget accumulates through lamination and drilling, so the fine-line capability of a six-layer board is set by the fabricator’s registration control rather than by the artwork alone.

Thickness follows the same logic. The finished thickness is the sum of the cores, the prepreg, the copper and the finish, and if the design has a hard thickness limit, the stackup has to be designed around it from the start.

The fabricator also needs to know how the layers will be used before the stack is fixed, because the copper distribution on each layer affects how the prepreg flows and how evenly the panel presses. A stack with one layer almost entirely copper and the next almost empty will press unevenly, and the resulting thickness variation shows up directly as an impedance variation on the outer layers. Balancing the copper, even with a thieving pattern on a sparse layer, is a cheap way to keep the six-layer stackup predictable. Multilayer PCB advantages in high speed designs explains why that predictability is worth more as the edge rates rise.

When Six Layers Is the Right Answer

Six layers is the point where the extra cost buys a real electrical improvement. Below it, a four-layer board with a ground and power plane can usually serve, provided the routing fits and the impedance can be controlled. Above it, the additional layers buy more routing or more planes, but the first step from four to six is the one that most often pays for itself.

The trigger for going to six layers is usually one of three things: a bus that needs controlled impedance on more than one layer, a component density that will not route on two signal layers, or a supply architecture with several rails that need low-impedance distribution. Layer stackup design for one to eight layers sets out the trade-offs at each count.

Common Mistakes

The most frequent mistake is to treat the inner layers as free routing space and forget the planes. A six-layer stack that routes signals on all four inner layers and relies on a fragmented ground is nothing more than a four-layer board with extra steps.

The second mistake is to change the stackup after layout. A different prepreg thickness changes every controlled impedance on the board, and the error is invisible until the board is measured. The third is to leave the power plane whole and cut the ground, which inverts the intended reference for most of the routing. Power plane splitting rules covers how to divide a plane without breaking the return path.

Six-layer boards are also frequently used with SMT-only assemblies, and the stackup has to survive two reflow passes if parts are placed on both sides. A symmetrical stack, which is automatic with an even count, keeps the board flat through both passes.

FAQ

Is a six-layer board much more expensive than four layers? It costs more, but less than most designers expect, because the fabrication steps are similar and only the material and lamination change. The premium is usually small compared with the cost of adding another design iteration.

Do all four inner layers have to be used? No. An unused inner layer can be a second ground plane, which lowers the plane impedance and improves the return path. Leaving a layer empty is wasteful but not harmful.

What thickness should the dielectric be? Whatever the impedance calculation requires, chosen from the prepreg thicknesses the fabricator actually stocks. Confirm the final stackup before production and check the impedance coupon results against it.