Rigid-Flex PCB Stackup Construction: Flex-in-Core vs Flex-on-Outer

A rigid-flex stackup is not a rigid stackup with a flex board glued on. The flex core runs continuously through the rigid sections and emerges as the bending layer — which means where you place the flex layers inside the stack decides almost everything about reliability, lamination complexity, and cost. This guide covers the construction choices that define a rigid-flex stackup: flex-in-core versus flex-on-outer-layers, symmetric build, no-flow prepreg, the bookbinder and air-gap techniques for multiple flex layers, and how layer count drives lamination cycles.

TL;DR

• The flex core is continuous. It runs through the rigid sections and is exposed in the bend region. Where it sits in the stack is the central design decision.
• Flex-in-core puts the flex layers at the center of the stack — the most common, most reliable build for 4-8 layer rigid-flex.
• Flex-on-outer-layers exposes the flex as the outermost conductors — used when the flex must be very thin or peel away to one side.
• Build symmetric. Center the flex core, balance copper and dielectric about the centerline, or the board warps.
• No-flow prepreg bonds rigid to flex without resin squeezing into the bend.
• Multiple flex layers in a bend are normally kept as independent sets (air-gap, also called loose-leaf), and for tight bends the sets can be cut to stepped lengths (bookbinder), so each set can slip past the others instead of being bonded into one stiff laminate.
• Layer count drives lamination cycles, and each cycle adds cost and risk.

This guide is rigid-flex specific. For pure multilayer flex (all-polyimide, no rigid FR-4 sections), the construction is different — see the multilayer flex PCB stackup guide. For the broad design rules that govern the whole board, see the rigid-flex PCB design guidelines and the rigid-flex service overview.

Where the Flex Layers Sit: Two Constructions

The defining choice in a rigid-flex stackup is where the continuous flex core sits relative to the rigid copper layers.

Flex-in-Core (Internal Flex)

The flex layers sit at the center of the stack. The rigid FR-4 layers and prepreg build out symmetrically above and below the flex core, and the flex core is the only thing that continues into the bend region.

This is the standard, most reliable construction for most 4-, 6-, and 8-layer rigid-flex boards because:

• The flex sits naturally near the neutral axis of the bend.
• The build is symmetric, so the board stays flat.
• The flex emerges from the center of the rigid stack, giving a clean transition.

Flex-on-Outer-Layers (External Flex)

The flex conductors are the outermost layers, with rigid build-up only on the inside. The flex peels away from one face of the rigid stack.

Use this when:

• The flex must be extremely thin (only one or two layers cross).
• The flex needs to fold back over the top of a rigid section.
• The mechanical design requires the flex to exit one face, not the centerline.

The trade-off is that an outer-layer flex sits off the neutral axis and the build is harder to keep symmetric, so it is reserved for cases where the mechanics demand it.

Symmetric Construction

A rigid-flex stackup must be balanced about its centerline or it will warp after lamination and shift the neutral axis off the copper. Rules:

• Center the flex core. For a 2-layer flex section, the flex copper should straddle the centerline.
• Mirror layer count and copper weight above and below the center.
• Match dielectric thicknesses on either side.

Layer symmetry is one of the core design rules covered in the rigid-flex design guidelines; the stackup is where you actually enforce it. Model symmetric candidate builds with the stackup builder before committing.

No-Flow Prepreg: Bonding Rigid to Flex

Standard FR-4 prepreg has high resin flow — under heat and pressure the resin spreads. In a rigid stackup that is exactly what you want. At a rigid-to-flex transition it is a disaster: the resin squeezes into the flex region and locks it solid.

No-flow (or low-flow) prepreg is formulated to stay in place during lamination. It bonds the rigid FR-4 layers to the flex core without flowing into the bend region. The layout must define a resin-flow keepout at the transition so the flex emerges clean. This single detail — specifying no-flow prepreg and its keepout — is the difference between a board that bends and a board that arrives stiff. The mechanics of that boundary are detailed in the rigid-flex transition zone design rules guide.

Multiple Flex Layers: Bonded, Air-Gap and Bookbinder

When more than one flex layer has to bend in the same region, the first decision is whether those layers are bonded together or kept apart. There are three constructions, and they build on each other.

• Bonded: all the flex layers are laminated into one stack with adhesive, in the rigid sections and in the bend. It is the simplest to draw and the stiffest to bend.
• Air-gap (loose-leaf, unbonded): the flex layers are kept as independent sets that are not bonded to each other in the bend region. Each set bends on its own thickness. In the rigid sections the sets are separated by FR-4-based prepreg rather than flex adhesive.
• Bookbinder: an air-gap build in which each unbonded set is also made slightly longer than the one under it, so the sets have room to nest around a tight bend without pushing on each other.

Air-gap and bookbinder are therefore not two competing techniques: bookbinder is air-gap plus stepped lengths. The reason to step the lengths is geometric. Layers on the outside of a bend travel a longer path than layers on the inside, and if they are bonded or cut to the same length, the inner set is forced into compression and the outer into tension.

Why bonded flex layers bend poorly

Stacking flex layers makes the bend region thicker, and a thicker section strains the copper more at the same radius. Design guides from fabricators give a common minimum-bend guideline of about 10 times the flex thickness for a one- or two-layer flex, rising to 20 to 30 times for a laminated four-layer flex section (Epec, air-gap benefits; Sierra Circuits, air-gap construction). These are guidelines, not limits for your board: your fabricator's design rules and the bend count you must survive decide the final radius.

Bonded stacks also put flex adhesive inside the rigid section. Flex layers and their adhesives expand more with heat than FR-4, so thermal cycling can load the plated vias, which is the via-reliability concern that air-gap construction was developed to remove (Sierra Circuits). Epec reports that in an IPC interconnect stress test an air-gap construction passed the cycle count while a rigid-flex with four flex layers laminated together failed (Epec).

How air-gap construction is limited

• Set size: the usual rule is independent sets of no more than three flex layers each. Two layers per set is described as optimum, and IPC-2223 allows three when the design has controlled-impedance lines in the flex area (Sierra Circuits).
• Extra process: each flex pair needs coverlay on both sides, which adds fabrication steps and cost (Epec).
• Bend shape: in a tight U-shaped bend, the pair on the outside can constrain the pair on the inside until it buckles. S-shaped bends are preferred because they cancel out the difference in bend radius between pairs (Epec).
• Set count: the more sets in the bend, the more they limit how tight it can go.

What bookbinder construction adds

Bookbinder is for the case where air-gap alone does not give enough room. A paper presented at an IPC conference by engineers from IBM, Lenovo and Amphenol describes a 15-layer server rigid-flex that had to make a 90-degree bend in a flex section only 15.25 mm long (Dangler, Taylor and Verbrugge, IPC). The authors work through the numbers:

Splitting the stack into unbonded sub-composites cut the guideline radius to about a third. The sub-composites were also made progressively longer from one to the next, so each had room to rest under the next without interference; in that design the widths stepped down by roughly 0.77 mm per group, from 47.46 mm to 45.16 mm across the groups. Those lengths belong to that board. Your fabricator calculates the step for your bend radius and layer thicknesses, so do not copy them.

A worked example with assumed numbers

Assume four flex layers of 0.1 mm each in the bend, all at one bend line. Laminated as a single 0.4 mm section, the 20 to 30 times guideline for a laminated four-layer flex gives 8 to 12 mm. Kept as two air-gap pairs of 0.2 mm each, the 10 times guideline for one or two layers gives 2 mm for each pair. If the enclosure only allows a 3 mm radius, the bonded build does not fit and the two-pair build does. If it allows 1.5 mm, even the pairs are too stiff, and the answer is fewer flex layers crossing the bend, a bookbinder build, or a different mechanical design. These are assumptions to show the arithmetic, not a design rule.

Choosing the construction

Both air-gap and bookbinder add fabrication steps and cost, so use them only when the flex genuinely needs multiple bending layers. For most boards the right answer is to minimize the flex layer count first by routing only what must cross, covered in the rigid-flex layer count guide. For the general bend-radius rules, see the flex PCB bend radius design guide.

What to put on the fabrication drawing

• The construction for the bend region: bonded, air-gap sets, or bookbinder, and which layers belong to which set.
• The number of flex layers crossing the bend and the thickness of each set.
• The required bend radius, whether the bend is static or dynamic, and the bend shape (U or S).
• Coverlay on both sides of each pair, and where the unbonded region starts and ends.
• For bookbinder, that the fabricator will calculate and confirm the stepped lengths from your radius and thicknesses.
• The IPC classes and the test evidence you expect; see the fabrication drawing guide and IPC-6013 inspection checklist.

Layer Count Drives Lamination Cycles

A rigid-flex board can require multiple lamination cycles: the flex core is built and coverlaid first, then the rigid layers are laminated around it, sometimes in stages, sometimes with sub-lamination of HDI build-up layers. Each additional cycle:

• Adds process time and cost.
• Adds a thermal-stress event the flex must survive.
• Increases the risk of registration error and delamination.

The number of cycles depends on the stackup and the fabricator's process, so ask for it instead of assuming a number. In general, a plain flex-in-core build needs the fewest, HDI build-up layers add cycles, and an air-gap or bookbinder build adds fabrication steps again because each flex set is processed with its own coverlay.

This is why lamination-cycle count is a primary cost driver — quantified in our rigid-flex PCB cost drivers guide. When you choose a layer count, you are also choosing a lamination-cycle count, so the two decisions are linked: see how to choose 4 vs 6 vs 8 layers.

FAQ

What is flex-in-core construction?

Flex-in-core means the continuous flex layers sit at the center of the rigid-flex stackup, with rigid FR-4 building out symmetrically above and below. It is the standard construction for most 4- to 8-layer rigid-flex boards because the flex sits near the neutral axis of the bend and the symmetric build keeps the board flat.

When do you use flex-on-outer-layers instead?

When the flex must be very thin (one or two layers crossing), when it has to fold back over a rigid section, or when the mechanical design requires the flex to exit one face of the board rather than the centerline. The trade-off is that the flex sits off the neutral axis and the stackup is harder to keep symmetric, so it is used only when the mechanics demand it.

What is bookbinder construction in rigid-flex?

Bookbinder construction is an air-gap (unbonded, loose-leaf) build in which each flex set is made slightly longer than the one under it, like the pages of a book, so the sets have room to nest around a tight bend without pushing on each other. It is used when air-gap sets alone cannot reach the radius, and it adds fabrication steps and cost. The fabricator calculates the stepped lengths from your radius and layer thicknesses.

What is the difference between bonded, air-gap and bookbinder flex layers?

Bonded layers are laminated together, which is simplest but stiffest and puts flex adhesive in the rigid section. Air-gap layers are kept as independent sets, usually two layers per set and no more than three, which bend on their own thickness. Bookbinder is air-gap with stepped set lengths for tight bends. For the bend radius you need and the number of layers that cross it, the sections above show how to choose.

Why does rigid-flex need no-flow prepreg?

Standard FR-4 prepreg flows under heat and pressure, and at a rigid-to-flex transition that resin would squeeze into the flex region and stiffen it. No-flow (low-flow) prepreg stays in place during lamination, bonding the rigid layers to the flex core without flowing into the bend. The layout must also define a resin-flow keepout at the transition.

Get a Rigid-Flex Stackup Built

Tell us your layer count, where the flex has to bend, and your radius target, and we will propose a symmetric, manufacturable stackup with the right lamination plan. Request a quote or contact our engineers.

References:

1. IPC — Association Connecting Electronics Industries. IPC-2223 Sectional Design Standard for Flexible Printed Boards
2. IPC-6013 Qualification and Performance Specification for Flexible Printed Boards