Rigid-Flex PCB Design Guide: Bend Radius, Stiffeners, and Coverlay

A static (flex-to-install) design bends a handful of times during assembly and service and then stays put. A dynamic design flexes repeatedly in operation—a printer head, a hinge, a wearable strap. The distinction drives every downstream rule: layer count, copper type, bend radius, and where the copper sits relative to the neutral axis.

IPC-2223 expresses minimum bend radius as a multiple of total flex thickness at the bend. The ratios below are starting points for layout; final values depend on copper weight, adhesive, and construction, so confirm them with the fabricator. To turn a thickness into a radius, use the flex PCB bend radius calculator.

State the motion requirement explicitly in your drawings: number of flex cycles, bend angle, and radius. A fabricator reviewing "dynamic, 100k cycles, 180° over R6" can validate the construction. A drawing that just shows a fold line cannot be reviewed meaningfully—and optimistic assumptions default in.

Beyond the radius minimums, what you route inside the bend zone matters as much as the geometry. Run traces perpendicular to the bend axis, never parallel. Stagger traces on adjacent layers so they do not stack into a stiff column. Use curved, teardropped entries instead of right angles. Keep vias, pads, and plated holes completely out of the bend zone plus a margin of at least 1 mm—plated copper in a flexing region is the classic crack initiator.

For two-layer flex in dynamic applications, consider I-beam avoidance: offsetting top and bottom traces so their neutral axes do not align. Cross-hatched ground planes instead of solid copper reduce stiffness and help the flex survive—at a predictable cost to impedance control, which is why controlled-impedance dynamic flex needs early fabricator involvement.

Traditional flex constructions bond copper to polyimide with acrylic adhesive; adhesiveless laminates deposit copper directly on the film. Adhesiveless constructions are thinner, tolerate tighter bends, resist delamination through reflow, and have become the default for anything beyond simple static designs. Specify them when bend performance matters; accept adhesive-based cores only where cost dominates and bends are gentle.

Keep flex layer counts as low as the design allows—every added flex layer thickens the bend section and multiplies the minimum radius. A common pattern that works: route high-density interconnect in the rigid sections and let the flex carry only the signals that genuinely travel between them, on one or two layers. Bookbinder constructions (progressively longer flex layers) exist for thick multi-layer flex bends but add real cost; treat them as a last resort. Worked layer-by-layer examples are in rigid-flex PCB stackups for 4, 6, and 8 layers.

Stiffeners are bonded to flex areas that must not bend: under components and connectors, and at ZIF contact tails. FR-4 supports component areas and is the cheapest. Polyimide builds a ZIF tail up to the connector thickness, and it is the material that can be profiled with the outline to meet ZIF width and finger-position tolerances. Stainless steel or aluminum is used when space is tight and the area needs more support than an equivalent FR-4 thickness, or when aluminum doubles as a heat spreader.

Thermally bonded stiffeners, laminated with the same flexible adhesive used for coverlay, are the preferred attachment: permanent, reflow-safe, and usually cheaper. Pressure-sensitive adhesive (PSA) is the fallback when the design prevents thermal bonding; check that the PSA is rated for your reflow profile.

Placement matters more than material. Never end a stiffener exactly at a bend-zone boundary, because the stiffness step concentrates stress where the flex starts moving. On ZIF tails, do not line up the stiffener edge, the coverlay edge, and the pad-to-trace junction; let the pad overlap the coverlay opening and the stiffener overlap the pad-to-trace junction so insertion does not crack traces. Radius stiffener corners, and open holes in the stiffener slightly larger than the flex holes to absorb bonding misregistration.

Coverlay is the flexible equivalent of solder mask: polyimide film with adhesive, laminated over the circuits, with openings punched or laser-cut for pads. Openings should overlap pad edges (encroachment) so adhesive anchors the pad against peel forces—dimension them deliberately rather than copying rigid-board mask rules. In transition zones, coverlay from the flex side should extend into the rigid area by a controlled amount; your fabricator will have a preferred overlap, so ask rather than guess.

A reviewable rigid-flex package contains: a stackup drawing showing rigid and flex regions layer by layer; bend lines with radius, angle, and cycle count; stiffener locations, materials, and thicknesses; coverlay opening strategy; ZIF tail specs where applicable; and a 3D model or folded-state drawing if the installed geometry is complex. Include impedance tables if any flex segment carries controlled-impedance signals.

Also declare the IPC-6013 Type and Class on the drawing. Most rigid-flex boards are Type 4 (multilayer rigid-flex with plated through-holes), and the Type determines which plating, coverlay, and inspection requirements apply. A Class callout without a Type is incomplete.

Rigid-flex quotes without mechanical detail come back slower and wider—fabricators price uncertainty. The teams that get fast, tight rigid-flex pricing are the ones that treat the mechanical drawing as a first-class deliverable, equal to the Gerbers.