Flex PCB Bend Radius: Rules for a Circuit That Folds

A flexible circuit is designed to be folded, and the number that decides whether it survives that fold is the bend radius. Specify it correctly and the flex behaves like a hinge for the life of the product. Get it wrong and the copper cracks at the outside of the bend after a few hundred cycles, usually in a unit that has already passed every test.

Why the Radius Matters

When a circuit is bent, the material on the outside of the curve is stretched and the material on the inside is compressed. Nothing in the middle changes length, which is why that middle plane – the neutral axis – is where conductors want to be.

The strain in the copper depends on how far it sits from the neutral axis and how tight the bend is. A thin layer of copper far from the neutral axis, wrapped around a small radius, is stretched far beyond its elastic limit, and copper does not stretch: it cracks. The failure is a fatigue crack that starts at a grain boundary, and it grows with each cycle until the trace opens.

Two other failure modes accompany it. The adhesive or the bond between the copper and the polyimide can delaminate, particularly at a sharp inside corner, and the coverlay can separate from the base material at the same place. Both reduce the mechanical support that keeps the copper intact, and both are accelerated by a tight radius.

Static and Dynamic Flexing

The single most important distinction in flex design is whether the circuit will be bent once or many times, because the two cases allow very different radii.

A static flex is bent during assembly and then fixed in position. The copper only has to survive one deformation, plus whatever creep and thermal movement occurs in service. Static designs commonly use a bend radius of about ten times the total thickness of the flex stack, and some constructions allow less.

A dynamic flex is bent repeatedly in operation: a hinge on a laptop, a print head cable, a medical device, a robot arm. Here the copper is subjected to fatigue, and the radius has to be far larger, typically in the region of one hundred times the material thickness for a long life, with the exact figure depending on the copper type, the thickness and the number of cycles required.

The difference between those two numbers is an order of magnitude, and it is the reason why a flex design should always state which category it belongs to before any routing is done.

How the Radius Is Calculated

The radius is quoted as a multiple of the total thickness of the flex stack, including the base film, the adhesive, the copper and the coverlay. Two conventions are used, and both should be stated explicitly to avoid a factor of two error.

The inside radius is the radius of the mandrel or the surface the flex is formed around, measured to the inside face of the flex. The neutral axis radius is the radius of the middle of the stack. The second is the one that governs the strain in the copper, and the two differ by half the stack thickness, which matters when the stack is a few tenths of a millimetre thick.

The calculation itself is a ratio. Take the distance from the neutral axis to the copper layer you are concerned about, divide it by the neutral axis radius, and you have the strain. Copper tolerates roughly 0.3 percent strain before it starts to fatigue in a static bend, and far less for a long fatigue life, which is why a dynamic design needs a radius an order of magnitude larger than a static one.

In practice the design rule is expressed as a rule of thumb rather than a calculation. For a static bend, a radius of ten times the stack thickness is a safe working figure for a single sided flex with standard copper. For a dynamic bend, a hundred times the thickness is a reasonable starting point for a long cycle life, and the specialist supplier should be asked for the figure that applies to the specific construction.

Materials and Construction

The choice of copper and of construction changes the achievable radius as much as the geometry does.

Rolled annealed copper has a grain structure that allows it to deform further before cracking, and it is the standard choice for dynamic flex. Electrodeposited copper is cheaper, has better dimensional stability and is acceptable for static bends, but it fatigues sooner under repeated flexing.

Adhesiveless laminates place the copper directly on the polyimide rather than bonding it with an acrylic layer. Removing the adhesive reduces the total thickness, brings the copper closer to the neutral axis, and removes a layer that is prone to delamination. For a dynamic application this is usually worth the price.

Coverlay versus solder mask is the other major choice. A polyimide coverlay with an adhesive provides mechanical support and is standard for flex, while a photoimageable coverlay or a flex-compatible solder mask is thinner and allows finer features but offers less protection at a bend. In the bend region the coverlay material and its thickness both matter, because they sit at the outside of the stack and contribute to the strain.

The base film thickness is the last variable. A 12 micrometre polyimide flexes far more readily than a 50 micrometre one, and where a tight radius is unavoidable, thinning the base and the coverlay is often more effective than adding layers to move the copper to the middle.

Layout Rules for the Bend Area

Inside the bend region, a different set of rules applies than anywhere else on the circuit.

• Route conductors perpendicular to the bend line. A trace that crosses the bend at an angle, or worse runs parallel to it, experiences a much longer strained path and fails sooner.
• Keep the copper near the neutral axis. On a multilayer flex this means placing the conductor layers symmetrically in the stack, and on a single sided design it means keeping the copper thin.
• Use the thinnest copper that meets the current requirement. A 35 micrometre copper foil can survive bends that a 70 micrometre foil will not, and the current carrying difference between them is often smaller than expected.
• No plated through holes in the bend region. A plated barrel is a rigid inclusion in a flexible material and will crack the surrounding copper. Route vias outside the bend or, where that is impossible, accept a very generous radius.
• No pads, no component sites and no stiffeners inside the bend. A stiffener turns the bend into a hinge at its edge, concentrating strain exactly where the stiffener ends.
• Stagger conductors rather than bunching them. Parallel traces in a bend stiffen the region and reduce the flexibility that the design intended.
• Use a cross hatched or thinned ground plane in the bend area rather than solid copper, since a solid plane raises the effective stiffness and moves the neutral axis.
• Add fillets where a trace meets a pad outside the bend, so the same stress concentration rules that apply on a rigid board apply here.
• Keep the bend line clear of any feature that cannot flex, and define the bend area on a mechanical drawing so the assembly instruction and the fabrication drawing agree.

The bend area is also defined by its length, not only by its radius. A bend needs a region over which the curvature develops, otherwise the flex is forced into a sharp crease at one line. Marking that region on the drawing tells the fabricator, the assembler and the mechanical designer the same thing, which matters because the flex is often formed inside a housing by someone who has never seen the layout.

Verifying the Design

Two tests matter, and they are different from the tests a rigid board receives.

The first is a bend test on a coupon. A sample of the same construction, with the same copper type and the same stack, is bent to the specified radius repeatedly until it fails, and the number of cycles is recorded. This is the only reliable way to know whether a dynamic flex will survive its service life, and it should be run on the construction rather than assumed from a published figure.

The second is a continuity and resistance check during and after bending, on the finished circuit. A trace that has started to crack may still conduct at rest and open when the flex is moved, so the measurement should be made with the flex held in its worst case position. For a dynamic application, a test rig that flexes the circuit while monitoring resistance is the practical way to catch a marginal design before it ships.

Where the design cannot pass, the fixes are ordered by cost. Increasing the radius is usually free if the mechanical design allows it. Thinning the copper or the base film is a material change. Moving the conductor layers to the neutral axis is a stackup change, and if a via or a stiffener is inside the bend, relocating it is a layout change. Any of those is cheaper than discovering the problem in the field.

FAQ

• What is a safe bend radius for a static flex? Around ten times the total stack thickness is a common working rule, with the exact figure depending on the construction.
• How is that different for a dynamic flex? The radius is much larger, in the region of a hundred times the thickness for a long cycle life.
• Can a flex be folded sharply once? A crease is worse than a bend because the strain concentrates at a single line. Even a once-folded design benefits from a defined radius.
• Can a rigid flex have a bend radius like a flex? The flexible region behaves like a flex, but the transitions to the rigid sections need their own relief, which is why the stackup matters. Our guide to rigid flex stackup covers the transition.

Summary

Bend radius is the parameter that decides whether a flexible circuit survives its service. The strain in the copper depends on how far the conductor sits from the neutral axis and how tight the bend is, and the difference between a static fold and a repeated flexing is an order of magnitude in the acceptable radius.

The rules are consistent: keep the copper thin and near the middle of the stack, cross the bend line at right angles, keep vias, pads and stiffeners out of the bend region, and define the bend area on a drawing. Then verify the construction with a bend test rather than trusting a table. Where the layout and the mechanical envelope disagree, the mechanical design usually wins, which is why the radius should be agreed with the fabricator and the flex assembly partner before the circuit is routed, and confirmed against the capability of the specific construction being used.