Why does a static bend need a different design from a dynamic one?

A static bend is formed once during assembly and then held, so the copper only has to survive that single deformation; a radius of about ten times the flex thickness is a working starting point. A dynamic bend moves continuously, so the copper is loaded on every cycle and the radius has to be much larger, the copper thinner and the construction simpler. The only way to know the cycle life is to bend-test the actual construction.

What is the difference between ED and rolled annealed copper?

Electrodeposited (ED) copper is plated and has a columnar grain structure that cracks early under repeated bending. Rolled annealed (RA) copper is rolled, has elongated grains and is far more ductile, which is why it is preferred for dynamic flex. If RA copper is used, the grain direction is called out on the fabrication drawing so the grain runs along the bend direction.

Why does a flex need a better fabrication drawing than a rigid board?

Because more of the part is decided outside the Gerber. Coverlay openings, stiffener material and thickness, which side each stiffener is on, the bend line, copper grain direction and the bend radius all have to be stated. On a rigid board a missing note is usually cosmetic; on a flex it changes how the part behaves, and the fabricator is left guessing.

Flex Circuit Design: The Rule Set

1. Decide the motion first

Everything else follows from one question: does the bend happen once, or continuously?

Static (form-once): the flex is shaped during assembly and fixed. The design has to survive installation, not service motion.
Dynamic: the flex moves in use — a hinge, a sliding mechanism, a moving head. The copper sees millions of cycles.

Static and dynamic versions of the same circuit are different products: different copper, different radius, sometimes a different construction. Deciding this late is the most expensive mistake in flex design.

2. Choose the construction

Single-sided flex: one copper layer, simplest and cheapest, good for a static bend.
Double-sided flex: two copper layers, allows crossing conductors and controlled impedance — and needs plated through-holes, which must stay out of the bend.
Multilayer flex: up to 12 layers here, for dense circuits that still have to move.
Rigid-flex: when both sides of the board carry components. As a rule of thumb, one-sided SMT suits an all-flex design; two-sided SMT usually means rigid-flex, so the components sit on rigid sections.

3. Copper: type, weight and grain

Copper is where a flex fails first.

Rolled annealed (RA) copper is ductile and preferred for anything that bends repeatedly. Electrodeposited (ED) copper is plated, harder and more brittle; it is acceptable for static bends and cheaper.
Weight: 0.5–2 oz is the plant range. In a dynamic bend, thinner is better — 1 oz sees far less strain than 2 oz at the same radius.
Grain direction: call it out, and orient it along the bend so the grain runs with the bend line, not across it.
Solid copper planes in the bend are cross-hatched where possible, which lets the film flex instead of resisting.

4. Rules for the bend area

No plated through-holes, no stiffener and no coverlay window inside the bend.
Conductors cross the bend at right angles, never running along it.
Teardrops and gentle tapers at pad transitions; no sharp corners or acute junctions.
Copper balanced around the neutral axis: layers directly opposite each other put one of them in tension on every cycle.
The bend radius is agreed in DFM, with the fold angle and the cycle count.

5. Coverlay, not solder mask

A flex is covered with a laminated film rather than a printed mask. Two consequences:

Design it as its own layer, the way a mask layer is drawn, with openings that allow for adhesive squeeze and cut tolerance.
Make the openings generous. A tight window leaves adhesive on the pad; too close a spacing exposes the neighbouring trace and bridges during assembly.

6. Stiffeners where the part must not move

Connector tails, component areas and anything clamped or screwed down need a stiffener. Practical rules:

Place it on the opposite side from the components it supports, unless a through-hole connector needs access from the other side.
Choose the material for the job: PI for connector tails and ZIF contacts, PET for cost-sensitive areas, FR-4 for a genuinely rigid section, stainless steel where thin and stiff is required.
Specify thickness as a stack — stiffener plus adhesive plus flex — because that is what the mating connector sees.

7. The fabrication drawing is part of the design

State the base material and thickness, copper type and weight, coverlay thickness and openings, stiffener material, thickness and side, bend line and radius, grain direction, impedance targets and the acceptance class.

A flex has more moving variables than a rigid board, and the fabricator who has to assume any of them will assume wrong at least once.

8. Cost and lead time follow the process

Flex is slower to build than rigid: thin film handling, different drilling and plating chemistry, and hand work on coverlay and stiffeners. A two-layer flex typically needs a few working days, and higher layer counts or rigid-flex sections take longer because the stack-up is planned per job rather than from a standard recipe.

That is also why flex costs more per square metre than FR-4 — the material, the handling and the yield all point the same way.

9. Test the bend, not the datasheet

Bend-radius tables are a starting point. The real number comes from bending the actual construction at the actual radius, for the actual cycle count — which is why the prototype should use the same material as production.

If the design cannot be qualified by analysis, qualify it by test: a bend fixture and a cycle count are cheaper than a field return.

The build limits are on the flexible PCB page and the capability page; the rigid-flex-specific rules are in the rigid-flex note, and the failure modes in the flex issue note.