Sunking PCB

Flex PCB Assembly Carriers: How Fixturing Decides Whether an FPC Can Run SMT

Design Guide2026-10-04Sunking Circuits Technical Team
Flex PCB Assembly Carriers: How Fixturing Decides Whether an FPC Can Run SMT

Why an SMT line cannot accept a limp board

Every station on an SMT line carries the same hidden assumption: the board does not move while it is being processed, its surface is flat, and that surface sits in a known relationship to the machine datum. Printing needs the board pressed flat against the stencil underside to release an even paste volume. The placement machine reads fiducials on the board and derives every pad coordinate from them. Reflow needs a rail or mesh belt to carry the board through the zones.

For a rigid board, all three assumptions are free — FR-4 holds itself. For a flex circuit, none of them hold. Polyimide base material is thin, limp, and not self-supporting; it curls on the print table, shifts when a nozzle touches it, and buckles when heated. So a flex circuit is not "unable to run SMT" — it is unable to run SMT until it is held.

The engineering consequence is that the real subject of flex assembly is not the flex circuit at all. It is the carrier (a removable rigid frame or pallet) that holds one or more FPCs flat so that the whole assembly looks like an ordinary rigid board to the equipment. What the carrier looks like determines whether this FPC can be assembled, how accurately, and whether the parts are still where you put them after reflow.

A carrier is not a stiffener, and the two get confused constantly

The most frequent confusion in drawing review is treating carriers and stiffeners as the same class of thing. Physically both "make the flex rigid," but they belong to completely different domains:

CarrierStiffener
OwnershipProcess tooling, does not shipPart of the product, permanently bonded
LifeReusable across one or many lotsSame life as the product
CoverageUsually supports the whole panelLocal only (connector area, screw holes, component zones)
Who defines itManufacturer, per line and board typeDesigner, on the drawing; manufacturer follows
Where it appearsProcess documents, tooling listStackup drawing, mechanical drawing, BOM

Once separated, the responsibility boundary becomes clear: the stiffener is a design input; the carrier is a manufacturing solution. But the designer cannot ignore carriers entirely, because a carrier needs something on the board to grip, and that something — tooling rails, locating holes, fiducials — can only come from the designer. That is the next section. Stiffener drawing conventions themselves are covered in Flex PCB Stiffener Drawing Review and are not repeated here.

Three basic fixturing mechanisms

Real carriers vary endlessly, but the holding mechanism reduces to three families, each with its own cost.

Mechanical clamping. A rigid outer frame (aluminium or a high-temperature engineering material) plus clamp bars presses the FPC edges or tooling rails against a pallet. It is structurally direct, repeats well, and consumes no consumables. The cost: the area under the clamp bars cannot carry components, and clamping only constrains where it touches — the centre of the board can still dome when heated. Suited to modest panel sizes with components concentrated centrally and adequate tooling rail around the perimeter.

Vacuum hold-down. The pallet carries an array of vacuum ports and negative pressure pulls the FPC flat. The whole surface is constrained and no layout area is sacrificed. The cost: it demands a continuous board surface — cutouts, windows, and large openings leak, and once it leaks the hold-down is gone. Vacuum channel sealing also degrades through the reflow heat, so vacuum is typically used for print and place, with a different scheme for the oven.

Reversible adhesion. The pallet uses a heat-tolerant, re-tackifiable coating or film and the FPC is bonded over its full area. It is insensitive to board outline, constrains everything, and can carry through reflow. The cost: the adhesive layer ages and picks up contamination, requiring scheduled cleaning and replacement — and the peel stress at the end is a real process risk. Peel it wrong and you tear a thin board or lift a soldered chip component with it.

In practice these are combined rather than chosen: vacuum for locating, mechanical clamping to hold, an adhesion pallet for the oven pass. The basis for combining them is board thickness, panel size, cutout area, component distribution, and whether both sides are populated — all board-level parameters that no generic recommendation can cover. Which is why the right move is not to specify a carrier type at design time, but to make the board fixturable at design time.

Four design inputs you must reserve

1) Tooling rail and grip area. An FPC outline usually hugs the mechanical envelope with no spare edge. A carrier needs grip area. So panelisation must reserve tooling rail and must state: how wide, whether the connection is continuous or broken by tabs, which side may be clamped, and whether separation is punched or laser cut. Tooling rail is area you are allowed to sacrifice; if the designer reserves none, the manufacturer has to encroach on the product area instead.

2) Datum strategy. The FPC and the carrier need a defined relative position, or the coordinates the placement machine computes will not land on the pads. The standard approach is locating holes in the tooling rail mating with pins on the carrier, plus optical fiducials on the board so the machine can correct for actual position. Locating holes solve "roughly right"; fiducials solve "precisely corrected" — neither substitutes for the other. Fiducial placement must also avoid any area a clamp bar or frame will cover; a covered fiducial does not exist.

3) Components and keep-out zones. Nothing — no component, no pad — may sit where a clamp bar presses, where a vacuum port opens, or where a carrier boss rises. This requires the manufacturer to feed back the carrier layout during process review, and the designer to mark keep-outs accordingly. It is a two-way process: if the designer never asks, the conflict surfaces just before production, when a revision costs the most.

4) A declared double-sided sequence. On a double-sided flex, the second reflow pass puts the first side's joints through the oven again, and that side now faces down or must be supported. Whether the board flips, which side provides the datum after flipping, and whether first-side components need adhesive staking — all of this belongs in the data package, not in a line operator's judgement call.

Thermal behaviour: the carrier's real difficulty is inside the oven

All three mechanisms flatten a board at room temperature. They diverge in reflow.

The core conflict is CTE mismatch. The polyimide base, the copper foil, the coverlay, and the carrier body (metal or engineering material) all have different expansion coefficients, so they grow by different amounts as temperature rises. If the carrier locks the FPC rigidly on all sides, that differential has nowhere to go, and the stress transfers to the weakest point available — usually copper in the bend area, the coverlay edge, or a joint that happens to straddle the constraint boundary.

So carrier design carries a pair of requirements that look contradictory: constrain tightly enough during placement that parts do not shift, yet leave the expansion a path during reflow. The common engineering answer is to make the constraint directional — fixed in one axis, allowed to slide in the other, rather than locked on four sides.

Two other things change at high temperature: vacuum channel sealing and retention degrade, and adhesive tack decays across repeated thermal cycles. Both mean a carrier that suits print and place does not necessarily suit the oven. Changing carriers between stations is normal process planning, not a failed design.

One more item that gets overlooked: carrier thermal mass. A heavy metal pallet absorbs heat and makes the FPC it holds lag the oven's programmed profile. That is not something you simply compensate for by extending time; the profile must be measured on the "carrier plus board" assembly, not inherited from the bare-board profile. Specific zone settings vary with carrier material and board type and require measurement during engineering confirmation; no values are given here.

Moisture and pre-bake: an extra step for thin material

Polyimide absorbs moisture, so pre-reflow bake is not optional for flex. The carrier-relevant question is when the bake happens. Bake the board already mounted on an adhesive-coated carrier and the coating ages first. Bake it and then leave it waiting too long before the line and the moisture comes back. The sensible flow puts the bake before carrier loading and controls the queue time, which needs process scheduling support rather than a single station's decision.

Failure modes and the station each one points back to

When assembly goes wrong, reasoning backwards from the symptom is faster than checking everything:

  • Whole-panel systematic offset — locating hole to pin fit, or how the carrier seats on the machine datum. Check fixturing, not placement accuracy.
  • Local offset, worse near the board edge — uneven constraint; the surface moved when the nozzle touched down. Check clamp bar layout and vacuum port distribution.
  • Uneven paste volume, thin toward the edges — the surface was not fully flat against the stencil underside. Check pallet flatness and whether constraint is missing in that zone.
  • Tombstoning or shift concentrated in one region — that region deformed during ramp-up. Check the thermal release design of the constraint.
  • Coverlay edge lifting, abnormal copper in bend areas — concentrated expansion stress. Check whether the carrier is locked on all four sides.
  • Board hard to de-panel after soldering, components missing after peel — adhesive ageing or wrong peel technique. Check carrier maintenance interval and the peel work instruction. Joint-level failure mechanisms are covered separately in PCBA Solder Joint Failure Causes.

The value of this list is not completeness. It is a habit: when flex assembly misbehaves, look at the fixturing before you touch machine parameters, because fixturing is the only variable on this chain that a human introduced.

Putting this back on the project timeline

Every engineering difficulty in flex assembly fixturing comes down to one sentence: it has to be discussed before the design freezes. Tooling rail, locating holes, keep-outs, and fiducial positions are geometric features on the board. Once Gerbers are released and tooling is cut, adding them is a revision.

So the correct cadence is: before releasing drawings, the designer puts "how is this board going to be held" to the manufacturer as an explicit question; the manufacturer returns carrier layout and keep-out zones; both sides confirm, then the design freezes. Bare-board stackup and manufacturing limits are covered in Flex PCB Construction and Manufacturing Limits, and bend-area routing rules in Flex PCB Conductor Routing in Bend Areas. This article addresses what happens after both are settled and the board reaches assembly. Board capability and process confirmation start from the Flex PCB product page and PCBA services.

FAQ

Does a flex circuit always need a carrier for assembly?

In practice, yes. Every SMT station assumes a fixed, flat board, and a flex circuit does not satisfy that on its own. Even a very small FPC needs some form of fixturing — possibly a simple pallet rather than elaborate tooling, but "unheld, straight onto the line" is not a real option.

Do we supply the carrier or does the factory?

Normally the manufacturer designs and builds it, because the carrier must match the clamping datum of specific equipment. What the designer supplies are the mating features on the board: tooling rail, locating holes, fiducials, keep-outs. A few long-running volume programmes have customer-specified or customer-supplied carriers; that needs to be declared early, because it constrains the board outline in return.

We already drew stiffeners. Do we still need a carrier?

Yes. A stiffener is part of the finished product and provides rigidity locally, usually over far less than the full panel. The stiffener's own placement accuracy also depends on the board being held during assembly. The two solve problems at different stages.

Can one carrier serve every station?

Not necessarily, and it need not. Printing, placement, and reflow ask different things of a carrier — printing wants flatness and stencil contact, placement wants locating accuracy, reflow wants temperature tolerance and expansion relief. Swapping carriers between stations is a common arrangement. If you want one set to cover the whole flow, raise it during engineering confirmation so the manufacturer can assess feasibility.

Does double-sided assembly add carrier requirements?

It does. When the second side is populated, the first side already carries components, so the carrier must provide clearance for them — typically a recessed or pocketed pallet — or it will crush them. The first side's joints also see a second reflow, so whether adhesive staking is needed must be stated in the data package. A double-sided plan should be declared at design time, because it directly limits component placement freedom.

Does panelisation affect the carrier?

Strongly. How many pieces one carrier holds, how pieces connect, and where they separate all determine the carrier's opening layout and locating hole positions. Panelisation and carrier strategy should be settled together — not panelisation first, carrier later.

Prototyping a flex or rigid-flex board?

Bend radius, coverlay and stiffener placement decide reliability. Tell us the bend cycles and assembly method and we will propose a construction with pricing.