Bow and Twist: 5 Rules for Measuring PCB Flatness
Bow and twist are the two ways a printed circuit board fails to be flat, and they are measured separately because they come from different causes. Bow is a curvature across the board, so the whole panel bends like a shallow dish. Twist is a corner out of plane, so the board rocks on three corners and the fourth lifts. Both are expressed as a percentage of a board dimension and both are limited by the workmanship standard.
Board flatness matters because assembly equipment assumes a flat part. A board that rocks on the printer table prints paste with a varying gap, a board that bows in the mounter can miss a placement, and a board that twists during reflow can lift a corner out of a BGA site. Flatness is a fabrication property that decides assembly yield.
Bow, Twist and Warpage: Three Different Things
Bow and twist are the measurements defined by the standard. Warpage is the general term that covers both, plus the local distortion that appears when a board is heated. A board can pass a board flatness check at goods-in and still warp during reflow, because the measurement was taken at room temperature on a laminate that still holds stress.
The distinction matters when a supplier and a customer disagree. A board measured at 20 degrees Celsius and the same board measured on a hot plate are two different test conditions, and both are legitimate. The specification has to say which condition applies, and the assembly process cares about the hot one. A value for bow and twist that does not name its condition is not a specification.
How the IPC Limit Is Calculated
The limit is a percentage rather than an absolute dimension, which is why a large board is allowed to deflect further in millimetres than a small one. Bow is calculated from the maximum deviation from a reference plane over the length of the board, and the result is expressed as a percentage of that length. Twist uses the height of the free corner after the other three are held down.
Common acceptance values sit below one percent for boards that will be populated with surface mount parts, and are looser for boards that will only be through-hole assembled. The exact figure depends on the acceptance class and the product, and it should be quoted on the fabrication drawing rather than assumed. The measurement method and the calculation are defined in the IPC test methods.
The Measurement Fixture and the Test Method
The bow and twist measurement is simple, which is exactly why it is often done badly. A granite surface plate or a ground plate is used as the reference, with the board resting on it or with three corners restrained for the twist measurement. The measurement fixture should be flat to a small fraction of the tolerance you are trying to measure, or the fixture becomes part of the result.
Two details decide whether bow and twist readings are comparable. The board must be at equilibrium, which means it should be measured after it has been left flat for a defined period rather than straight out of a rack, and it must not be pressed by hand or clamped by the fixture. The result should be recorded with the temperature and the humidity of the room, because a laminate absorbs moisture and moves with it.
Why Lamination Creates Laminate Stress
A multilayer board is a stack of materials with different expansion rates, bonded at a temperature above the glass transition. As it cools, each layer tries to contract by a different amount and the resin locks the mismatch in place. That stored energy is the laminate stress that appears later as bow, twist or reflow warpage.
The press cycle decides how much laminate stress survives. A fast cooling rate freezes the board while it is still hot and soft, and the stress is worse than a slow, controlled cool that allows the stack to relax. Cooling under pressure to below the glass transition before the press opens is one of the most effective single changes. Our notes on press cycle control describe the sequence.
Copper Balance and Layer Symmetry
A stack with copper on one side of the core and none on the other will bow, because the two sides shrink by different amounts. The same applies to a panel with a large plane on one layer and a sparse pattern on the mirror layer. Symmetry in copper distribution and in prepreg placement is the first design rule for flatness.
Where a design cannot be symmetric, the fabricator can compensate to a degree by choosing prepreg thicknesses that balance the build, but the room for that is limited. Resin content and flow also play a part, because a layer that flows more ends up thinner than the layer above it. Our resin content notes explain the variables.
Reflow Warpage and Solder Defects
When the board goes through the reflow oven, it crosses the glass transition temperature and the laminate softens. Stored stress is released, the board takes up a new shape, and it may not return to its original one when it cools. The result is a board that was flat at goods-in and is not flat after the first pass.
The defects that follow are subtle. A BGA site that warps during the peak can lift a ball out of contact or leave it in contact with a joint that is too thin to be reliable. A thin board with a large connector can bow enough to make paste printing uneven on the second side. This is why reflow warpage is measured on a sample that has been processed, not on incoming material.
Twisted Boards in Assembly Equipment
A twisted board is a handling problem before it is a soldering problem. It does not sit in the magazine, it does not sit flat on the conveyor, and it may not sit on the printer table at all. Where the twist is local, a board support with adjustable pins can compensate, but the fix is expensive and it hides the real cause.
The practical check is to place a sample from each lot on a flat surface and look for the rocking corner before the boards are loaded. That takes seconds and it catches the material that will otherwise stop the line. Equipment that clamps a twisted board flat will only do so until the clamps are released, and the solder joints are made while the board is held.
Fixing Flatness Problems at the Source
Work in order of cost. First check storage, because panels stored leaning against a wall or stacked unevenly will take a set, and the remedy is a rack. Then check the press cooling profile and the press plate condition. Then review copper balance and layer symmetry with the designer, because that change is free if it happens before the panels are made.
Changing laminate is the last resort, but it is the right one where the board is thin, the temperature is high and the design is inherently asymmetric. A laminate with a higher glass transition and a lower Z-axis expansion behaves better through reflow, and the difference is measurable. Our notes on warpage control follow the same sequence.
What to State on the Drawing
The drawing should give the maximum bow and twist as a percentage, name the measurement method, and state whether the limit applies before or after assembly. Where the product will be reflowed, a separate limit for reflow warpage is worth having, even if it is only checked on a sample per lot.
It should also state the panel size and the grain direction if the design is sensitive, because twist behavior differs between the length and the width of the panel. A short note here prevents a long argument later, and it costs nothing at the time the stack-up is released. The glass transition of the chosen material belongs in the same note.
FAQ
Is bow and twist measured before or after assembly? The bow and twist limit applies to the board as delivered, which is before assembly. Many products need a second limit after the first reflow, because the heat releases stress and the board can move. Both conditions should be stated if the product is sensitive.
Can a board that fails the bow and twist limit be straightened? A board can be pressed flat and baked, and it will hold its shape for a while, but the stress remains and returns on the next thermal cycle. Straightening is a temporary measure that hides a stack-up or press problem, and it should not be used as a production fix.
Does a thicker board warp less? Generally yes, because stiffness rises with thickness, but the stress does not disappear. A thick board with a badly unbalanced stack can still twist, and a thin board with a symmetrical stack can be surprisingly flat. Balance matters more for board flatness than thickness does.
