Component Placement Accuracy in SMT
Placement accuracy is the difference between where a component was intended to be and where it actually landed. It is quoted as a machine specification, but the number that reaches the board is the sum of several contributions, and only some of them belong to the placement head. Understanding which contribution dominates is what makes it possible to improve a process rather than to replace a machine.
What the Specification Means
A machine is quoted with a placement accuracy and a repeatability, and the two are different. Accuracy is the distance from the target, and repeatability is the spread of a series of placements. A machine can be repeatable and still placed slightly off target, in which case the correction is a calibration rather than a repair.
The figure is usually quoted in microns at a defined speed, with a defined component and often with vision alignment enabled. Removing any of those conditions changes the number, so a machine that meets its specification in a demonstration may not meet it in production with a different feeder and a heavier component.
The practical question is not the machine number but the fraction of the pad that the placement consumes. A 40 micron error on a 0.5 mm pad is eight percent of the pad, which self alignment can correct. The same error on a 0.2 mm pad is twenty percent, which it may not.
Vision Alignment
Vision alignment measures the component and the board separately and applies the difference. The component is imaged on the way to the board, and the board is imaged at its fiducials, and the machine then rotates and offsets the placement to match.
The accuracy of the correction depends on the fiducials. Global fiducials at the panel corners handle the board level offset, and local fiducials near a fine pitch device handle the local distortion that appears after reflow and lamination. A design that provides no local fiducials leaves the machine correcting to an average that may not apply where the fine pitch part is.
The vision system needs contrast to see the feature. A fiducial that is covered by solder mask, that is the same colour as the surrounding copper or that has been partially plated will be found inconsistently, and the error appears as an occasional placement offset rather than as a systematic one.
Nozzle and Component
The nozzle picks the component and holds it, so the nozzle condition sets part of the placement error. A worn or contaminated nozzle allows the component to shift, and a nozzle that is too large for the part gives a placement that wanders within the tip.
The component itself contributes. A body that is at the edge of its dimensional tolerance, a ball array that is offset within the package or a lead frame that is skewed will all be placed where the machine measured the body rather than where the design intended the terminals to be.
Nozzle selection is therefore part of the process rather than a fixed setting. Each package type should have a defined nozzle, and the choice should be recorded so that a change in placement error can be traced to a nozzle change rather than to a machine fault.
Feeder and Material Effects
A feeder presents the component at a repeatable position or it does not. A tape that has been spliced, a reel that is wound unevenly, a pocket that has been opened and a feeder that has not been calibrated all introduce an offset that the vision system may or may not correct.
Where the machine picks blind, without imaging the component, the feeder accuracy appears directly in the placement. Where the machine images the component on the fly, the feeder error is corrected but the cycle time increases, and the choice between the two is a production decision.
Material condition matters as well. A component that has absorbed moisture, a tape whose cover has been left open and a part that sticks in its pocket all change the pick position. These effects appear after a pause in production rather than at the start of a run, which makes them easy to misattribute.
Self Alignment During Reflow
Molten solder pulls a component toward the centre of its pads. The force is proportional to the imbalance in the wetting area, and it can correct a placement error that is a meaningful fraction of the pad width. This is why an 0402 part can be placed with a modest accuracy and still form a good joint.
The force has limits. A component that is displaced beyond the pad, a pad that is contaminated on one side, a paste deposit that is missing or a thermal imbalance that melts one joint before the other all prevent the correction and can produce a tombstone instead.
The lesson is that placement accuracy and paste deposition are a single system. Improving the placer while the printing process has a volume variation does not improve the outcome, and the reverse is also true. Both have to be in control for the self alignment to do its work.
Measurement and Control
Placement accuracy is measured with a glass board or a dedicated test component, and the result is a distribution rather than a single number. The distribution should be recorded at intervals and compared with the specification and with the previous measurements.
In production, the useful measurement is the pad coverage from the automated optical inspection system and the position data that the placer itself logs. A gradual change in the position distribution is the earliest indication of a worn nozzle, a drifting feeder or a fiducial that is being found less reliably.
Where an error is confirmed, the analysis should separate the contributions. A machine calibration, a nozzle change and a feeder adjustment are three different actions, and applying all three at once makes it impossible to know which one mattered.
Practical Rules
Match the nozzle to the package, provide local fiducials beside fine pitch devices, and keep the feeders calibrated and the tapes intact. Measure the position distribution rather than a single sample.
Record the nozzle, the feeder and the fiducial data with the build records and the printing defect history, and review the AOI data and the component shift causes whenever the placement distribution moves.
Process Control and Verification
On a design of this kind, vision alignment is the item that decides how the rest of the board is arranged. The process window is set by the narrowest step in the flow, so an improvement anywhere else shows up as margin rather than as yield until that step is addressed. A short note on the drawing about handling, storage or packaging is often worth more than an extra decimal place on a tolerance.
FAQ
Is accuracy the same as repeatability? No. Accuracy is the distance from the target and repeatability is the spread of repeated placements. A repeatable machine that sits off target needs a calibration, not a repair.
How much placement error can reflow correct? Enough to be useful, but only while the component sits on both pads with balanced wetting. Beyond the pad edge, or with a missing deposit, the force cannot recover the placement.
Why do local fiducials matter? They correct the local distortion of the board after lamination and reflow. Without them the machine corrects to an average that may not apply where a fine pitch device sits.
