Depanelization Stress: Strain Gauge Measurement and Limits
A multilayer ceramic capacitor that cracks during depanelization usually passes every test performed at the time, and fails months later as a short or as an open. The way to see the damage before it happens is to measure the strain the board experiences, and the strain gauge is the instrument that does it.
Why the Measurement Is Needed
Ceramic components are brittle and their terminations are rigid, so a small bend in the board concentrates stress at the corner of the termination. The crack that follows may be a hairline that does not cross the electrodes, which is why the part passes an electrical test and fails later.
The bending that causes it is easy to produce: snapping a scored panel by hand, driving a screw into a board that is not supported, seating a connector with a press, or simply handling a large panel by one corner. Measuring the strain turns each of those from an opinion into a number.
What the Standards Ask For
The usual reference is a limit on the strain measured at the component location, with a figure in the region of 500 microstrain for boards with leaded solder and a tighter figure, around 300 microstrain, where lead free joints and brittle components are involved. The figure applies to the surface of the board at the component rather than to the panel as a whole.
Some standards also limit the strain rate, because a fast bend damages a component more than a slow one of the same magnitude. The rate is read from the same trace, and it is the reason the measurement system has to sample quickly enough to see the event at all.
Strain Gauge Placement and Orientation
A single axis gauge measures strain in one direction, and a rosette of three gauges resolves the direction and the magnitude of the principal strain. Where the loading direction is known, a single gauge oriented along it is enough and is easier to bond in a tight space.
The gauge is placed on the board surface as close to the component of interest as the layout permits, with the axis aligned to the direction in which the board bends. Placement on the opposite side of the board describes a different surface and a different strain, which is a common source of a result that cannot be reproduced.
Measurement Setup
The gauge is wired into a bridge circuit and the signal is amplified and recorded, with a sampling rate high enough to capture the event; a depanelization snap lasts a few milliseconds, so a rate in the kilohertz range is normal. The leads are routed so that they do not carry the bending load themselves.
The system is calibrated against a known signal before the run, and the gauge factor and the bridge configuration are recorded with the result. A reading without those two items cannot be converted into strain, and it cannot be compared with a reading taken on another line.
Which Operations to Measure
The operations worth measuring are depanelization, screw and standoff mounting, connector insertion and extraction, board handling on a conveyor and any fixture that presses on the assembly. Each has a characteristic strain profile and a different peak.
The measurements are taken on a board that carries dummy components in the positions that matter, because the gauge cannot be bonded where a real component sits. The dummy has the same mass and the same footprint, so the loading case is preserved even though the parts are expendable.
Reading the Result
The trace is read for the peak strain, for the rate of change and for the duration above the limit, and the three together describe the severity of the event. A short peak at 400 microstrain is a different case from a plateau at 300 for fifty milliseconds.
The location of the gauge is part of the result, and a limit that is written for one component position does not automatically apply at another. Where a board carries several brittle parts, the measurements are repeated at each of them or at the worst case that the layout presents.
Reducing Strain in Production
The countermeasures are support under the board at the point of load, a machine that cuts rather than snaps the panel, a sharper blade, and a fixture that spreads the force over an area instead of a point. A depanelization machine that supports both halves of the panel during the cut reduces strain considerably compared with a hand snap.
Screw mounting is the same problem in a different form, because a screw driven into an unsupported board pulls the surface down and then releases it. Supporting the underside at the screw position, and using a torque driver rather than a hand tool, changes the measured strain more than any other single measure.
Verification and Corrective Action
The verification is a measurement on the first article with the production tooling and the production method, and it is repeated after any change to the tool, the method or the support. A result above the limit is a process change rather than a training note.
Where a corrective action is taken, the measurement is repeated to show that the strain has fallen, and the number is recorded rather than the fact that the operation was changed. Our warpage control notes describe the board’s own flatness, which is part of the same stress budget.
Records and Process Control
The record carries the gauge position and orientation, the gauge factor, the sampling rate, the event that was measured and the peak result. Those items make the measurement repeatable, which is the only way a limit can be enforced over time.
Where the product is built on several lines, the measurement is repeated on each of them, because the tooling and the handling differ. Our board failure notes describe how a cracked component is recognised later, and the strain record is what shows the operation that caused it.
Points to Confirm at First Article
Consumables have a life measured in cycles, and the replacement point should come from the measurement rather than from a failure. The sequence of operations is part of the specification, because a different order produces a different result from the same steps.
Where the process window is narrow, the measurement resolution has to be better than the window, or the data cannot distinguish a good part from a marginal one.
Related reading: our fabrication notes, board quality and design release notes cover the same ground.
FAQ
Can a cracked capacitor be detected after assembly? Sometimes, by acoustic microscopy or by a bend test on a sample, and not by electrical test. That is why the strain is measured during the operation rather than the damage being looked for afterwards.
Does the limit depend on the component? It does. A large ceramic capacitor in a rigid package is more sensitive than a small one, and the limit is set from the most sensitive part on the board rather than from an average.
Why does a slow bend matter less than a fast one? Because the material has time to distribute the stress, which is why the standard limits the strain rate as well as the magnitude. The measurement system has to sample fast enough to see the difference.
