Depaneling Strain Measurement: Gauge Placement, Limits and Data
Depaneling is the last mechanical operation a board sees, and it is the one most likely to crack a solder joint that passed every inspection before it. Strain measurement turns that risk into a number, which is the only way to compare one tool, one program or one operator against another.
Why Depaneling Strain Matters More Than Visual Inspection
A board that has been bent beyond its limit rarely shows the damage immediately. The crack starts at the interface between a solder joint and a pad, or inside a ceramic capacitor body, and it propagates only when the assembly is thermally cycled in service. Optical inspection after depaneling sees nothing, and the defect appears months later as a field failure.
The stress comes from the separation method. Routing applies a cutting force, V-score separation applies a bending moment, and hand breaking applies whatever the operator’s wrist produces. Only the third is variable between operators, but all three produce a flexure that the board transmits to every component near the separation line. Measuring that flexure is the only way to know whether the process is inside a limit that reliability testing has validated.
How a Strain Gauge Measures Board Flexure
A strain gauge is a foil resistor bonded to the board surface whose resistance changes as the surface stretches or compresses. Bonded in a rosette of three elements at 0, 45 and 90 degrees, it measures the strain in three directions at one location, which is what is needed because a board bends differently along and across the separation line.
The output is microstrain, one part per million of length. A board that stretches by 1 micron over a 1 mm gauge length reads 1000 microstrain. The gauges are small, typically 1 to 3 mm, so they measure a local condition, and the reading depends on where the gauge is placed relative to the bend. That makes placement the single most important decision in the measurement.
Gauge Placement: Where the Board Actually Bends
The gauges belong where the components are most at risk, not where they are convenient to bond. Those locations are next to the separation line, at the corner of a large component, on the body of a ceramic capacitor, and along the edge of a ball grid array package where the joint row nearest the cut sees the largest flexure. Place at least one gauge on each side of the separation line.
Placement relative to the bend axis also matters. A gauge oriented parallel to the cut measures the strain along the separation direction, while one oriented perpendicular measures the bending across it, and the perpendicular direction usually carries the larger signal. With a rosette both are captured, but the orientation of the rosette still has to be recorded, because a result quoted without the orientation cannot be compared with another test. The number of locations follows from the board. A small board with one separation line needs two or three gauges; a large panel with several cut outs and a mix of component types needs eight or more to find the worst case.
Microstrain Limits and the Standards Behind Them
The commonly used limit for a populated assembly is 500 microstrain at any gauge location, with a tighter limit of 300 microstrain for components that are known to be fragile, such as multilayer ceramic capacitors in a large case size and thin packages with a low stand off. These figures come from reliability testing and they are not universally applicable; where a product has its own qualification data, use it.
The limit applies to the highest reading, not to the average. A process that reads 200 microstrain everywhere except one gauge at 900 microstrain has failed, even though the average is comfortable. This is why the number of gauges matters: a measurement that misses the worst location reports a passing result for a failing process. Note the difference between the strain during separation and the residual strain afterwards. Most of the flexure is elastic and disappears when the board is released, so the measurement has to be made during the operation rather than after it. A gauge read after the board is free measures almost nothing. Where the board carries a heavy component near the cut, a limit derived from solder joint fatigue may be lower than the ceramic limit, so check both against the design rather than applying a single number.
Principal Strain: Converting Three Readings Into One Number
A rosette gives three strain values, and they are combined into the principal strains, which are the maximum and minimum strains at that point in the plane of the board. The larger of the two in absolute terms is the number that is compared against the limit, and it is what most specifications mean when they quote a figure without qualification.
The calculation requires the three readings and the orientation angles, and it is easily automated. The useful output is not only the peak value but the time trace, because the shape of the trace shows how the strain develops. A single spike at the moment of separation indicates a brittle release; a slow rise and fall indicates a controlled bend. The peak value sets compliance, and the shape guides the fix.
Test Setup: Bonding, Wiring and Sampling Rate
The gauge has to be bonded to the board surface with a thin, uniform adhesive layer, and the bond has to be cured before the measurement. A thick or partial bond absorbs part of the strain and reads low, which is a dangerous error because it makes a failing process look compliant. Cyanoacrylate is the common choice for a short test, with a full cure of at least an hour before use.
Wiring and sampling rate are the other practical limits. Lead wires stiffen the board locally and change the measurement unless they are routed away from the bend and taped down; a wire that pulls on the gauge produces a reading that comes from the wire, not the board. The sampling rate has to be high enough to capture the event, which for a fast separation means at least 1 kHz, and for a slow press driven cut 100 Hz may be adequate. Record the gauge type, the adhesive, the amplifier settings, the excitation voltage, the sampling rate and the channel to location mapping. A measurement without that metadata cannot be reproduced, and an unreproducible measurement settles no argument.
Running the Measurement on a Depaneling Process
Run the measurement on production boards, with the tooling and the program that production will use, and with the same operator technique. Run several boards rather than one, because the peak varies with panel placement, with tool wear and with the position on the panel. Five boards at minimum, ten where the process is marginal.
Measure the separation of a single part at a time so that the source of the strain is unambiguous. Where a panel is separated in a sequence, the release of the first part changes the stiffness of the panel for the second, so the last part separated may see a different strain from the first. Record the sequence and the position of each measurement on the panel. Where the tooling includes a fixture that holds the board, check that the fixture itself does not pre load the board. A clamp that is tightened before the cut applies a static strain that adds to the separation strain, and the total is what the joints experience.
Interpreting Results and Finding the Cause
A reading above the limit has a short list of causes. The tooling does not support the board close enough to the cut, so the board acts as a cantilever. The separation is done too quickly or in one pass when two would be gentler. The operator is applying a bending moment rather than a controlled separation. The board is thinner or more warped than the sample that was used to set the process.
Work through them in that order, because the tooling support and the separation speed are the two that are cheapest to change. Support closer to the cut reduces the unsupported span and therefore the strain directly; a slower, controlled separation reduces the dynamic component.. Support closer to the cut reduces the unsupported span and therefore the strain directly, and a slower controlled separation reduces the dynamic component. The same warpage conditions that concentrate stress at depaneling are the ones described in our warpage control notes, and the same panel tooling questions are covered in the panel tooling design review, so a fix that addresses both usually holds.
Records, Re qualification and Tooling Changes
Keep the setup, the traces, the peak values and the tooling identification with each measurement. Requalify when the panel design changes, when the tooling is replaced or modified, when the separation method changes, and at a defined interval so that gradual wear is detected. Compare a new trace with the previous one rather than only against the limit, because a slow rise that has not yet reached the limit is the earliest warning available.
Where the measured strain is well below the limit, keep the number rather than discarding the test. A baseline of 150 microstrain gives room to evaluate a tooling change, and it makes the eventual requalification a comparison rather than a fresh judgement. Record the microstrain peak per gauge location, so that a change in the pattern across the panel is visible as well as a change in the maximum.
FAQ
What microstrain limit should a depaneling process meet? A common limit for a populated assembly is 500 microstrain at any location, tightened to about 300 microstrain near fragile components such as large multilayer ceramic capacitors. Apply the limit to the highest reading, not the average, and use product specific qualification data where it exists.
Where should strain gauges be placed for a depaneling test? Next to the separation line on both sides, at the corners of large components, and on the body of ceramic capacitors that sit close to the cut. Orient a rosette so that one element is perpendicular to the cut line, and record the orientation, because a result without it cannot be compared.
Can depaneling strain be measured after the board is separated? No. Almost all of the flexure is elastic and is released when the board is freed, so a gauge read afterwards measures nearly nothing. The measurement has to be made during the separation itself, with a sampling rate high enough to capture the event.
