When a PCBA shows cracked BGA joints or MLCC cracks at test or in the field, the usual suspects are flexure steps in assembly: depaneling, the ICT bed-of-nails press, press-fit connectors, screw mounting and enclosure fitting. Each of these bends the board briefly. Afterwards there is no visible warpage, but the joints and ceramic parts have already taken the strain.
Strain gauge testing turns "is this step too harsh?" into data: how much strain each step produces near a component, and how fast. With that data you can judge against a limit, change the fixture or process, and re-measure to confirm.
Where stress comes from and how to specify dimensional stability on a drawing is covered in PCB mechanical stress and dimensional stability. This article covers how to measure strain, where limits come from and what to change afterwards.
1. Why measure strain instead of warpage
Warpage is a static shape: bow and twist with the board resting on a surface plate. Process damage mostly comes from dynamic flexure: the moment the ICT press plate comes down, the cutter passing through a tab, the pins going into a press-fit connector. The board springs back afterwards, so a later warpage check shows nothing.
Solder joints and MLCCs respond to two quantities:
- Strain magnitude: how much the board surface stretches or compresses near the part, in microstrain (µε).
- Strain rate: how fast the strain changes. At the same strain, faster loading makes brittle joint fracture more likely.
So the measurement has to run during the process step, not as an inspection afterwards.
2. Method: three-element rosettes and principal strain
The strain direction on the board surface is not known in advance, and a single-element gauge only reads one direction. The usual approach is a three-element rosette (0°/45°/90°). From the three readings you calculate:
- Principal strain: maximum and minimum principal strain, the largest stretch and compression at that point.
- Diagonal strain: shows whether the board is twisting.
- Strain rate: the slope of principal strain over time.
The data acquisition system needs a sampling rate high enough to catch millisecond-scale peaks from depaneling and press-fit. IPC/JEDEC-9704A, the strain gauge testing guideline for printed circuit assemblies, describes rosette selection, gauge attachment, acquisition and data reduction, and is a sound basis for a test plan.
3. Where to place the gauges
A gauge measures surface strain at its own location, so placement decides what the data represents:
- Corners of critical parts: near the corners of large BGAs and fine-pitch CSPs, close to the outer ball rows.
- Near large MLCCs: especially capacitors close to board edges, tabs and mounting holes.
- Near load points: around depanel tabs, press-fit connectors, screw holes and ICT push-rod contact points.
- Same side first: usually on the component side. Where there is no room, use the matching spot on the opposite side and note it in the report.
Mark every gauge position and number on the board drawing and photograph it, so re-tests use the same locations.
4. Step-by-step coverage
A PCBA passes several loading steps between the line and the enclosure. The test plan should cover each of them, not only the one that looks worst:
| Step | How load is applied | What to focus on |
|---|---|---|
| Depaneling (V-score cutter, router, punch) | Local bending as the tab breaks | Parts near tabs; comparing depanel methods |
| ICT bed-of-nails | Press plate force and probe reaction | Spans between fixture support posts |
| Press-fit connectors | Insertion force transferred to the board | Parts around the connector; support fixture fit |
| Screw mounting / enclosure | Forced deflection on uneven mounting faces | Around screw holes; tightening sequence |
| Handling, cable insertion | One-handed holding, plugging cables | Board edges and connector areas |
Measure several boards per step and record the equipment, fixture ID and settings. Without that, before-and-after data cannot be compared.
5. Where limits come from
The most common question is "how much is too much?" There is no single number. The limit depends on:
- The component: ball size and pitch, package size, MLCC size and termination design.
- Solder and finish: lead-free and tin-lead joints tolerate different strain.
- Board thickness: the same strain means different curvature on thick and thin boards.
- Strain rate: limits usually vary with strain rate, with lower allowable values at faster loading.
Limits typically come from three places: component supplier application notes, customer or industry internal specifications, or in-house tests using bend characterization methods such as IPC-9702. The test report should state the limit source and revision. Where no usable limit exists, start with a baseline: record current strain per step, compare it with failure analysis results, and build an internal control line from there.
6. What to change when strain is too high
When results exceed the limit, start with the process, because it does not require a board change:
- ICT fixture: add support posts under push-rod and dense probe areas, balance top and bottom forces and shorten support spans.
- Depaneling: compare router, V-score cutter and punch on the same gauge positions. Sensitive parts near tabs should get the method with the lowest strain.
- Press-fit: use a support fixture that matches the board surface, with support directly under every insertion point.
- Screw mounting / enclosure: check mounting face flatness and specify tightening sequence and torque.
If the process cannot be changed enough, go back to the design:
- Keep sensitive parts away from tabs, screw holes and press-fit zones.
- Change the number, position and type of breakaway tabs, see PCB panelization design.
- Place MLCCs with their long side parallel to the break line or main bending line, which reduces tension between the two terminations.
- Evaluate a thicker board or a different support structure.
Re-measure at the same gauge positions after each change and confirm the improvement with data.
7. What the PCB fabricator can contribute
Strain testing happens mainly at PCBA and box build, but the results often lead back to PCB design and fabrication inputs:
- Panelization: tab position, V-score depth and residual web, and the number of mouse-bite holes directly affect depanel strain. State them on the panel drawing at order.
- Thickness and tolerance: thickness sets stiffness. Confirm thickness and tolerance against the current tolerance capabilities.
- Press-fit holes: hole size and plating set insertion force, see press-fit connector hole manufacturing inputs.
- Gauge sites: boards planned for strain testing can reserve flat, mask-free gauge areas next to critical parts.
How SKPCB handles it
Every SKPCB PCB order is 100% electrically tested. On the PCBA side, SKPCB runs SPI, inline 2D AOI, post-reflow 3D AOI, X-ray, ICT and FCT, see PCBA services. Automotive orders are built at the IATF 16949:2016 certified site, see quality assurance.
If a project has strain test requirements, or panelization, thickness and press-fit holes need adjusting based on strain results, attach the test plan or report to your quote request and engineering will confirm the fabrication inputs that can be met.
FAQ
Q: Is it enough to test only the ICT step?
A: No. Depaneling, press-fit, screw mounting and enclosure fitting all bend the board, and a failure can come from any of them. Measure each step, then decide which one is the main source.
Q: Is there a standard strain limit?
A: There is no single number. IPC/JEDEC-9704A defines the test method. The limit depends on component, solder, board thickness and strain rate, and should come from supplier guidance or customer specifications, with the source stated in the report.
Q: Should gauges go on the component side or the opposite side?
A: Usually on the component side near the part corners. Where there is no room, use the matching position on the opposite side, note it in the report and keep it the same for re-tests.
Q: Does a high result always mean a board redesign?
A: Not necessarily. In most cases you can change the process first: more ICT support posts, a lower-strain depanel method, a press-fit support fixture, a defined tightening sequence. Change layout or panelization only when the process cannot be changed enough.
Q: What can the PCB fabricator do to reduce assembly strain?
A: Mainly build to the drawing: tab layout, V-score residual web, board thickness and press-fit hole requirements. These need to be stated explicitly at order.