Issue link: https://iconnect007.uberflip.com/i/1547450
60 SMT007 MAGAZINE I NOVEMBER 2026 and stop, cruder force-and-timer setups push until the clock runs out, and the board absorbs whatever the plastic does not. Board flexure is the failure mode that matters, and its victims are well documented. Multilayer ceramic capacitors crack under board strain, and the cracks are often invisible at test, opening up later in the field under thermal cycling. 5 Pad cratering (fracture in the laminate resin under a BGA pad) follows the same script. Neither failure announces itself at the staking station. Both trace back to strain, which is why IPC/JEDEC-9704A exists to measure it 6 and IPC/JEDEC-9702 exists to characterize how much board-level interconnects can take. 7 Component suppliers commonly cap allowable board strain at a few hundred microstrain; a thin board bridging an unsupported span in a staking fixture can blow through that with room to spare. 8 The layout defenses are cheap: • Treat every stake site the way you already treat screw holes and depaneling edges. Keep large-case MLCCs and BGA corners clear of them, and orient nearby chip capacitors so their long axis runs parallel to the expected bend line rather than across it. • Ask whether the fixture supports the board near each stake site. The boss shoulder itself is the first support; a fixture that backs up the board directly under the stake turns bending into compression, which FR-4 handles far more gracefully. • Prefer equipment that forms multiple stakes in one balanced stroke and verifies displacement. A press with position feedback—good systems resolve 0.1 mm—confirms each stud is present and stops the head at the programmed height, cycle after cycle, instead of leaning on the board until a timer expires. If strain-sensitive components must be within reach of a stake site, indicate this on the assembly drawing and request a strain-gaged first article per IPC/JEDEC- 9704A. It is a one-day test that has saved more than one program from a very expensive education. The Plastic Sets the Temperature, Not the Board A reasonable worry: The tool runs at 200–320°C, and standard FR-4 has a Tg between roughly 130–180°C. 9 Is the board being cooked? Figure 4: Where staking force goes. Left: An unsupported span turns tool force into board bending, and strain cracks nearby ceramic capacitors. Right: Support directly under the stake site turns the same force into compression, which the laminate tolerates easily.