Solder paste inspection is the only check on an SMT line that looks at the joint before it exists. Every other station judges a result that has already been made. By the time AOI sees a bridge, the paste that caused it has been reflowed and the board has been through a full thermal cycle.
That is why a line without paste inspection spends its rework budget on defects it could have stopped at the printer. The cost of a printed board with low paste is a wash and a reprint. The cost of the same board after reflow is a rework ticket, a heated pad and a unit with a history.
What goes wrong at the printer
Paste defects come from four places, and all four are visible to a 3D scanner before placement: aperture fill, stencil release, paste condition and board support.
An aperture that fills short gives a joint that opens after reflow, and on a 01005 part at 0.4 by 0.2 mm the margin is so small that a small volume loss is the whole joint. Paste that smears between apertures bridges at reflow, which is the classic cause of a short on a 0.3 mm pitch array. Paste that has been open too long changes viscosity, and the deposit slumps before placement. A board that is not supported flat holds the stencil off the surface at one corner, and the deposits at that corner are thick while the rest are right.
What the four measurements predict
| Measurement | What it tells you | Defect it prevents |
|---|---|---|
| Volume against aperture | Whether the aperture released the paste it was cut for | Open joint, weak fillet, head in pillow on a BGA |
| Area and coverage | Whether the deposit sits on the pad or partly off it | Tombstone, drift, solder ball beside the joint |
| Height | Whether the stencil sat flat and released cleanly | Bridging on fine pitch, uneven standoff under a large part |
| Offset from pad centre | Whether the printer is aligned to the board | Partial contact on one side, weak joint on a chip part |
Shape is the fifth reading and the one that matters most on a thermal pad: a large aperture with good volume but a hollow centre prints a ring of paste and leaves the middle thin, and the voids under that pad appear after reflow, not at print.
Why a limit set once does not hold
Paste volume limits are not a station setting, they belong to the build. A stencil cut for a 0.5 mm pitch connector and a stencil cut for a 01005 part do not share a window, and a limit copied from a previous product will either pass bad boards or stop good ones.
The workable approach is to set the window from the aperture area ratio and the paste type at first article, then check the process capability on the first fifty boards before the window is locked for the lot. Reference for deposition acceptance is IPC-7527, and the printed result is judged against the pad and the paste, not against a single global number.
How SUPERB PCBA uses SPI on the line
Printing runs on an automatic printer with a 400 by 340 mm working area, and every board is scanned on a 3D SPI system after print, before placement. The measurements are volume, area, height and offset per pad, and the printer settings are corrected from the scan rather than from a periodic sample.
Builds run on seven SMT lines down to 01005, with 3D AOI after reflow, DIP AOI after wave or selective solder, and X-ray read to IPC-7095 with voiding under five percent. The flow is seventeen steps with ten QC stations, held to IPC-A-610 Class 2 and Class 3, under ISO 9001 and ISO 13485 for medical work. Reflow is a twelve zone nitrogen oven under 500 ppm oxygen with SAC305, so the paste that passed inspection sees the profile the paste maker specified.
The scanner and the printer it feeds are described under 3D SPI and automatic stencil printer, and the station sequence is set out under PCBA quality control.
What SPI cannot tell you
It cannot tell you about a joint that has the right volume and still fails: a cold joint from a profile that ran too fast, a void that formed during reflow, a component that lifted on one side after placement. Those belong to the profile check, AOI and X-ray. Paste inspection narrows what reaches them, it does not replace them.
It also says nothing about paste that was printed correctly and then sat too long before reflow. That is a clock question, not a scanner question, and it belongs with the working life record for the lot.
Solder paste inspection questions
Is every board scanned, or only samples?
Every board, at print. Sampling catches a drifting printer late and lets a bad tray of boards through between samples. For fine pitch and 01005 work the scan is also the fastest way to see a stencil that is starting to clog, because the trend shows in the volume before the board fails.
Can the limits be set to my product instead of a default?
Yes, and they should be. Send the pad geometry, the stencil thickness and the paste type with the build, and the window is set at first article and confirmed on the first fifty boards. The stencil side of that decision is discussed under SMT first article check.
What happens to a board that fails the scan?
It goes to wash and reprint before placement. Paste is removed with the approved cleaning method, the stencil is wiped, and the board is scanned again before it goes on. A board is never sent to reflow with a deposit that failed the window, because the reflow turns a washable board into a rework unit.
Next step
Send the BOM, the Gerber set and the pad detail for the fine pitch parts, or send the build you are about to release and the paste and stencil you have specified. The reply returns the stencil call, the inspection window for your apertures, and the process route for the lot.