SPI Calibration Verification: Trusting a Paste Volume Number

An SPI system reports a solder paste volume for every pad, and that number is used to accept or reject a print, to adjust the printer and increasingly to release a board without further inspection. SPI calibration is what decides whether the number describes the deposit or describes the machine.

The distinction matters because the two look identical on a report. A machine that has drifted low by ten percent produces a report in which every pad is ten percent short, which prompts a printer adjustment that makes the true deposit ten percent too heavy. Verification is what breaks that loop. The machine that causes trouble is not the one that reports a failure, but the one that reports that everything is fine.

What the System Reports

The system reports height, area and volume for each pad, and the three are linked through the profile of the deposit. Volume is derived from the height map over the measured area, so an error in either the height reading or the area boundary changes the volume that is reported. That dependency matters when a single pad reports an unusual value while its neighbours look normal.

Height is measured optically, and the reading depends on how much light the paste returns. Fresh paste reflects differently from paste that has started to dry, and a bright or a dark solder mask beneath changes the reference surface. Those are measurement conditions as much as process conditions. A change of mask colour between two products is enough to shift the reading with no change at all to the printing process.

The Reference Target as a Standard

A reference target is a sample whose features have known and stable dimensions. It is measured by the system, and the reported values are compared with the certified values. Because the target does not change, any change in the reported values belongs to the machine. Where the target is certified externally, the certificate and its revision date belong with the measurement record.

The target has to be clean, at the working temperature and presented in the same way each time, because a target that is rotated or placed at a different height is a different measurement. Keeping one dedicated fixture for the target removes that variation. Handling matters too, because a target held by its optical surface collects oils that afterwards change the reading.

Gauge Repeatability and Reproducibility

Gauge repeatability is the spread of repeated measurements of the same target by one operator, and reproducibility is the spread between operators or between machines. Both are expressed as a fraction of the tolerance the system is used to judge.

A common rule is that the measurement system should consume less than ten percent of the tolerance, with twenty or thirty percent marking the point at which the system becomes a significant part of the variation the process is judged against. The study should be run with the operators and the shift pattern that will really use the system.

Height, Area and Volume Accuracy

Once repeatability is established, accuracy is checked against the certified values of the target. Height errors and area errors behave differently, so separating them needs targets with different footprints rather than a single pad. Measurement accuracy is then reported per channel. A target that is correct in height and wrong in area will send the investigation in the wrong direction.

A target with a small footprint and a nominal height shows height accuracy with little influence from area. A target with a large footprint and a shallow deposit shows the area boundary and the way the software decides where the paste ends.

Zero and Height Offset Drift

Most systems work from a reference plane established on the board surface or on the mask. If that plane is set incorrectly, every height on the board is offset by the same amount and the volume follows it. The error appears as a constant addition to every reading, which is exactly the signature a genuine process shift does not have.

An offset of a few micrometres is small against a deposit of a hundred and twenty micrometres, and it is not small against the tolerance of a fine pitch print. The zero check is quick, it uses the bare board, and it belongs at the start of a shift.

Verification Frequency and Triggers

Frequency is set by how quickly the system drifts and by how much the product depends on the result. Weekly verification on a target suits a stable machine, and daily verification suits a machine that is used to release product. Verification performed only when a customer asks documents the state of the machine on the day of the visit.

Triggers sit alongside the schedule. A lens or a light source replacement, a move of the machine, a software update, a change of calibration paste or a new finish on the target all justify a verification before the next product is measured.

Correlation Between Machines

Where two SPI systems inspect the same product, each is verified against its target and the two are then compared on the same printed board. The comparison shows whether their differences fall inside the tolerance the product allows.

Correlation matters when one machine releases a board that the other rejects. Without a correlation record that difference is settled by discussion, and with one it is settled by a number both sides accept. The record should name the board used, its print parameters and the date, so the comparison can be repeated on the same input.

Effect on Print Feedback

When SPI data is fed back to the printer, the accuracy of the system becomes part of the control loop. A system reading low pushes the printer to overprint, and the loop can settle at a point where the reported volume is correct and the true volume is not.

That is why a closed loop should be checked against an independent measurement at intervals. The reference methods are described in the notes on paste volume measurement, and the feedback side is covered in the notes on print feedback from SPI data.

Records and Limits

Records hold the target identity and its certificate, the date, the reported values in each channel, the operator and the limit that was applied. With those fields, a change in a reported value can be attributed or excluded. A file holding the raw readings is more useful than a certificate that holds only a pass. The schedule itself should be recorded as well, because a missed check is invisible in a file that contains only the checks that were done.

Limits belong in the record as well, because a verification result without a limit is a number without a decision. Where the machine releases product, the input side is kept stable by the practices in the notes on stencil underside cleaning.

FAQ

Can calibration be replaced by inspecting a real board? A real board carries too many variables to isolate the machine. A reference target plus a periodic board check covers accuracy and relevance together.

How much of the tolerance may the measurement system consume? Ten percent is a common target. Above twenty percent the system starts to dominate the variation being judged.

Should the zero check use a bare board? Yes. The bare board establishes the reference plane, and it is faster and more repeatable than checking a printed board.