Volume Accuracy: 6 Checks for Paste Inspection

Volume accuracy is the agreement between the volume a paste inspection system reports for a deposit and the volume that is actually there. It is not the same as repeatability, which only says that the machine gives the same answer twice, and a system can be perfectly repeatable while being confidently wrong.

The distinction matters because the inspection result is used to control the printer. A machine with a measurement bias will push the process in the wrong direction, and the operator will see a trend that looks like a printer fault while the printer is doing exactly what it was asked to do.

What Volume Accuracy Actually Measures

A paste inspection system does not measure volume directly. It measures the height of the deposit across its area, sums the result and converts it to a volume, so any error in the height calibration appears as an error in volume.

That chain has several links: the sensor, the height reference, the algorithm that decides where the deposit ends and the threshold that separates a real deposit from noise. Accuracy is the product of all of them, which is why a calibration on a single flat plate proves less than it appears to.

Setting the Reference Pad

Most systems learn a reference from a known good board and then judge every later board against it. The learned value becomes the target, so a board that was itself marginal teaches a marginal reference and the process is then controlled around the wrong number.

The reference should come from a board whose printing has been confirmed independently, by height measurement or by cross section on a sample. Where the reference is taken from a production board chosen for convenience, the inspection system is validating the printer against the printer.

Deposit Height, Area and the Volume Between

Two deposits can have the same volume and very different shapes. A tall, narrow deposit releases poorly into the joint, while a flat, wide one with the same volume behaves properly, so a volume figure on its own does not describe whether the deposit will solder.

Height and area should be reported alongside volume, at least on the small apertures where the shape matters most. Our aperture area ratio notes explain how the geometry of the opening sets the shape that can be produced.

Paste Inspection in Line or Off Line

An in line system inspects every board at production speed and stops the line when a limit is crossed. An off line system inspects a sample with more time and better resolution, and is mainly used for setup and for investigations rather than for control.

The choice is not either or. A line with an in line system still benefits from an off line check when a new stencil or a new solder paste is introduced, because that is the point where the reference and the limits are being established rather than used.

Trend Data and Control Limits

Volume data is most useful as a trend. The absolute value of a deposit depends on the stencil, the paste, the printer settings and the board support, while the change over time depends mainly on the things that drift, such as paste condition and stencil cleanliness.

Control limits should be set from the distribution of a stable process rather than from the specification. A limit set too wide produces no reaction, and one set too narrow produces constant alarms, and both teach the operator to ignore the system. Our inspection correlation notes describe how the limits link to the optical check that follows.

Aperture Size and the Expected Volume

The expected volume for an aperture follows from its area and the stencil thickness, with an allowance for the release efficiency of the paste. Small apertures release less of their theoretical volume than large ones, so a single target percentage across a board is not a realistic expectation.

Where the printer software stores a per aperture target, the comparison becomes meaningful for each pad. Where it stores one target for the board, the small apertures will read permanently low, and the natural reaction is to over print the large ones to bring the average up.

Noise, Calibration and False Rejects

A false reject costs time on every board it touches, and it usually comes from a threshold that is too close to the noise rather than from a printer problem. Dust on the sensor, a dirty reference or a fluorescent lighting change can all move the reading.

Calibration should be done with the artefact the maker supplies and at the interval the maker states, and the result should be recorded. A system that is calibrated but whose result is not written down cannot be defended when a customer asks how the data was produced.

What to Do When Volume Drifts

The first checks are the stencil and the paste, because they change faster than the machine. Apertures that are partially blocked by dried paste produce a slow fall in volume on the small pads, and a paste that is warming up or drying out produces a change that follows the shift rather than the board.

Only after those are excluded should the printer settings be touched. Changing squeegee pressure or speed to compensate for a blocked stencil hides the real fault and usually makes the next shift worse, and our skipped deposit notes describe the pattern to expect.

Reporting Volume to the Customer

Where a customer asks for inspection data, the report should state the system, the sampling, the reference used and the limits applied. A percentage figure quoted without those items cannot be compared with a figure from another supplier, and the comparison will be made anyway.

The most useful report is a distribution rather than an average, because an average hides the small number of pads that are outside the limit. A histogram for the smallest aperture on the board tells the assembler where the risk actually sits.

Records and Correlation With X-Ray

Volume data becomes much more valuable when it is correlated with the joint that results. Comparing the inspection record for a panel with the X-ray image of the same panel shows which volume actually produces a good joint, and that is the number the control limit should be built on.

That correlation is a project rather than a routine, and it only has to be done once for each product family. Our paste release notes explain why the release efficiency differs between apertures, and the SMTA process library covers the inspection methods in more detail.

FAQ

Is repeatability enough to trust a system? No. A repeatable system with a bias gives consistently wrong data, and the process is then controlled to the wrong target. Accuracy against an independent measurement is what makes the data usable.

Should every board be inspected? For a mature product, sampling is usually sufficient because the process is stable. For a new stencil, a fine pitch board or a first run, inspecting every board for a shift is worth the time and gives the data to set the limits.

Why does the volume read low on the smallest pads? Because release efficiency falls as the aperture gets smaller, and because the measurement itself is more affected by the edge of the deposit. The per aperture target should reflect that rather than treating every pad the same.