Checking Field-Position Error in Optical PCB Measurements

An optical measurement program can report different dimensions for nominally identical PCB features simply because they occupy different positions in the image. Before adjusting manufacturing compensation, determine whether the discrepancy follows the board feature or the measurement field.

Separate image scale from edge detection

Mitutoyo distinguishes optical magnification accuracy from measuring accuracy and explains how illumination affects dimensional observation. A sharp picture and a calibration label do not establish the uncertainty of every measurement program.

Measure a suitable reference feature near the image center and at several off-center positions without changing the feature itself. Repeat at the working magnification and illumination used for boards. If the apparent dimension changes with field position, investigate calibration, distortion correction and the measurement configuration before blaming etching or routing. Include the reference artifact’s uncertainty when judging whether an observed difference is meaningful.

Use the actual surface and height conditions

Copper, solder mask and laminate edges produce different contrast. A threshold chosen for bright copper can shift the detected boundary on a dark mask or translucent substrate. Board warp also changes object height. Even with telecentric optics, focus and edge contrast still need to be controlled within the instrument’s intended operating range.

• Retain magnification, exposure and illumination settings.
• Record which physical edge the software is detecting.
• Check repeated placement and rotation of a common sample.
• Use reference features that bracket the dimension and field area of interest.

Apply the result to the manufacturing decision

For example, a panel inspection might suggest that connector lands near one image corner are narrower. Move those same lands to the center. If their result improves while previously acceptable lands become narrow at the corner, a manufacturing change would be premature. If the error stays with the physical feature, the process investigation has stronger evidence.

Preserve both raw images and reported coordinates so another reviewer can inspect the edge choice. Relate the verified method to trace-width definition and hole-position datums. The measurement record should describe uncertainty at the relevant geometry, rather than quoting pixel size as if it were dimensional accuracy. A reliable comparison links a defined physical boundary to a demonstrated measurement method before it drives a process correction.