On this page
A narrow fired conductor can appear wider, narrower or discontinuous depending on printing direction, paste transfer, edge roughness, illumination and the threshold used to convert an image into a width. A useful process study must therefore keep the physical line and the measurement definition separate. This guide compares line orientations on the same ceramic panel, measures connected width rather than isolated bright pixels and links optical findings to electrical continuity without converting a test pattern into a universal company capability.
Key design decisions
- Place orthogonal and turning features in a common test field so printing direction can be compared under the same pass.
- Lock the edge-detection rule before ranking process conditions.
- Correlate minimum connected width and discontinuity location with a defined electrical path.
Define direction relative to the print mechanics
Label every test line by its direction relative to the squeegee travel and screen geometry, not merely horizontal or vertical on a photograph. Include straight lines in both principal directions, bends, necks and junctions. Keep nominal width and length matched where the comparison is intended to isolate direction. Record panel orientation so an apparent directional effect is not actually a position effect.
Paste experiences shear during passage through the mesh and then rebuilds and levels after separation from the screen. Aperture orientation, mesh crossings, snap-off and substrate support can influence the deposited edge. These interactions depend on the actual paste and screen system. Do not carry a line-width result from one material, mesh or drying history into another design without a new comparison.
Write the width measurement rule first
Specify optical system, pixel scale, illumination, focus method and threshold algorithm. A fixed grey-level threshold may shift the detected edge when surface reflectance changes, while an adaptive threshold may hide a real film-thickness difference. Store the original image and the derived mask. Report whether width is measured perpendicular to the local centerline or along fixed image rows.
Use a traceable scale and verify distortion across the field. Sample enough cross-sections to represent edge waviness rather than selecting the most attractive position. Define how isolated specks, pinholes and edge satellites are treated. The same raw image must produce the same result when reprocessed; otherwise algorithm adjustment can be mistaken for process improvement.
Measure the connected path, not only mean width
Average width can remain close to nominal while one neck creates a high-resistance or open path. Extract local width along the line and identify the minimum connected section after applying the documented noise rule. Preserve the position of each minimum so it can be compared with mesh phase, a bend, a substrate feature or an observed paste defect.
A simple continuity ratio divides the centerline length having connected conductor above the stated minimum width by the evaluated centerline length. This ratio is meaningful only with the same optical resolution and threshold. It does not replace electrical measurement because a visually connected region can be thin, porous or poorly fired, while a dark optical boundary may still conduct.
Cw = Lconnected(w >= wmin) / Levaluated
- Cw: optical connected-width ratio
- Lconnected: centerline length meeting the stated local-width rule
- wmin: drawing- or study-specific minimum connected width
- Levaluated: total inspected centerline length
Registered image scale, fixed segmentation rule and adequate resolution; the ratio is not an electrical rating.
Separate threshold sensitivity from physical spreading
Reprocess the same image at a small, predefined set of defensible thresholds and plot width change. Large threshold sensitivity indicates a weak optical edge, glare, surface texture or a gradual thickness shoulder. If all process groups move similarly with threshold, their ranking may be robust. If the ranking reverses, the study has not established a physical width difference.
Confirm ambiguous edges with an independent view such as controlled oblique illumination or a profile through representative locations. Do not choose the threshold that best agrees with the artwork. The artwork is a target, not an optical calibration standard. Keep physical calibration and segmentation validation separate from acceptance limits on the finished conductor.
| Observation | Competing cause | Next check |
|---|---|---|
| Both directions widen equally | Paste leveling or common threshold shift | Profile and raw-image reprocessing |
| Only parallel lines neck periodically | Mesh or print-direction interaction | Map neck pitch to screen phase |
| Width changes with illumination | Reflectance boundary | Lock lighting and inspect profile |
| Mean width passes but resistance rises | Local neck or film-thickness loss | Map minimum width and voltage drop |
| Bends fail before straight lines | Corner transfer or segmentation geometry | Use local normal-width analysis |
Link optical position to electrical continuity
Use dedicated terminals that isolate the evaluated line from long shared buses. Record test current, contact positions and specimen temperature. For low resistance, consider four-wire sensing where geometry permits. The electrical measurement should cover the same physical region as the optical analysis, with fiducials or coordinates that allow a high-drop segment to be revisited.
Resistance depends on resistivity, fired thickness, width and length. A width-only calculation can screen whether a measured change is plausible, but it cannot infer paste composition or local porosity. Compare lines printed in both directions with matched length and nominal geometry. Repeat contacts to separate probe instability from a genuine intermittent conductor.
Rline ≈ ρf × L / (tavg × weff)
- Rline: resistance of the defined conductor segment
- ρf: effective resistivity of the processed film
- L: evaluated path length
- tavg: representative conducting thickness
- weff: effective connected width
Approximately uniform ohmic film; not valid for isolated necks, cracks, contact-dominated paths or unknown thickness.
Design a direction comparison without confounding position
Distribute both orientations across multiple panel regions rather than placing all parallel lines on one side and all perpendicular lines on the other. Include repeated structures and randomize their measurement order. Record paste identity and condition, screen identity, print direction, support, drying delay and thermal history. Hold these constant when direction is the variable of interest.
Use an unchanged reference condition in every session. If screen wear is suspected, compare early and late prints while retaining panel position balance. A trend that follows print order rather than orientation is not a direction effect. Report missing or damaged test structures; silently excluding them inflates apparent continuity and removes the strongest failure evidence.
Interpret directional failure signatures
Periodic edge scalloping aligned with mesh phase, repeated necks on one side of lines, tails at print termination and directional corner rounding each point to different transfer interactions. A broad halo around every feature may instead reflect optical threshold or paste leveling. Random isolated voids require review of contamination, screen condition and material state before direction is blamed.
A true process interaction should repeat by orientation across separated panel positions and remain visible under a validated measurement rule. Escalate when a neck reaches the drawing limit, creates unstable resistance, approaches a high-current region or changes isolation to a neighboring conductor. Do not publish a minimum printable width from a single test image or one successful panel.
Release a measurement recipe with the artwork
The handoff should include test-pattern revision, nominal geometry, print direction, screen and paste identities, image acquisition settings, segmentation rule, sampling interval and electrical terminals. Store raw images as well as summary numbers. A table of average widths without the local traces cannot show whether continuity is controlled by one short neck.
Reopen the study when routing rotates, line width or spacing changes, the screen or paste system changes, a new ceramic surface is introduced or inspection software is updated. State which conclusions are optical and which are electrical. The result supports a drawing-specific decision; manufacturing limits and acceptance ownership remain subject to engineering review.
Send the narrow-line artwork and measurement boundary
Provide the inputs needed to review printing direction and continuity.
- Layer artwork, nominal widths, bends, spacing, print orientation and panel map.
- Ceramic, conductor system, screen, print, drying and firing identities or proposed conditions.
- Raw calibrated images, threshold method, local-width traces and excluded-feature record.
- Electrical path, terminals, test current, temperature and drawing-specific acceptance limits.
The drawing-upload form loads as you reach this section.