Membrane Switch Film Shrinkage: One Scale or Two?

Define the dimensional-change record before comparing percentages

Mark the film's machine direction (MD, along the web) and transverse direction (TD, across it), then show how they map to drawing X and Y. Keep that orientation with each sample and layer. A sheet rotated for nesting changes this mapping; a drawing's horizontal dimension is not automatically the same film direction on the next order.

ASTM D1204-25 addresses linear dimensional change under specified elevated temperature and time. Its public listing also cautions against assuming equivalence with a liquid-medium method or ISO 11501. A method name without the actual applicable procedure and specimen conditions is therefore insufficient comparison evidence. Obtain the authorized full method where required; this article does not reconstruct it.

For the illustrative calculation, both initial and final lengths are read in the same agreed measurement state, after the stated processing sequence. That convention isolates the assumed net change for the model; it does not predict dimensions while hot. A laboratory exposure is not a safe production recipe, and a material result is not a finished-part tolerance. Ask for actual material temperature, time, restraint, conditioning and measurement details rather than copying an oven display setting.

Worked example: one factor fixes X but overcorrects Y

Hypothetical HYR editorial example—not a factory measurement: an initial X span is 120.00 mm and a Y span is 80.00 mm. Assume uniform reductions of 0.15% in X and 0.05% in Y after one defined process history. These rates are invented solely for arithmetic and do not describe a named film, material lot or HYR process. The resulting spans are 119.82 mm and 79.96 mm.

In this ideal model, final length equals initial length multiplied by (1 − s), where s is the assumed fractional reduction, not its percentage number. To target 120.00 mm in X, the inverse factor is 1 ÷ 0.9985, approximately 1.001502. If that same factor is applied to the original 80.00 mm Y span, its modeled final Y dimension is 80.00 × 0.9995 ÷ 0.9985 = approximately 80.0801 mm. X fits the target, but Y is about 0.0801 mm above its target—not automatically acceptable or unacceptable without a project tolerance.

Separate inverse factors would give theoretical starting spans of approximately 120.1803 mm in X and 80.0400 mm in Y. This is a counterexample to a universal scale factor, not a recommendation to release those artwork dimensions. It assumes the percentage changes remain constant after resizing and ignores every nonuniform effect. Verify any proposed correction with actual repeatable data and controlled sample features.

Compare graphic and circuit layers without resetting the datum

Extend the hypothetical example with a circuit layer that also starts with a 120.00 mm X span but has an assumed reduction of 0.05%. Its modeled final span is 119.94 mm. The graphic layer's 119.82 mm and circuit layer's 119.94 mm differ by 0.12 mm across that span. Equal starting artwork dimensions did not produce equal final spans in this model.

That difference is not automatically a 0.12 mm shift at every key. If the two near ends are fixed to the same ideal origin, the far endpoints differ by 0.12 mm. If both spans are symmetrically centered about a common origin, the endpoint differences are 0.06 mm at each end. Real locating, rotation, local distortion and assembly errors require their own records; a span alone cannot identify them.

Nor should every successive heat exposure receive another identical percentage deduction. The real material's history may change its response. Keep separate raw records for each film, lot and manufacturing stage instead of multiplying an assumed rate by the number of print passes. A proposal to rotate a layer by 90 degrees also needs review: in this simple example it swaps the assumed rate assigned to the drawing axes.

What this means for OEM engineering and sourcing

Send each quoting supplier the same dimensional baseline and ask which dimensions are controlled finished-part requirements, which are provisional artwork settings and which require sample evidence. If a supplier proposes a different grade, orientation or stabilization route, ask for a documented comparison rather than accepting a generic low-shrink label. The nominal film family alone does not close print-surface, optical, bond or forming requirements.

Retain multiple controlled feature locations, not just overall width and height. They let the responsible engineer distinguish a scale pattern from translation, rotation or local deformation. Specify the measuring fixture, supported state, conditioning, repeat measurements and decision rule, including how uncertainty is handled near a limit. Do not silently turn the four decimal places shown by this arithmetic into a manufacturing precision claim.

A corrected loose pattern still needs the separate mounted-clearance and function checks. A display border, key center, spacer opening and circuit contact can have different functional consequences. Use the window-registration guide for its clearance budget; do not count the same dimensional effect twice under both process compensation and assembly tolerance.

Five ways the scale decision can fail

Use the observed dimensional pattern to decide what evidence is missing before changing the artwork. These are plausible failure modes, not reported HYR defects.

• An X-axis correction is applied isotropically: X approaches its target while Y moves away. Compare both axes under the same measurement convention.
• A nested sheet is rotated but its orientation record is not: the assumed relationship between film direction and drawing axes changes. Recheck the actual cut orientation.
• A graphic-layer correction is copied into a circuit layer with a different process history: the individual outlines can look plausible while relative feature locations differ.
• Only outer dimensions are checked: local registration error or rotation remains hidden. Retain additional functional coordinates and a common datum scheme.
• An attractive calculation replaces sample approval: ink adhesion, window appearance, key response or mounted fit changes remain untested. Close those checks on the proposed construction.

Release an RFQ that keeps the correction provisional

Attach the controlled artwork and drawing, layer map, film orientation and critical-feature coordinates. Include the actual grade and lot where known, material thickness, print and processing sequence, restraint, temperature/time evidence and the defined initial/final measurement state. Ask the supplier to return raw dimensional records, the proposed correction, limitations and a plan for comparing the corrected sample with the released drawing.

Name change triggers: film grade or lot behavior, thickness, surface treatment, sheet orientation, heat history, ink/process sequence, restraint or fixture, artwork, and measurement state. The responsible engineers decide what must be reverified after each relevant change. Retain the accepted sample and superseded settings so a repeat order does not inherit an unexplained scale value.

HYR's secure RFQ can start that bounded membrane-switch and graphic-overlay discussion. Send the files you have and label missing data; do not invent process settings to complete the enquiry. Feasibility, compensation and project-specific acceptance remain engineering decisions supported by the actual sample and agreed evidence.