Silicone Keypad Tolerances and Assembly Stack-Up

Specify silicone keypad tolerances from the installed functional window, not a blanket plus/minus value. Establish assembly datums, define how the flexible part is measured, and calculate whether dimensional extremes leave release clearance and enough travel for electrical closure before the hard stop. Dimensional conformity and functional acceptance are related, but they are not interchangeable.

The silicone rubber keypad design guide provides the broader design framework. This article develops a tolerance-allocation method for a silicone key with a conductive pill contacting a PCB.

Choose Datums from Assembly and Function

Start with the surfaces and locators that actually position the assembly. SiTECH’s dimensioning guidance emphasizes functional features and references that can be measured repeatably.

One possible keypad datum scheme uses a rigid chassis support plane as A, a housing locator as B for in-plane position, and a second locating feature as C for rotation. A round-hole and relieved-slot arrangement is one concept to review, not a mandatory construction. Define the locating clearances rather than assuming that a pin removes all movement.

Connect the keypad’s seating features, PCB locators, and housing dimensions to that assembly reference. Merely naming unrelated surfaces “A” on separate drawings does not establish a common datum. A flexible keymat needs a defined datum-simulation fixture, not an instruction to flatten it by hand.

Separate critical relationships from dimensions that do not control function:

A trimmed perimeter is not automatically non-critical. Its role in the assembly decides its treatment.

Define How Flexible Parts Are Measured

The public ASTM D3767 significance-and-use text explains that measuring pressure affects the observed thickness of soft flexible materials. Its scope excludes sampling and where samples are taken. Define the measurement locations for individual features in the project inspection plan.

Use explicit states rather than a single undefined “keypad height”:

For contact measurement, record the foot geometry and applied force or pressure. For noncontact measurement, validate the support, viewing direction, and edge-detection method. Compare repeated removal and refixturing, not just repeated readings without moving the part.

The public ISO 23529 abstract covers general preparation and conditioning of rubber test pieces, but excludes special requirements for testing whole products. Define the whole-keypad method separately. Record material and cure history, conditioning, temperature, instrument, fixture revision, measurement locations, and sample identity. Agree how measurement uncertainty affects acceptance; displayed resolution alone is not an uncertainty estimate. Keep free-supported dimensions separate from installed deformation.

Worked Example: Allocate Tolerances to a Functional Window

Establish the model and inspection state

Illustrative example: every dimension, tolerance, and functional limit below is hypothetical. None represents JASPER capability or a measured result.

Consider one centrally pressed conductive-pill key. All lengths are in millimeters. Let +z point from Datum A toward the user. Use an illustrative reference temperature of 23 °C, after thermal equilibrium and completion of the declared cure process.

The housing has a keypad seating plane M. The keymat is located and its perimeter retained against defined assembly stops, with no external load on the key at rest. Assume axial, one-to-one key/pill movement before first contact. Lateral rubbing and tilt are outside this one-dimensional model and need separate checks.

P is a displacement, not a force or an assumed gasket-compression percentage. Determine it from the defined free-supported and installed states; do not automatically equate perimeter compression with pill movement.

The coordinates give the dimension chain directly:

Pill face at installed rest:  z_pill = H − K − P
PCB contact plane:           z_pad  = S + T
Rest gap:                    G      = H − K − P − S − T
Nominal gap:                 G_nom  = 4.40 − 1.80 − 0.10 − 0.40 − 1.60
                                    = 0.50 mm

Calculate both dimensional extremes

The smallest gap combines the lowest H with the highest subtractive dimensions. The largest gap reverses those choices:

G_min = 4.35 − 1.87 − 0.12 − 0.43 − 1.65 = 0.28 mm
G_max = 4.45 − 1.73 − 0.08 − 0.37 − 1.55 = 0.72 mm
Total worst-case half-width = 0.05 + 0.07 + 0.02 + 0.03 + 0.05
                            = 0.22 mm

These are separate limiting assemblies, not two conditions occurring simultaneously in one part. The bounded calculation does not require statistical independence, although coupled dimensions can make it conservative.

Translate closure and recovery requirements into a gap window

For this example, assume the following functional allocations:

The 1.0 s observation time is also hypothetical, not a standard requirement. Verify q and r using agreed press/release conditions and electrical thresholds; they cannot be inferred from component dimensions alone.

The two geometric conditions are:

Release:  G_min − r_max ≥ g_release
Closure:  D_min − (G_max + q_max) ≥ m

Therefore the permitted rest-gap window is:
G ≥ 0.25 + 0.05 = 0.30 mm
G ≤ 0.90 − 0.12 − 0.10 = 0.68 mm

At a nominal gap of 0.50 mm, a symmetric worst-case tolerance budget cannot exceed the smaller distance to either boundary:

Permitted half-width ≤ min(0.50 − 0.30, 0.68 − 0.50) = 0.18 mm

The initial 0.22 mm budget is too large. Moving the nominal alone cannot fit its 0.44 mm-wide range inside the 0.38 mm-wide permitted window.

Reallocate, then verify feasibility

For calculation only, reduce K’s tolerance to ±0.04 mm and T’s to ±0.03 mm, leaving everything else unchanged. The new half-width is 0.17 mm, producing G = 0.33–0.67 mm.

The initial calculation identifies inadequate allocated clearance and reserve; it does not establish that a tested key fails to switch. Likewise, the revised calculation is conditional on every assumed bound being valid. Positive clearance does not prove that the key actually rebounds or the circuit reopens.

The remaining distances to the permitted gap boundaries are only 0.03 mm and 0.01 mm. Measurement decision rules and service-condition changes have not received separate allowances here. This is a closure-and-release screen, not proof of safe maximum overtravel, contact stress, or service life.

Before adopting tighter values, confirm molding, PCB, and inspection feasibility. Alternatives include reducing the number of stack contributors or changing support and stop geometry. Do not turn a mathematical allocation into an unsupported purchasing specification.

Use Statistical Stack-Ups Only with Suitable Data

For a linear chain with uncorrelated dimensional variations:

σ_G = √(σ_H² + σ_K² + σ_P² + σ_S² + σ_T²)

NIST’s uncertainty guidance documents the variance-propagation mathematics and the need to consider covariance. Applying that mathematics to manufacturing variation requires actual process data; the NIST page is not a keypad production-acceptance standard.

A drawing’s ± tolerance is not automatically a standard deviation. Establish means, variation, correlations, and the distribution assumptions used to estimate tail risk. Keep cavity and assembly-state effects visible. In particular, P may depend on the same geometry already represented elsewhere in the chain.

Use covariance terms when inputs are correlated. Preserve shared variables when deriving stop travel and gap, rather than treating duplicate contributions as independent. Neither an RSS calculation nor a small predicted standard deviation proves an acceptable assembly yield.

Link Dimensional Checks to Functional Acceptance

Measure force in N, local key displacement in mm, and electrical state during press and release. Mark-10’s switch-testing discussion distinguishes force captured at electrical activation from a later bottoming-force spike. A peak-force reading alone is therefore not a closure measurement.

Define probe geometry, press point, speed, displacement zero, support, fastening condition, electrical test level, thresholds, and release-observation time. Check that the displacement channel represents key movement rather than uncorrected fixture deflection.

Use the tooling and prototyping guide to connect these results to controlled sample and production revisions. A passing nominal sample is not evidence for every permitted combination.

Frequently Asked Questions

Request a Keypad Design Review

For a project purchased as an integrated module, review silicone keypad assemblies and identify who owns the rubber, circuit, housing, and inspection interfaces.

Send the keypad drawing or sample, PCB/contact layout, enclosure section, support and fastening details, datum scheme, current tolerance stack, and any force/electrical records. Include the equipment application, prototype and annual quantities, critical acceptance requirements, and the supply boundary. Mark unresolved inputs TBD so they can be addressed before requirements are frozen.

Request a Keypad Design Review.