Topre & EC Dome Force Curves: How to Read and Compare Domes

A force–travel curve is not a complete model of how a keyboard feels. It is a controlled mechanical measurement: the reaction force recorded while a key mechanism is moved through a stated displacement path. The curve can be reduced into useful quantities, but no single quantity—not peak force, Snap %, slope, or area—is a universal measure of perceived weight or tactility.

That distinction matters for Topre and Topre-compatible electrostatic-capacitive (EC) domes. A Topre patent application identifies the classic 30 gf and 45 gf variable-weight values as targets for the peak press load, and places the switch-on region after that peak.^[1] Current REALFORCE documentation likewise lists key-press-force settings separately from adjustable electrical actuation positions.^[2] Those sources support a peak-force interpretation for Topre's 30/45 designations; they do not establish a universal naming convention for every aftermarket dome.

The Topre Force Curve Bench therefore reports the curve as a set of complementary measurements. This article defines those measurements, explains the Dome Lab's metrics-v4.1 processing rules, and states what the results can—and cannot—support.

The bench and the house protocol

The Dome Lab rig is based on bluepylons' open-source Open-Switch-Curve-Meter Gen 2 and was commissioned as a custom build. The original design uses a motorized linear stage for displacement and a load cell for force. With its 1 mm-lead stage and 200-step motor, one full motor step corresponds to a nominal 0.005 mm of motion.^[3] The Dome Lab adds its own Topre/EC fixture, acquisition procedure, and analysis software.

The current house protocol is:

• Nominal displacement increment: 0.005 mm.
• Press speed: approximately 0.15 mm/s, making the test quasi-static rather than representative of normal typing speed.
• Force measurement: a load cell calibrated and checked with known test weights under the project procedure.
• Stroke directions: press and controlled return are recorded in the raw files.
• Replicates: event detection and nonlinear metrics are computed on each accepted physical run first; the run-level results are then averaged. The mean curve is a visualization, not the source of canonical peak, valley, crossing, or maximum-slope values.
• Numerical precision: calculations use unrounded data; rounding occurs only for display.

The 0.005 mm value is a commanded step increment, not a claim of 0.005 mm measurement accuracy. Position is inferred from motor steps, as in the source design, and force accuracy also depends on calibration, alignment, drift, fixture geometry, and the tested assembly. The published results characterize the particular specimens and configurations under this house protocol; they are not a product-line tolerance specification or an accredited calibration certificate.

First identify the curve landmarks

Let F(x) be the press force in gram-force (gf) at travel x in millimetres. The analysis identifies three mechanical landmarks before calculating the derived metrics:

Travel uses the project’s travel-v3 rule on the landmark-detection curve: after the valley, find the first 0.005 mm step whose force increase is greater than 1% of the current measured force, require that condition for three consecutive steps, and apply a 10 gf noise floor. This is an operational house definition chosen to make bottom-out onset reproducible across the dataset. The steep hard-stop wall after that onset is excluded from full-stroke press work.

Mechanical collapse, the post-collapse valley, bottom-out onset, and electrical actuation are different events. The force bench does not infer a fixed electrical actuation point from the mechanical curve. That separation is especially important because current REALFORCE APC models can assign several electrical actuation depths to the same key mechanism.^[2]

Drop size and drop rate are different measurements

The central v4.1 change is to decompose the tactile event instead of asking Snap % to represent all of it.

ΔF = Fc − Fv

Δx = xv − xc

The 0.10 mm search for STEEPEST DROP (GF/MM) is restricted to [xc, xv]; both ends of every candidate window must stay inside the tactile descent. This prevents seating features, non-monotonic tails, or the bottom-out wall from becoming false maxima. At the bench’s nominal spacing, 0.10 mm covers 20 increments. The wider finite span is less noise-sensitive than differentiating adjacent 0.005 mm samples, but it remains a method-defined descriptor and is calculated per run before averaging.

Snap % is still useful. The same formula appears as “click ratio” in a silicone-keypad manufacturer’s force–travel terminology.^[4] But it contains no travel term. Historical ASTM F2592-16 also treated force–displacement slope as useful for differentiating tactile response and replaced earlier tactile-ratio terminology with slope-angle measures; that standard was withdrawn in 2023 with no replacement.^[5] The Dome Lab therefore does not claim current ASTM compliance.

Worked example: Sony BKE Gray 1

Using the current metrics-v4.1 canonical values F_c = 78.0 gf at x_c = 1.290 mm and F_v = 29.8 gf at x_v = 3.125 mm:

• DROP: 78.0 − 29.8 = 48.2 gf
• DROP TRAVEL: 3.125 − 1.290 = 1.835 mm
• DROP RATE: 48.2 / 1.835 = 26.3 gf/mm
• SNAP: 100 × 48.2 / 78.0 = 61.8%
• NORM. DROP RATE: (48.2 / 78.0) / 1.835 = 0.337/mm

Canonical derived metrics are calculated from the underlying per-run values before display rounding, so recomputing them from fewer-decimal chart labels can differ slightly. This example demonstrates numerical decomposition only; it is not a human-perception validation.

RAMP measures the pre-collapse rise

RAMP (GF/MM) is the average slope through the central 10–90% of the baseline-to-peak rise. It is designed to describe force buildup without anchoring the result to the initial seating transient.

For each run, metrics-v4.1 defines F_b as the median force on the landmark-detection curve over the fixed 0.05–0.15 mm displacement band. This band is a documented Dome Lab house-method assumption, not a universal standard. The two thresholds are:

T10 = Fb + 0.10(Fc − Fb)
T90 = Fb + 0.90(Fc − Fb)

The software finds the last upward crossing of each threshold before collapse, linearly interpolates the crossing travel, and calculates:

RAMP = 0.8(Fc − Fb) / (x90 − x10)

RAMP is an average 10–90% rise slope. It is not the local derivative at the peak and has not been validated as a standalone heaviness score. The baseline is used only for this ramp definition; it is not silently subtracted from collapse force or the work integrals.

Force–travel area is press work, not “actual weight”

The area under a press curve is mechanical work along the measured path:

W(a,b) = ∫ab F(x) dx

The Dome Lab retains two endpoint-specific work quantities:

• PRE-COLLAPSE WORK (GF·MM): ∫0xc F(x) dx, the gross press work supplied up to the collapse peak.
• FULL-STROKE PRESS WORK (GF·MM): ∫0xbo F(x) dx, the gross press work supplied up to detected bottom-out onset.

The previous label “Total Energy” has been retired because it could be mistaken for stored elastic energy, energy released by collapse, or energy dissipated during an entire press–return cycle. FULL-STROKE PRESS WORK says exactly what was integrated and where the integration stops.

Gram-force millimetre is a work unit: 1 gf·mm = 9.80665 µJ.^[6] Work may be relevant to repeated effort, especially for a user who travels to bottom-out, but it is not a universal definition of perceived weight. A user who reverses shortly after actuation does not perform the full-stroke work, and quasi-static area does not contain the dynamic impact of bottoming out.

The silencing-ring correction in metrics-v4.1

Topre’s JP2012-138254A patent describes an annular cushioning member positioned between the upper surface of the plunger and the housing’s upper stop. Its illustrated embodiment uses a PET layer with a softer foamed-urethane layer and is approximately 0.5 mm thick; the patent’s stated function is to reduce the sound produced when the plunger returns to that upper stop.^[7] The patent does not, by itself, establish the force-curve or perceived-feel consequences of every aftermarket ring and slider combination.

An earlier Dome Lab draft claimed that ring precompression steepened the pre-collapse buildup. That conclusion came from an early-anchored slope contaminated by the initial seating/preload force. Reprocessing the same Topre Slider Black comparison with metrics-v4.1 changed the numerical result. In the canonical records, all three rows use the same specimen catalogued as a Topre 45g Aged dome, standard Topre slider and housing, and Topre conical spring; the two comparison rows add the named ring:

These are canonical per-run-then-average values from the extracted metrics-v4.1 dataset. Within these tested assemblies and this low-rate protocol, both ring configurations had a lower central rise slope than the bare-slider configuration. The former “rings steepen buildup” statement is withdrawn. The 0.3 mm assembly’s collapse force was slightly lower than baseline, while the 0.5 mm assembly’s was higher; the result is not a monotonic “rings increase collapse force” effect. These three measurements do not isolate ring material, preload, slider geometry, travel, sound, and speed as independent causal variables, and they do not establish a universal threshold for how a ring will feel during typing.

How force slope relates to dynamic feel

DROP RATE and STEEPEST DROP are spatial slopes: change in force per change in displacement. They are not change in force per second. During a moving press, the two are related by the chain rule:

dF/dt = (dF/dx)(dx/dt)

The same force–travel slope can therefore produce a different force rate when traversed at a different speed. A controlled study of ten experimental click switches treated force-drop size and force-drop rate as separate design variables and found both associated with perceived feedback clarity, with drop size showing the stronger relationship in that experiment.^[8] That study supports reporting magnitude and slope separately; because it used long-stroke experimental mechanisms with a common peak force, it does not validate NORM. DROP RATE as a Topre perceptual score.

Rubber mechanisms are also rate dependent. Nagurka and Marklin tested rubber-dome keyboard keys at constant speeds from 0.5 to 80 mm/s and reported peak forces at 80 mm/s more than 12% above the quasi-static 0.5 mm/s values for the keys shown.^[9] The Dome Lab’s approximately 0.15 mm/s curves are valuable controlled comparisons, but they are not exact predictions of force, vibration, sound, or impact during natural typing.

How to compare two domes responsibly

Start with the question you are trying to answer, then read the corresponding group of metrics:

• Peak-load and early-stroke comparison: COLLAPSE force and travel, RAMP, and PRE-COLLAPSE WORK.
• Size of the tactile release: DROP and SNAP.
• Width and spatial abruptness of the release: DROP TRAVEL, DROP RATE, NORM. DROP RATE, and STEEPEST DROP.
• Full-stroke comparison: TRAVEL and FULL-STROKE PRESS WORK, while remembering that bottom-out impact is dynamic.
• Return and live typing: inspect the return curve and consider speed, hysteresis, vibration, sound, keycap, housing, springs, silencing parts, and user technique. Those factors are not reducible to the press-curve metrics above.

Compare absolute and normalized quantities together. DROP RATE preserves the actual force change per millimetre; NORM. DROP RATE divides by the dome’s own peak. A light dome can therefore rank higher on the normalized descriptor while a heavy dome has the larger absolute force drop. Neither ranking is an error—they answer different questions.

Also keep the assembly boundary visible. A dome-only comparison, a complete-key comparison, and a slider/ring experiment are not interchangeable. Aftermarket parts, aged parts, and individual rubber specimens can vary. A curve describes the tested unit and configuration; unmatched specimens cannot isolate age, material formulation, or any other single cause.

Data processing and reproducibility

The v4.1 canonical rule is per run, then aggregate. Peak, valley, crossings, travel, local maxima, slopes, and work endpoints are found independently on every accepted run. Averaging curves first can smear sharp features when their positions differ slightly, particularly the STEEPEST DROP maximum and threshold crossings. The implementation applies a seven-sample moving average to mechanical-event and force-shape calculations; force–travel work is integrated on the unsmoothed press values. The v4.1 dataset metadata defines runfilter-v1 as excluding a run when collapse force differs by more than 1.0 gf or collapse position differs by more than 0.10 mm from the batch median. These choices are part of the method version, not universal constants.

The released schema uses explicit field names such as collapse_force_gf, collapse_travel_mm, valley_force_gf, valley_travel_mm, full_stroke_press_work_gf_mm, precollapse_work_gf_mm, steepest_drop_0p10mm_gf_per_mm, and ramp_10_90_gf_per_mm. Method-defining choices—including the nominal step, 0.10 mm window, ramp fractions, baseline band, interpolation, smoothing, aggregation, run filtering, and travel rule—are recorded in data/schema_meta.json. Per-test quality_flags carry named computation failures; review status and assembly notes remain separate provenance fields.

Raw press and return CSV files and the public tools are maintained in the Dome Lab data repository.^[10] Use the version displayed by the bench when recording or citing a result; algorithm changes can change a measurand even when the underlying raw run is unchanged.

The related EC Switch Explorer is an educational cross-section informed by the project’s parts data and measured curves. It is a visualization, not a manufacturing drawing, finite-element model, impact simulation, or acoustic simulation.

The practical conclusion

For Topre’s classic 30/45 convention, collapse force is the relevant rated peak-load quantity. For the work required to reach collapse, use PRE-COLLAPSE WORK. For a full low-rate stroke to detected bottom-out onset, use FULL-STROKE PRESS WORK. For the tactile event, report both force-drop magnitude and travel-dependent slope. None of those quantities alone is “the perceived weight.”

The purpose of the Dome Lab is not to replace subjective descriptions with a single score. It is to give each physical question a named, reproducible measurement—and to keep the measurement separate from conclusions that require a controlled human study.

Sources and method notes

1. T. Nagashima / Topre Corporation, Keyboard, Japanese Patent Application Publication JP2001-216070A (2001). The abstract identifies 45 gf- and 30 gf-centred press-load peaks and a switch-on region after the peak. JST J-GLOBAL record.
2. Topre Corporation, REALFORCE Features. Current page listing key-press-force settings separately from APC actuation depths and stating that bottoming out is not required.
3. bluepylons, Open-Switch-Curve-Meter. Open-source hardware repository and Gen 2 documentation; source for the stage, step-count position method, load-cell design, and nominal 0.005 mm full-step increment.
4. Shin-Etsu Polymer Europe, Force–travel characteristic. Manufacturer terminology for peak force, contact force, end-stop force, travel, and click ratio.
5. ASTM International, ASTM F2592-16, Standard Test Method for Measuring the Force-Displacement of a Membrane Switch. Historical methodological context only; ASTM marks it withdrawn in 2023 with no replacement.
6. A. Thompson and B. N. Taylor, Guide for the Use of the International System of Units (SI), NIST Special Publication 811. Unit basis using standard gravity, 9.80665 m/s².
7. Topre Corporation, JP2012-138254A, Keyboard switch. Annular plunger/housing cushioning-member placement and the illustrated PET/foamed-urethane embodiment.
8. J. Hyeong and J. Lee, “Effects of Design Parameters of Haptic Profiles on the Feedback Clarity at a Switch,” Journal of the Ergonomics Society of Korea 40(3), 161–171 (2021), doi:10.5143/JESK.2021.40.3.161.
9. M. L. Nagurka and R. W. Marklin Jr., “Measurement of Stiffness and Damping Characteristics of Computer Keyboard Keys,” Journal of Dynamic Systems, Measurement, and Control 127(2), 283–288 (2005), doi:10.1115/1.1902823. Author-hosted paper.
10. BuddyOG / Unreal Keyboards, Topre Force Curves — The Dome Lab data repository. Project raw data, viewer, and version history.

Method version documented here: metrics-v4.1 (baseline-relative ramp-10-90) + travel-v3 (step1pct) + runfilter-v1. Project measurements cited in the worked example and silencing-ring correction are Dome Lab data, not independent replications.

© 2026 Brian “BuddyOG” Gebo — Unreal Keyboards. All rights reserved.