Self-capacitance measures an electrode’s apparent capacitance to the sensing circuit’s ground; mutual capacitance measures coupling between transmit (TX) and receive (RX) electrodes. Choose between them by proving the required controls in their installed stack—not by equating self-capacitance with buttons and mutual capacitance with screens.
Neither label establishes superior sensitivity, lower cost, or waterproof operation. For an OEM custom capacitive touch panel project, compare supported controller configurations against the same functions, enclosure, power states, and acceptance criteria.
- What Each Method Measures
- Compare Six Engineering Dimensions
- How Electrode Geometry and Routing Differ
- What Water and Nearby Ground Can Change
- How Buttons and Coordinate Sensors Use These Methods
- What to Confirm with the Controller Owner
- Compare Both Methods on the Same Fixed-Key Task
- Request a Design Review
What Each Method Measures
Self-capacitance tracks the electrode’s ground-referenced load; mutual capacitance tracks a defined TX–RX coupling path.
Microchip’s self-capacitance model includes sensor, parasitic, and ground-return capacitances. A finger introduces an additional return path and increases apparent capacitance. Local circuit ground is not necessarily bonded to earth, so battery operation and an externally connected setup are not equivalent measurement conditions.
Microchip’s mutual-capacitance model describes competing finger effects: additional coupling between electrodes and diversion of charge through the user’s return path. The latter commonly reduces measured mutual coupling. Defining both electrodes therefore does not remove the influence of the user, ground, or power arrangement.
Record untouched and touched readings, the controller’s delta definition, and signal polarity. Do not assume a physical capacitance decrease must appear as a negative displayed delta: the controller converts capacitance into its own measurement representation. TI’s conversion-count explanation shows why gain and offset settings belong with that record.
Compare Six Engineering Dimensions
Use the comparison to identify design evidence, not to assign a universal winner. The distinctions draw on the measurement models above and TI’s sensing architecture guidance.
| Engineering dimension | Self-capacitance | Mutual capacitance |
|---|---|---|
| Measured quantity | Apparent capacitance from a sensing electrode to circuit ground. | Coupling between a selected TX and RX pair. |
| Electrode relationship | A discrete key can use one sensing pad; independently sensed keys require channel mapping. | A key uses a TX–RX interaction region; electrodes may be shared across nodes. |
| Routing and parasitics | Pad, trace, connector, and surrounding conductors contribute loading and unwanted sensitivity. | Unwanted TX–RX coupling adds to the intended node; TX/RX loading to ground still matters. |
| Ground and water paths | User return coupling and liquid connections can change the reading. | Liquid can change electrode coupling or create a touch-like return path. |
| Controller support | Confirm sensing channels, input-loading limits, shield options, and concurrent-key logic. | Confirm TX/RX assignments, node capacity, scan timing, and supported rejection features. |
| Validation method | Log each key’s signal and unwanted adjacent responses through press, hold, and release. | Log each node and shared-electrode interactions through the same events and conditions. |
The last two rows are review requirements: obtain answers and evidence for the actual configuration rather than treating a family-level feature list as approval.
How Electrode Geometry and Routing Differ
Keep the same user-facing control task, but let each method use geometry appropriate to its controller. Identical icons do not require identical copper patterns.
Consider an illustrative panel with Power, Mode, Increase, and Decrease keys. Hold the icon positions, cover stack, enclosure, and operating requirements constant. Compare separate sensing pads with a TX–RX layout inside the same permitted footprint. Do not force both candidates to reuse one electrode pattern merely to make them look comparable.
For mutual sensors, TX–RX spacing and interleaving interact with the cover: Microchip’s touch-cover guidance explains that increasing cover distance changes coupling and the geometry needed to retain a useful touch signal. Check candidate layouts through the specified stack, not an uncovered board.
Figure 1 — Self-capacitance fixed key, illustrative schematic description. One sensing pad lies beneath a fixed icon and connects to a measurement channel. A simplified cross-section identifies the cover, finger coupling, local circuit ground, and capacitive return through the user and surroundings. Earth and local ground remain separately labeled.
Figure 2 — Mutual-capacitance fixed key, illustrative schematic description. Separate, non-contacting TX and RX conductors occupy the same icon footprint. Their facing or interleaved edges define the sensing region. A cross-section identifies coupling through the cover and the finger-related return path. These are conceptual layouts, not product photographs or production dimensions.
Check the entire path to the controller. TI’s routing guidance treats traces and connectors as electrical contributors, not neutral extensions. For mutual sensing, inspect unintended TX–RX adjacency outside the key as well as each conductor’s loading to ground.
Compare candidate artwork with installed tail routes, neighboring signals, and mounting conductors. Archive the electrode revision, pin map, connector, routing restrictions, and measured loading or calibration status. A layout change should remain visible in the comparison record.
What Water and Nearby Ground Can Change
Water tolerance depends on the available electrical paths, not just the measurement label. First distinguish local circuit ground, protective earth, a floating metal part, and a driven shield.
Both measurement models include return coupling through the user and the equipment. Test the intended battery or adapter arrangement, chassis connections, and realistic handling states; document whether debugging or communications cables are attached.
TI’s moisture discussion explains that an isolated droplet can increase mutual coupling and produce a change opposite to a normal touch. Liquid extending to a nearby ground-referenced conductor can instead create a touch-direction change in either method. An isolated-droplet demonstration therefore does not establish spill performance.
Compare isolated droplets, a film across adjacent keys, and a liquid path toward an actual bezel or grounded region. Record liquid identity, placement, amount, ground connections, sensor readings, and host events. State whether each condition requires continued operation, rejection, or temporary lockout.
A driven shield is not simply ground copper. Microchip’s driven-shield guidance distinguishes drive waveforms and warns that excessive shield coupling can impair calibration or signal quality. Confirm compatible hardware, waveform, loading, and geometry before adopting it.
Use the water and glove tuning guide for detailed state definitions. Keep enclosure ingress acceptance separate from wet-touch acceptance: require evidence for each rather than inferring one from the other.
How Buttons and Coordinate Sensors Use These Methods
The sensing method does not determine whether the interface delivers fixed commands or coordinates. Either method can support fixed-function controls with suitable hardware and processing.
Individually measured self-capacitance keys are not the same architecture as a row/column self-capacitance matrix. In the latter, simultaneous active rows and columns can leave ambiguous intersections. A mutual matrix measures distinct TX–RX nodes, addressing that particular ambiguity; it does not guarantee every required multi-touch behavior. TI’s matrix explanation describes this distinction.
For fixed keys, specify press, hold, release, permitted key combinations, suppression, and host events. For a projected-capacitive coordinate sensor, separately specify position, edges, tracking, touch count, and display mapping. Do not accept a coordinate demonstration as evidence that a particular command is safely handled. Ordinary touch keys must not substitute for required safety functions.
What to Confirm with the Controller Owner
Approve the exact controller configuration before freezing electrode artwork. Support must cover the selected pins, measurement mode, software, and required operating states—not merely the device family’s marketing description.
TI’s Design Center documentation illustrates configurable modes, pin assignments, scanning, and diagnostic access. Treat these as examples of questions to resolve, not features promised for every controller.
| Project condition | Question for the controller owner | Evidence to retain |
|---|---|---|
| Controller already selected | Does this device, package, and software support the proposed electrodes and mode? | Device identity, supported configuration, pin map. |
| Long tail or dense routing | What loading, resistance, settling, and coupling limits apply? | Layout review, calibration and scan evidence. |
| Metal bezel or changing power setup | Which reference, shield, and inactive-electrode states are supported? | Schematic and configuration for each state. |
| Water or an approved glove required | Is the target continued operation, rejection, or lockout, and how is recovery handled? | State requirements and matched test logs. |
| Simultaneous keys required | Which suppression rules operate, and are all required combinations reportable? | Combination and release-order tests. |
| Tight power or response budget | What are the complete scan schedule, wake behavior, and end-to-end delay? | Timed host events and measured power. |
| Production transfer planned | Who owns tuning files, programming, diagnostics, updates, and change approval? | Versioned release package and responsibility assignment. |
Assign the cover owner its dielectric stack and tolerances; the sensor owner its geometry and tail; the controller hardware owner pin and loading limits; and the firmware owner scanning, calibration, and event processing. Assign display integration to its electrical/optical owner and command interpretation to the host owner. Name who validates their interaction. The custom capacitive touch panel design guide provides the wider handoff framework.
Compare Both Methods on the Same Fixed-Key Task
Select the candidate that satisfies the same installed acceptance plan with documented margin and manageable implementation effort. A larger raw-count change is not a cross-platform performance ranking.
Use the four-key example as an illustrative comparison protocol, not a claimed test result. Hold external requirements constant; permit method-specific electrodes and tuning, recording every difference. Include the weakest intended input and the most disruptive realistic operating states.
TI’s SNR measurement tools expose filtering and data-collection choices, while its gain documentation shows that configuration changes measurement scale. Record the noise definition, acquisition window, gain, filtering, baseline, and thresholds with each result. Compare usable detection and rejection margin alongside response time—not counts from unlike configurations.
| Test on both candidates | Decision evidence | Self result | Mutual result |
|---|---|---|---|
| Dry and approved-glove operation at every key | Misses, false events, press/release timing. | ||
| Required key combinations and release orders | Independent reporting and correct host commands. | ||
| Installed power, display, and cable states | Noise, baseline movement, threshold margin. | ||
| Isolated liquid, cross-key film, and relevant ground path | Required operation or rejection; no unintended command. | ||
| Startup while touched or wet; subsequent removal | Agreed startup handling and recovery without spurious events. | ||
| Allowed cover, bond, and tail variation | Repeatable behavior across documented assembly variants. |
Illustrative recording template. Blank cells are for actual measurements, not implied passes.
Before testing, agree event limits, timing limits, repetitions, sample coverage, and recovery criteria. Preserve raw or diagnostic data where accessible, then align controller states with host timestamps. A good sensor trace does not replace evidence that the host executed—or rejected—the intended command.
Record the selected architecture, rejected alternatives, hardware and firmware revisions, worst observed condition, remaining restrictions, and approval owner. When both candidates pass, compare total supplied scope and maintenance obligations instead of assigning a cost advantage to the sensing principle.
Frequently Asked Questions
The following questions help turn an architecture comparison into a reviewable specification.
Does a two-key shortcut require mutual capacitance?
Not necessarily. Ask for a demonstration of both keys being reported independently with the proposed sensor layout, controller, and suppression settings. Test both release orders and confirm how the host interprets overlapping presses.
Can existing electrode artwork be reused with a different controller?
Treat it as a candidate, not an approved replacement. Have the new controller owner review pin mapping, loading, routing, shield drive, and scan timing, then repeat installed validation before releasing unchanged artwork.
Can one controller use both sensing methods?
Some platforms can. TI documents self- and mutual-capacitance sensors within one CapTIvate design. Confirm support on the exact device and software version, including shared resources, scan scheduling, and diagnostic access.
Is the design with the largest raw touch delta better?
Not by that number alone. Compare each design against its own noise and thresholds under matched conditions. Keep gain, filtering, sampling, and response-time settings with the result so a larger count is not mistaken for more usable margin.
What should replace “works with gloves” in an RFQ?
Specify glove manufacturer and model, fit, layers, condition, required controls, and accepted or rejected behavior. Include the actual cover stack and power configuration. Approval should apply to that documented combination, not to gloves as a category.
What can be reviewed before a controller is selected?
Provide the function map, cover stack, enclosure and ground arrangement, tail route, power modes, environmental states, and required output behavior. Use these inputs to screen controllers; keep electrode artwork provisional until electrical constraints are agreed.
How should quotes using different sensing methods be compared?
Request matching scope for the physical panel, controller hardware, firmware, tuning, fixtures, diagnostics, and installed testing. Separate recurring costs from development and change costs. Compare proposals that satisfy the same acceptance plan, rather than assuming either method is cheaper.
Request a Design Review
Submit a defined control task and its electrical boundary, not just finished icons.
Request a Design Review with the key map and required combinations; cover, adhesive, and enclosure drawings; candidate electrode files; controller part number and configuration, or selection constraints; tail and connector details; power and ground arrangements; required water and glove states; sample and annual quantities; and acceptance criteria. Identify which party supplies the sensor, controller, firmware, display, and host integration so the proposed scope can be reviewed consistently.
Ready to Start Your Project?
Tell us about your membrane switch, keypad, or graphic overlay requirements. Our engineering team will review your specifications and provide a detailed quote.
