Silicone Keypad Matrix Design: Pin Count, Ghosting and Multi-Key Input

A silicone keypad matrix should be defined as part of the electrical interface, not as a property of the silicone rubber itself. The molded keypad provides the moving keys and conductive contacts; the row/column matrix, connector pinout, pull network, scan method, logic thresholds, and multi-key behavior belong to the circuit and controller design.

For a matrix with R rows and C columns, the keypad can provide up to R × C switch positions using R + C matrix signal lines. That pin reduction is useful, but it creates a design obligation: the OEM team must decide which key combinations are valid, which combinations may be ignored, and whether any required combination can become electrically ambiguous before the PCB, FPC, connector, or host-board pinout is released.

A useful matrix review therefore starts with the approved key-combination list, not with the lowest possible pin count.

1. Define Single-Key and Multi-Key Use Cases Before Choosing the Matrix

Start by classifying the expected user inputs.

A matrix is attractive because it reduces conductors. A row/column matrix uses one wire per row and one per column; pressing a key connects its row and column. A 3 × 3 matrix therefore represents nine intersections with six row/column lines; a 4 × 4 matrix represents sixteen intersections with eight lines.

That arithmetic does not prove that the matrix supports every multi-key combination. Pin efficiency and rollover behavior are separate requirements.

If the product has only a few critical keys and simultaneous detection is mandatory, direct GPIO inputs or another explicitly multi-key-capable architecture may be more appropriate than forcing every key into the smallest possible matrix.

2. Map Rows, Columns and Connector Pins on the Drawing

The released drawing should show the electrical key map and the connector map together. Do not leave row/column naming to firmware interpretation after tooling.

Illustrative 3 × 3 Matrix

The following example is illustrative only. It does not represent a JASPER customer design or a tested production circuit.

                 C0             C1             C2
                 |              |              |
R0 ----------- [K00] -------- [K01] -------- [K02]
                 |              |              |
R1 ----------- [K10] -------- [K11] -------- [K12]
                 |              |              |
R2 ----------- [K20] -------- [K21] -------- [K22]

For this example, define the controller side as:

• R0, R1, R2: sensed inputs with pull-ups.
• C0, C1, C2: scanned outputs.
• One column is driven low during its scan window.
• The other columns are placed in the controller-approved inactive state, commonly high impedance in a simple low-going scan.
• A closed key connects its row to the currently scanned column.

An illustrative six-pin connector could therefore be:

LED power, LED control, shield, ground, ESD structures, or other circuits are not part of the six matrix signals unless the approved schematic explicitly makes them so.

The drawing package should also state connector orientation, pin numbering view, net names, NC pins, shared returns, and who owns the controller. A keypad supplier should not have to infer whether “R1” means a physical top row, MCU port name, or connector position.

3. Define the Scan State, Floating Lines and Logic Margin

A matrix only works when the scan states are unambiguous.

A basic scan drives one axis and reads the other with pull-ups. For example, row inputs can use pull-ups while columns are swept with one column driven low at a time. The specific polarity may be reversed in another design, but the drawing and firmware must agree.

For each row and column, define:

• input or output state during idle;
• active scan level;
• inactive-line state;
• pull-up or pull-down source and tolerance;
• input high/low thresholds;
• output low/high limits;
• cable, connector and contact resistance included in the path;
• leakage and parasitic capacitance;
• scan period, precharge or settling time;
• debounce and key-release logic.

Do not assume that an inactive scan line can always be driven to the opposite logic level. With multiple keys held, pressed switches can connect matrix lines together. The controller design must avoid uncontrolled output contention or provide an approved protection/current-limiting method.

Input threshold margin also matters. A carbon pill, PCB contact, connector and trace create a real closed path rather than an ideal zero-ohm switch. JASPER’s carbon-pill guide therefore treats the closed contact as part of the complete input circuit and recommends defining the pull network, thresholds, inactive-line state, scan rate and settling behavior instead of approving the rubber contact in isolation.

The row/column RC time constant affects sampling; the ADP5586, for example, provides configurable precharge time. That is a useful reminder for long tails, higher resistance pull networks, external multiplexers, ESD parts, or high line capacitance: sample after the electrical node has had time to settle, not merely after the firmware changes a GPIO register.

4. Identify Ambiguous Key Combinations Before PCB Release

Ghosting is a matrix topology problem. It is different from contact bounce.

Ghosting has a characteristic failure pattern: with certain three-key combinations in a row/column matrix, the electrical path can make an unpressed fourth key appear active.

Using the same 3 × 3 example, assume these three keys are physically pressed:

• K00 = R0/C0
• K01 = R0/C1
• K10 = R1/C0

Now scan C1 low while C0 is inactive.

C1 LOW
  |
 K01  (pressed)
  |
 R0
  |
 K00  (pressed)
  |
 C0   (inactive path)
  |
 K10  (pressed)
  |
 R1

R1 can be pulled toward the active state through the three closed switches. The scanner may therefore infer that K11 (R1/C1) is also closed even though the user did not press it.

That is the phantom fourth corner.

Whether a particular controller reports, suppresses, or flags that condition depends on its scan architecture and firmware, but the basic ambiguity must be reviewed before layout approval. Right-angle combinations can create ghosting. Arrange keys likely to be pressed together so they avoid that problematic pattern.

Debounce does not solve this. Debounce filters short transitions from a real contact event. Ghosting can create a stable but electrically ambiguous state.

5. Review Controller and Circuit-Level Mitigations

There is no single mitigation that fits every OEM product.

Remap the matrix

If the required key chords are known, assign them so the combinations do not form ambiguous rectangles. Frequently combined keys can sometimes be placed on the same row or same column, or otherwise arranged so the approved combinations do not create the ghost path.

This is often the lowest-component solution, but only if the key map is frozen and tested.

Use more pins where required

A larger, less densely populated matrix can separate critical combinations. For a small number of safety-critical or modifier keys, dedicated inputs may be clearer than saving one or two connector pins.

Use controller-supported ghost handling

Some keypad controllers can detect suspicious multi-key states and suppress or flag them. That can prevent a false command, but it does not automatically tell the product which physical keys were actually pressed. Confirm whether the requirement is “do not issue a phantom command” or “identify every simultaneous key.”

Add a validated isolation architecture

If independent multi-key detection is mandatory, the circuit may need per-key isolation or another architecture designed for rollover. Any isolation components must be shown in the PCB/FPC schematic and validated against scan direction, voltage drop, input thresholds, leakage, ESD network, and available assembly space. Do not add an isolation component late and assume the original thresholds still have the same margin.

Change the architecture

For a limited key count, direct GPIO may remove matrix ambiguity. For larger interfaces, a dedicated keypad scanner or I/O expander can reduce the host-processor pin burden while keeping scan behavior inside a defined controller function. The component choice still has to satisfy the approved simultaneous-key requirement.

6. Test the Approved Key Combination List

Single-key continuity is not enough for a keypad that has multi-key functions.

The validation plan should exercise the exact approved combinations in both press and release sequences. Include the production-intent keypad, circuit, connector, host input configuration and enclosure support.

Use a controlled table such as this:

Also test the combinations that are not permitted. The required product behavior might be “ignore the newest key,” “report an error,” “retain the first two keys,” or “no action.” That decision belongs in the product requirement, not in an undocumented firmware assumption.

Where timing is important, record the scanned line, sensed line, input voltage and decoded key event together. This distinguishes contact bounce, slow settling, threshold margin and true ghosting.

7. Freeze the Key Map, Pinout and Responsibility Boundary Together

Before releasing the keypad drawing, PCB artwork or connector, approve these items as one package:

1. key ID and function table;
2. required single-key and multi-key combinations;
3. row/column assignment;
4. connector pinout and orientation;
5. controller-side input/output state table;
6. pull-up/pull-down and logic thresholds;
7. scan and settling timing;
8. ghost-handling or isolation strategy;
9. debounce and release rules;
10. multi-key validation table and acceptance behavior;
11. supplied boundary: rubber keypad only, keypad + circuit, or tested assembly.

For the conductive interface itself, review JASPER’s Conductive Rubber Keypads page and Carbon Pill Resistance guide. If the project needs the rubber keypad, circuit, connector and functional test controlled as one module, the Silicone Keypad Assemblies route is the more relevant sourcing boundary.

The key point is simple: do not approve the silicone tool and postpone the row/column decision. Matrix assignment can change connector count, circuit routing, controller firmware and the behavior of required multi-key commands.

Frequently Asked Questions

Review Your Contact and Circuit Requirements

Use request a keypad engineering review to submit the key layout, conductive-contact drawing, row/column proposal, host schematic, connector pinout, required multi-key combinations, input thresholds, scan/debounce requirements, assembly boundary, annual or project quantity, equipment use and validation expectations.

Items that are not final can be marked TBD. The important point is to expose the open decisions before tooling and PCB release rather than hiding them inside a generic “matrix keypad” note.