Smartphone Vibration Motor: X-Axis Linear Motor vs Z-Axis Eccentric

That crisp, springy click you feel on a flagship keyboard is not software magic. It is a precisely tuned mass-spring system driven by a high-voltage waveform. Most teardowns gloss over it, but the choice between an X-axis LRA, Z-axis LRA, and the legacy ERM explains why one phone feels premium and another buzzes like a 2014 pager.

In the last three years, Android flagships have migrated aggressively to large X-axis linear resonant actuators (LRAs). Still, cost-sensitive designs and compact foldables often retain Z-axis LRAs — and ultra-budget boards still ship the eccentric rotating mass (ERM) motor. If you are designing haptics or evaluating BOM tradeoffs, the physics, driver, and integration differences matter far more than the marketing sheet.

1. How Smartphone Haptics Actually Work

All three motor types convert electrical energy to mechanical vibration, but via completely different mechanisms.

ERM (Eccentric Rotating Mass): A DC motor spins an off-center weight. The centrifugal force shakes the phone. Simple, cheap ($0.30–$0.60), but inertia is high. Start time ~40–60 ms, brake time ~60–100 ms. You cannot shape a waveform — only on/off and voltage amplitude.

LRA (Linear Resonant Actuator): A magnet mass suspended on springs inside a coil. Apply AC at the resonant frequency (F0) and the mass oscillates linearly. LRAs are closed-loop resonant systems with fast transients and precise waveform control. They are divided by axis of motion.

2. Z-Axis LRA: The “Taptic Tap”

Z-axis LRAs dominate mid-range phones because they are thin (2.5–3.2 mm) and cheap to integrate. The mass moves along the Z-axis — perpendicular to the screen — like a tiny piston knocking against your fingertip. Typical specs from Samsung AAC 1016 or Jinlong Z6261:

• Resonant frequency: 205–235 Hz (higher = sharper but thinner)
• Mass: 0.8–1.5 g, Gpp: 0.8–1.2 G at 2.0 Vrms
• Rise time: ~18–25 ms, Brake time with active braking: ~30 ms
• Dimensions: ~10×10×3.0 mm or 12×12×2.6 mm (coin-type)

The feel is best described as a tap. It is precise for single events — a notification thud or camera shutter — but struggles with rich effects. Because the stroke is short (~0.25 mm) and F0 is high, the decay is fast and over-ring is minimal. On the downside, the human palm is far less sensitive to Z-axis vibration than lateral shear, so the perceived intensity is weaker despite similar acceleration. You also get more audible “tok” leakage through the mid-frame.

Power is modest: a typical haptic pop draws ~80–110 mA peak for 10–15 ms. This matters on thermally constrained devices — our A18 Pro deep dive noted how Apple budgets transient haptic current against SoC boost power.

3. X-Axis LRA: The Flagship Standard

X-axis LRAs — AAC 1225, 1216 X, Samsung 2030-series, NFP 0832 — move a heavier mass laterally, parallel to the display glass. The chassis becomes the radiator.

• Resonant frequency: 135–175 Hz (lower frequency = richer, deeper)
• Mass: 3.0–5.8 g, Gpp: 1.4–1.9 G at 2 Vrms, up to 3.0 G transient
• Rise time: 6–10 ms, Brake time with closed-loop: 12–18 ms
• Dimensions: ~12×25×3.6 mm (rectangular brick)

That 3–5× mass increase is everything. Acceleration is force over mass, but perceived haptics scale with transient energy: E ∝ m·x²·f². X-axis motors store dramatically more kinetic energy and couple efficiently into the phone’s X-bending mode. The result is a broad, skin-stretching sensation rather than a localized poke.

Designers can also render HD haptics: multiple harmonics, texture simulation, and tight sync with audio. This is why GPU-driven physics demos and keyboard libraries like Samsung’s Haptic Composer or Apple’s Core Haptics explicitly target X-axis profiles. A classic Android “pop” waveform on an X-axis LRA is a 7 ms rise, 9 ms sustain at F0, then an inverted braking pulse — impossible to cleanly render on an ERM.

4. Head-to-Head

From a system view, the cost delta is not just the motor. X-axis LRAs demand a smarter driver (TI DRV2624, Cirrus CS40L25-B, or Dialog DA7282), a dedicated 5–8V boost rail, and mechanical tuning — foam gasketing, mid-frame isolation, and mass balancing to avoid camera module resonance. That is why many camera autofocus stacks are validated against haptic sweeps during EVT.

5. Driver IC and Waveform — Where Feel Is Made

The motor is dumb metal without its driver. Modern haptics are defined in software waveforms, similar to audio samples.

Two details matter here. Resonant tracking: F0 drifts ±8 Hz with temperature and aging. Drivers like CS40L25 run auto-resonance tracking via back-EMF zero-crossing, adjusting drive frequency every cycle — without it, a Z-axis motor driven 10 Hz off-resonance can lose 40% amplitude. Overdrive and braking: To hit that 6 ms rise, the driver overdrives to 7–8 Vpk for 2–3 ms, then slams an inverted phase to cancel the mass. On ERMs this is impossible; the rotor just coasts.

This tight loop is analogous to on-device LLM inference: latency-sensitive, power-spiky, and best handled by a dedicated co-processor rather than the AP. Many flagships keep the haptic driver on the always-on island so taps still feel instant when the AP is gated.

6. Integration, Acoustics, and Reliability

An X-axis motor can make a phone sound hollow if poorly mounted. The brick excites the mid-frame’s first bending mode (typically 160–190 Hz) — intentionally. But if the back glass or camera deco is not damped, you hear a rattling “boing” at 300–500 Hz harmonics. Top OEMs solve this with pressure-sensitive adhesive rings, silicone gaskets, and intentional mass loading of the frame near the motor pocket.

Z-axis LRAs are more forgiving: their energy couples out-of-plane and dissipates quickly through the display stack. The tradeoff is that screen protectors and cases attenuate Z-axis feel more than X-axis feel. In drop tests, X-axis motors also stress solder joints more — the 5g mass at 1.8G creates ~0.09 N cyclic force; over 100,000 clicks that is a fatigue concern. Most vendors now use board-to-board flex with strain relief, not direct SMT.

Finally, consider battery impact. A day of heavy typing (2,000 haptic events × 15 ms × 90 mA avg) costs ~0.75 mAh — negligible. But poorly tuned continuous effects in games (40 ms buzzes) can draw 150 mA for seconds. Good haptic UX caps long effects and uses amplitude scaling based on battery state.

7. Which Should You Choose?

Choose X-axis LRA if you are building a flagship, gaming phone, or any device where typing, gestures, and UI feedback are differentiators. The BOM hit ($4–$5 extra vs Z) pays back in perceived quality scores — blind tests consistently rate X-axis devices 0.8–1.2 points higher on 5-point haptic quality scales. It also future-proofs you for Android’s Rich Haptic APIs.

Choose Z-axis LRA when thickness (<7.5 mm), weight, or cost is constrained — think slim foldables, mid-range A-series, or compact wearables companions. It still beats ERM dramatically in latency and power, and newer Z-axis parts (AAC 1027) have narrowed the gap with lower F0 (~185 Hz) and active braking.

Avoid ERM for new designs unless the target is sub-$120. ERM supply is shrinking, driver support is legacy, and Android 14+ haptic compliance tests will flag ERM devices as non-HD capable. If you must use ERM, isolate it from the camera OIS — the imbalance can introduce 2–5 µm jitter in long-exposure shots.

The bottom line: the motor axis is not an esoteric spec — it is a direct proxy for how “expensive” a phone feels in the hand. If you are doing a teardown, flip the board and look at the motor pocket. Its dimensions tell you the design intent before you even power the device on.