Most haptic hardware in phones, watches, game controllers, and touch panels relies on one of four mechanisms. Eccentric rotating mass (ERM) motors and linear resonant actuators (LRAs) handle the bulk of everyday vibration, piezoelectric actuators show up where a fast, sharp click matters, and a smaller family of nonlinear and impact designs aims to deliver a single tap with nothing ringing after it. Each one turns current into motion in its own way, and that mechanism sets the limits on what a device can make you feel.
Eccentric rotating mass motors
An ERM is a small DC motor with an off-center weight on its shaft. As the weight spins, the imbalance shakes the motor and whatever it's mounted to, and the frequency of that shaking matches the rotation speed. ERMs come as small cylinders, the classic pager motor, or as flat coin motors a few millimeters thick that sit behind a circuit board.
Their appeal is simplicity and cost, since an ERM runs on plain DC and a transistor and a microcontroller pin are enough to drive it. The catch is that speed controls frequency and strength together, so turning the voltage down for a gentler buzz also lowers its pitch, and designers can't set the two independently. ERMs are also slow to start and stop because the motor has to spin the weight up from rest and then coast down, which commonly takes tens of milliseconds unless the driver overdrives the start and applies reverse voltage to brake. That lag is a big part of why ERM feedback tends to feel soft and rumbly.
Linear resonant actuators
An LRA moves a magnetic mass back and forth along one axis on a spring, pushed by a voice coil much like the one in a loudspeaker. Because the mass sits on a spring, the assembly has a resonant frequency where it moves the most for the least power, and the driver feeds it an alternating signal tuned to that frequency. Coin and bar LRAs in phones and wearables commonly resonate somewhere between about 150 and 250 Hz, close to the band where the skin's vibration receptors are most sensitive.
LRAs start and stop faster than ERMs, use less power for the same output, and can change strength without changing frequency, which is why they've replaced ERMs in many phones and watches. The tradeoff is a narrow operating band. Output falls off steeply away from resonance, and the resonant frequency itself varies from part to part and drifts with temperature, wear, and mounting, so better drivers track it continuously. An LRA also keeps ringing for several cycles after the drive stops unless the driver brakes it, which the guide to haptic waveforms and drivers explains in detail.
Some recent phones and game controllers use larger linear actuators with a wider usable frequency range. These wideband parts give software room to render a wider variety of clicks and textures, at the cost of more internal space and a more capable driver.
Piezoelectric actuators
A piezoelectric ceramic changes shape when a voltage is applied across it. Haptic piezo parts are usually built as benders, where thin ceramic layers bonded to a metal plate flex when driven, or as multilayer stacks that expand by a tiny amount with considerable force. The ceramic responds almost immediately, so a piezo actuator can reach full output within a few milliseconds or less and reproduce a much wider range of frequencies than an LRA.
The costs are voltage, complexity, and price. Haptic piezo actuators typically need drive signals of several tens of volts, and some designs run above 100 volts, which means a boost converter and a dedicated driver, and the ceramic behaves like a capacitor that has to be charged and discharged on every pulse. Displacement is small, so piezo works best pushing on a stiff surface such as a trackpad, a button, or a touch panel, where a short, sharp movement reads as a click. The ceramic is also brittle, and the mechanical design has to protect it from shock and from bending loads it wasn't built to take.
Nonlinear and tap actuators
The fourth group is less standardized and covers designs built to produce a single, discrete tap. A conventional actuator behaves like a mass on a linear spring, so any push sets it ringing at its natural frequency, and nonlinear designs change that relationship. Bistable mechanisms snap from one stable position to another and stay there, impact designs drive a small mass into a stop or the housing, and others shape the spring force or add damping so the moving part settles in a single stroke.
The result is a sensation with a clear start and end, closer to a fingertip tapping the skin than to a buzz. These designs give up some flexibility to get it. A snap-through mechanism tends to deliver a force set by its geometry, so varying intensity is harder than with an LRA, and the nonlinear behavior can be sensitive to manufacturing tolerances. The article on nonlinear actuators and wearable interfaces goes deeper into the mechanics and the production challenges, and the evolution of haptic feedback traces how the field moved toward taps.
How the four compare
The table summarizes typical behavior for each family. Treat the figures as broad ranges, because a specific part's datasheet, its driver, and the product it's mounted in all shift the numbers.
| Actuator | How it moves | Drive | Speed | Frequency behavior | Typical uses |
|---|---|---|---|---|---|
| ERM | Off-center weight spins, shaking in the plane of rotation | DC voltage from a transistor or H-bridge | Slowest, tens of milliseconds to start and stop | Frequency and strength rise together with speed | Budget phones, controllers, toys, pagers |
| LRA | Mass on a spring moves along one axis | AC signal near resonance, usually 150 to 250 Hz | Faster, but rings without active braking | Strong near resonance, weak elsewhere | Phones, smartwatches, fitness bands |
| Piezo | Ceramic bends or expands | High voltage from a boost driver | Fastest, a few milliseconds or less | Wide usable band | Trackpads, touch panels, buttons |
| Nonlinear or tap | Snap-through, impact, or shaped spring | Short pulse, varies by design | Single event built to settle without ringing | Not a continuous vibration source | Discrete taps in wearables and research devices |
Choosing an actuator
For a low-cost product that only needs simple alerts, an ERM is still a reasonable choice, especially when the alerts run long enough that slow start and stop times don't matter. For a phone or wearable where notifications should feel crisp and battery life counts, an LRA paired with a closed-loop driver is the default, and a wideband linear actuator is worth the extra space if the software team plans a large library of effects.
When a flat surface has to feel like a physical button, piezo or a wideband actuator moving the whole surface is the better fit, since the click needs a fast, short impulse. Tap-based nonlinear actuators make sense when discrete taps are the point, for example when a product encodes information in tap rhythm, and the team can live with fewer off-the-shelf options. Whatever the choice, test in the real enclosure, because the same actuator feels different in a light watch and a heavy phone. The guide to measuring haptic quality covers how to run those tests.