Haptic Feedback for Wearables: LRA vs ERM vs Piezo

Concept render

Haptic feedback in a wearable usually comes from an ERM motor, a linear resonant actuator (LRA) or a piezo actuator. ERMs are cheap and easy to drive. LRAs respond faster and use less energy, but only near their resonant frequency, so they need a driver that tracks it. Piezo responds fastest and needs high voltage.

A vibration that people feel, and that does not drain the battery or rattle, depends on three choices: the actuator, the driver chip and the way the part is mounted. This guide is for founders and engineering leads choosing haptics for a wearable or a companion device. Every measured figure comes from a chip vendor application note or an actuator maker's guide, linked below; the battery example in the power section is our own arithmetic on those figures.

Test figures describe the parts in each source. Measure your own actuator in your own enclosure.

What is haptic feedback, and which actuators produce it?

Haptic feedback uses vibration to tell a person something through touch. TI describes it as mechanical feedback that varies frequency, amplitude, duration and direction to simulate events and effects (TI SLOA194A, as of November 2026).

Three actuator types cover most small devices:

• ERM (eccentric rotating mass). A small DC motor with an off-center weight. It is easy to drive and low cost.
• LRA (linear resonant actuator). A magnet and mass on a spring, moved by a coil. It vibrates along one axis and only works well near its resonant frequency.
• Piezo actuator. A ceramic that flexes under voltage. It is the fastest of the three and needs a high-voltage driver.

TI's note on mobile and wearable devices also covers BLDC coin modules, which put the motor driver inside the actuator. TI says these modules cannot be braked, cannot play clicks, and will not move below the power-up voltage of the internal IC, around 2.9 V (TI SLOA207A, as of November 2026).

LRA vs ERM vs piezo: how do they compare?

The table draws on TI's comparison table and actuator descriptions in SLOA194A (revised January 2022), with drive notes from TI SLOA207A and Precision Microdrives AB-028, all as of November 2026. Response time, acceleration, noise and energy are TI's rounded figures for typical parts.

Actuator maker Precision Microdrives adds wear and size data in its comparison guide. At rated voltage, it lists brushed ERMs with a mean time to failure of 100 to 600 hours and LRAs at 1,000 hours or more. It lists LRA diameters of 8 to 10 mm, rated frequencies of 150 to 205 Hz, about 2 V RMS rated voltage and vibration on a single axis (AB-028, as of November 2026). TI notes LRAs as thin as 2.5 mm with 8 mm diameters (SLOA207A).

An ERM still fits a simple buzz when no haptic driver IC is planned. Precision Microdrives says an ERM is much easier to drive if you are not using a haptic chip (design guidance for haptic wearables, as of November 2026).

Why does a linear resonant actuator need a driver that tracks resonance?

An LRA is a spring and mass. It gives strong vibration only in a narrow band around its resonant frequency. The sources describe that band in similar terms. TI's SLOA194A says to drive within about ±2 Hz of resonance, and its figure shows vibration dropping 25% over 2.5 Hz. TI's SLOA207A says performance drops at offsets of typically 3 to 5 Hz. Precision Microdrives says the drive signal should stay within ±5 Hz of the rated resonant frequency (AB-028). All as of November 2026.

What moves the resonant frequency

TI lists manufacturing tolerance, aging, temperature and mechanical mounting as causes of drift (SLOA194A). SLOA207A adds the mass of the product the LRA is mounted to, how the product is held, and drive voltage: at maximum voltage, many LRAs shift several hertz from mechanical compression. Precision Microdrives says the resonant frequency changes with what the LRA is mounted to (AB-003, as of November 2026). A bench frequency may not hold inside a housing.

What a closed-loop driver adds

A closed-loop driver reads the LRA's back-EMF and adjusts its drive frequency each cycle. TI tested 21 LRAs with a data sheet resonant frequency of 230 Hz. Driven at a fixed 230 Hz by a DRV8601 without auto-resonance, they gave mixed acceleration. Driven by a DRV2605 with auto-resonance tracking, average acceleration rose 51.8%, average power fell 20.6%, and power per g improved 45.0% (TI SLOA188A, revised November 2021, as of November 2026).

Braking matters as much as frequency. Without it, TI says an actuator keeps vibrating until its own damping stops it, which can last hundreds of milliseconds (SLOA207A). Braking an LRA means driving it 180 degrees out of phase at resonance. In TI's example, braking stopped the LRA 75 ms sooner, which reads as a crisper click (SLOA188A).

One driver as an example

The DRV2605L data sheet (SLOS854D, revised March 2018) shows what to look for in any haptic driver. As of November 2026 it lists a 2 V to 5.2 V supply, ERM and LRA support selected by a register bit, automatic overdrive and braking, and resonance tracking that locks on after half a cycle. If the LRA gives no valid back-EMF, the chip falls back to open-loop drive at a set frequency. It also includes a licensed effect library with 6 ERM libraries and 1 LRA library (TI SLOS854D). These figures apply to the DRV2605L only.

How much power does a haptic motor use?

TI measured energy per effect on one ERM, one LRA and one piezo module. Each actuator carried a 100 g metal mass to simulate a smartphone. The table is from SLOA194A Table 3-1, as of November 2026, with acceleration held near 0.9 to 1.0 g.

From this data, TI concludes that the LRA can save 60% to 80% of the energy an ERM uses. Piezo uses the least on short effects and about the same as the ERM on long buzzes. With auto-resonance on, TI's LRA buzz reached 1.64 g at 28.4 mA/g. With it off, the same LRA reached 1.00 g at 51.8 mA/g (SLOA194A, as of November 2026).

The idle current can outweigh the effects

A worked example makes the point. TI uses a 110 mAh battery as an example of a small wearable cell (SLOA207A). Take 100 LRA clicks a day at 0.57 µAh each. That is 57 µAh, about 0.05% of 110 mAh. Now look at the driver. The DRV2605L lists 0.5 mA typical quiescent current when it is enabled with standby off, and 4.1 µA typical in standby, at 3.6 V and 25 °C. Left awake all day, 0.5 mA becomes 12 mAh, about 11% of that battery. In standby it is about 0.1 mAh. Firmware that forgets standby costs far more than the vibrations. Put the driver's idle state in your power budget and measure it on the board; the guide on testing battery life claims covers how. For testing a small device with a battery emulator, see Battery emulator for device testing: what it does and when to use it.

How do mounting and the enclosure change the feel?

The actuator shakes the whole device, so the enclosure is part of the haptic system. Four points from the sources shape the mechanical design:

• Mass changes the result. TI's energy tests used a 100 g load. A ring or a clip weighs much less, and SLOA207A lists product mass as a cause of LRA frequency shift. Ask at what test mass any acceleration figure was measured.
• Give an ERM room to spin. Precision Microdrives says the rotating mass must not be impeded, and a casing that sits too close causes an audible rattle as the mass hits it each rotation (source, as of November 2026).
• Calibrate in the final assembly. The DRV2605L data sheet says to evaluate system performance during final assembly, because auto-calibration can affect actuator performance and behavior.
• Mind the axis. An LRA vibrates on one axis. Orient it so that axis moves the part of the device that touches skin.

Smaller is not always better. Precision Microdrives says a smaller motor has less torque and draws more current for the same amplitude, and advises testing a range of options early in prototyping. Adhesive and spring contacts also have to survive drops; see the guide on drop testing electronics.

How do you design silent alerts people notice?

A silent alert has two jobs: be felt, and be told apart from other alerts. Precision Microdrives says the level of vibration that is enough depends on whether the product is worn over clothes, during activities, and on its total weight. None of our sources gives a perception threshold, so treat this as a test you run on people.

• Vary the pattern as well as the strength. Change count, length and spacing. Braking stops each pulse quickly, which helps short pulses stay distinct.
• Keep long buzzes for urgent events. In TI's table, an LRA buzz used about six times the energy of a click.
• Test on the body. Try the wrist, the chest and over a jacket, while walking, with people outside the team.
• Plan for the table. A companion device may sit on a desk. TI lists device position, such as gripped tightly or set on a table, among the causes of resonance drift (SLOA194A). Pair vibration with light or sound where the device is not worn.

For older users, pair every alert with light and sound as well. The guide on wearable technology for seniors covers that in more detail.

What should you ask your hardware team?

Bring this list to the first electronics review, and write down who answered each item.

• Which actuator type and model is in the design, and what is its rated resonant frequency and rated voltage?
• Which driver IC runs it, and does that driver track LRA resonance in closed loop?
• Can the driver brake the actuator? A single FET cannot brake an ERM.
• Was the level calibration run with the actuator mounted in the final enclosure?
• Are calibration results stored per unit, rerun at power-up, or burned into the driver once?
• What does the driver draw when idle, and does firmware put it in standby between effects?
• At what test mass was any acceleration figure measured, and how heavy is our device?
• Which alert patterns have been tested on the body, over clothing and during movement, by people outside the team?

Companion and elder care devices may need to alert someone without sound. For that category, see AI companion device development.

If you want an actuator, driver and alert plan checked against your enclosure and battery, tell us about your device.

Sources. All pages accessed November 6, 2026. Test figures apply to the parts and setups in each source. Confirm current values before you decide.

• Texas Instruments, Haptic Energy Consumption, SLOA194A (May 2014, revised January 2022) (link)
• Texas Instruments, Haptic Implementation Considerations for Mobile and Wearable Devices, SLOA207A (November 2014, revised November 2021) (link)
• Texas Instruments, Benefits of Auto-Resonance Tracking, SLOA188A (October 2013, revised November 2021) (link)
• Texas Instruments, DRV2605L data sheet, SLOS854D (revised March 2018) (link)
• Precision Microdrives, AB-028: Vibration Motor Comparison Guide (link)
• Precision Microdrives, AB-003: Driving Linear Resonance Vibration Actuators (link)
• Precision Microdrives, Design Guidance for Haptic Wearables (link)