How do devices provide haptic feedback?

Video game controllers, cell phones, wearables, and dozens of other consumer electronic devices make use of vibrotactile feedback to increase user engagement. There are three different types of hardware most frequently used to provide haptic feedback: eccentric rotating mass motors, linear resonant actuators, and piezoelectric actuators.

Eccentric Rotating Mass Motor

The eccentric rotating mass (ERM) motor is similar to a regular DC motor: it uses the magnetic field of a direct electrical current to move an object in a circle. Unlike a motor you would use for a toy car, the ERM moves a small weighted object (rotating mass) that is off-center from the point of rotation (eccentric). Due to the uneven centripetal force produced by the rotation of the mass, the entire motor will move back and forth to produce a vibration from side to side (lateral vibration). Like most DC motors, an ERM motor will run its current through windings attached to the shaft of the motor. Since the current will run inside of a magnetic field created by magnets installed inside of the motor, the magnetic field produced by the current will apply a force to the rotating load proportional to the current running through the windings—typically, the intensity of vibration produced by an ERM motor will change according to the supplied voltage at its terminals, following Ohm’s law. Since the mechanism of an ERM motor is very simple, they come in a variety of shapes, sizes, and operating characteristics.

The most common type of ERM motor inside of consumer electronic devices is the pancake shaftless vibration motor (pancake motor). Pancake motors are used ubiquitously in cell phones and video game controllers to inexpensively provide a perceptible vibration to a user. Other types often have the eccentric mass exposed in a cylindrical shape similar to a standard DC motor. Larger ERM motors used in industrial systems will often consist of a larger implementation of the same fundamental design.
In consumer electronic devices, an ERM motor will be mounted to the enclosure of a device to propagate the vibration along the exterior of the device gripped by a user. Since the motor can be driven using a DC current, a simple switching circuit can turn the motor on or off as needed. Such a basic switching circuit will frequently be implemented using a transistor with a protection diode. A transistor allows a microcontroller or integrated circuit to provide sufficient current to drive the motor, and a diode will protect the integrated circuit from a reverse voltage; since the motor is an inductive load, its deceleration can produce a kickback voltage that can damage sensitive digital circuits.

With a transistor and diode, the perceptible strength and vibration of the motor may be controlled using a pulse-width modulated (PWM) signal that rapidly turns the motor on and off with a specific duty cycle (the percentage of time the signal is active). As the duty cycle increases, the proportion of time during which the DC motor receives a driving current increases as well. As a result, the PWM signal allows the vibration produced by the motor to transition more smoothly between its on-off states, and also allows a designer to vary the intensity of vibration over time. However, the technique of pulse-width modulation comes with its own limitations. Since an ERM motor must move an eccentric mass using the current it is provided, the frequency and amplitude of vibration cannot be modified independently. Instead, the frequency with which the eccentric mass rotates will correlate linearly with the linear acceleration (force) of the eccentric mass. As a result, the duty cycle of a PWM signal driving an ERM motor will proportionally change both the frequency and amplitude of the device’s vibration.

As applications utilizing ERM motors require more precise feedback, the driver circuit for the motor may be modified to provide more precise control over the strength of vibration produced by the motor. Instead of a PWM signal provided to a transistor, a higher or lower voltage within the specified range of the motor will more precisely change the speed of its rotation, thus changing the frequency and amplitude of the vibration that is produced. In spite of this benefit of using a variable voltage source, it constrains the variation in the amplitude of vibration to the operating voltage range of the motor. The more precise control over frequency and amplitude comes at the cost of a narrower window within which the motor may be considered “on” or “off”.

Another common implementation of ERM motors driven by a digital circuit consists of an H-bridge instead of a transistor to drive the motor. Since a DC current may be provided to either of the two electrical terminals of the motor (one forward, one reverse), the motor may be driven in either direction. When an ERM motor is turned on, the eccentric mass rotates with kinetic energy that must dissipate through the enclosure of the device when it is turned off. Although this dissipation does not pose any problems for many consumer applications, some devices may require the motor to stop immediately. Using an H-bridge driver circuit, the motor may be stopped more quickly by switching to a reverse voltage for a short period of time before turning off the motor. The reverse voltage will oppose the existing kinetic energy of the eccentric mass and prevent it from producing further vibrations as it slows down.

By combining both an H-bridge circuit and a variable voltage source, both the start and stop times of an ERM motor may be further optimized. In order to turn on the motor faster, the starting voltage may be set slightly higher than the motor’s normal operating voltage. This higher voltage will increase the rate at which the eccentric mass reaches its optimal speed for producing the desired vibration (overdrive). As the eccentric mass accelerates, the voltage will lower to the normal operating voltage to sustain the vibration at a desired rate. Afterwards, a braking voltage may be supplied through the H-bridge to slow the motor down at a faster rate when turning off the vibration.

Though simple switching circuits and H-bridges may be constructed from common electronic components (MOSFETs and diodes), the procedure of providing an overdrive and braking voltage along with precise control over the intensity of vibration can be implemented using several off-the-shelf integrated circuits. As demand for consumer electronics with haptic feedback has increased, semiconductor companies have developed black-box integrated circuits with small packages that automatically handle the voltage regulation and intensity control for a vibration motor. The DRV2603 and DRV2605 from Texas Instruments, MAX1749 from Maxim Integrated, FAH4820 from Fairchild Semiconductor, MTCH810 from Microchip, and dozens of other integrated circuits are available to drive ERM motors—each with their own performance characteristics.

Linear Resonant Actuators

Although eccentric rotating mass motors gained mass adoption in consumer electronics, a new kind of device can be used to produce haptic feedback: the linear resonant actuator (LRA). Unlike an ERM motor, a linear resonant actuator uses a voice coil instead of a DC motor. A voice coil takes an AC input and produces a corresponding vibration with a frequency and amplitude corresponding to the electrical signal it is provided. Since the device must be controlled with alternating current, the necessary circuit to drive the actuator is significantly more complex than a circuit used to drive an ERM motor with direct current. In spite of the increased complexity, the devices have several unique advantages. LRA’s will typically consume less power to produce a vibration than an ERM motor, and their performance characteristics allow for significantly shorter start-stop times in typical applications. In addition, LRA’s don’t produce as much noise because they do not have a spinning mass inside of them.

LRA’s also use magnetic fields and electrical current to create a force, but the voice coil remains stationary inside the device. Instead of moving, the voice coil presses against a magnetic mass attached to a spring. By driving the magnetic mass up and down against the spring, the LRA as a whole will be displaced and produce a vibration. The mechanism works much like a speaker producing sound. In a speaker, air is funneled through a cone and displaced at different frequencies by turning an AC frequency and amplitude into a vibrational frequency and amplitude—internally, a speaker accomplishes this task by moving a magnetic mass with a fast-changing alternating current. Unlike a speaker, however, an LRA is useful in haptic applications within a specific frequency range.

The “resonant” part of the term “linear resonant actuator” comes from its internal use of a spring. Instead of directly transferring the force produced by the voice coil to the skin, the device optimizes for power consumption by taking advantage of the resonant frequency of the spring. As the voice coil pushes the magnetic mass against the spring at the spring’s resonant frequency, the device can produce a vibration of higher amplitude more efficiently. Since the voice coil is driven by an AC current modeling the desired frequency and amplitude of vibration, the frequency and amplitude may be independently modified, unlike an ERM motor that couples the two properties of the resulting vibration.

Although the frequency can be changed, the LRA will typically be operated within a narrow frequency range to optimize its power consumption—if the device is driven at the resonant frequency of the spring, it will consume less power to produce a vibration of equal magnitude. Regardless, this improvement alone presents a unique advantage over ERM motors: a precise waveform of varying intensity over time can be reproduced in an LRA with a fixed frequency, whereas a waveform of varying intensity in an ERM motor will also produce a varying frequency of vibration.

The typical start time for an LRA is approximately 5-10ms, a fraction of the time required to produce a vibration with an ERM motor. This incredible speed results from the immediate movement of the magnetic mass as current is applied to the voice coil inside of the device. In an ERM motor, the vibration can only be produced after the motor reaches its operating speed—even when overdriving the motor to produce faster acceleration, the motor can require 20-50ms before reaching a desired intensity of vibration. Unfortunately, the stop time of an LRA can be significantly longer than an ERM motor. An LRA can take up to 300ms to stop vibrating due to the continued storage of kinetic energy in the internal spring during operation. Thankfully, an active braking mechanism can also be used for an LRA—by performing an 180-degree phase shift of the AC signal provided to the actuator, the vibration can be stopped very quickly (within approximately 10ms) by producing a force opposite to the oscillation of the spring.

Although it is possible to create a custom driver circuit for an LRA, the process of detecting the proper resonant frequency and performing active braking can be very complex. Instead, several off-the-shelf driver chips exist to simplify this process and automatically handle the process of delivering crisp haptic waveforms. The DRV2603 and DRV2605 from Texas Instruments can also drive LRA’s in addition to ERM motors, but several other driver IC’s are also available. The MAX11811 from Maxim Integrated provides a haptic driver in combination with a capacitive touch sensing circuit. The SEMTEC SX86 provides a similar functionality but provides a resistive touch circuit. The LC898302 from ON Semiconductor and FAH4840 from Fairchild Semiconductor may also be used to independently drive an LRA.

C2 Tactor

The C-2 tactor is “a miniature vibrotactile transducer that has been optimized to create a strong, localized sensation on the body.” Much like a linear resonant actuator, the C-2 takes a driving AC voltage and creates a vibration corresponding to the frequency and amplitude of the driving signal. Unlike an LRA, the C-2 uses a moving “contactor” that oscillates in and out of the housing of the device. As a result, the vibrations from the C-2 are designed to be delivered on a single point of the skin rather than a vibration across the entire device. The enclosure of the C-2 allows the vibration to be isolated to a small surface area, enabling haptic interfaces constructed from C-2 tactors to utilize effects reliant on the localization of vibrations by user. Frequently, the C-2 tactor is applied using a silicone or elastomer enclosure that furtner localizes the produced vibration to a specific point and reduces the propagation of the vibration. The device can also be sewn into clothing or integrated into surfaces that frequently make contact with the skin.

Since the C-2 tactor has been designed to provide detailed feedback to a localized point of vibration, it has been frequently used in different device form factors for psychophysical studies of human perception. Although the tactor itself is too expensive for easy integration into a consumer technology product, it has enabled a more detailed exploration of the somatosensory system.

Piezo Electronics

The piezoelectric effect refers to an electric charge that accumulates in a solid material in response to an applied mechanical stress—the process occurs in a linear electromechanical interaction between the electrical state of crystalline materials and their mechanical state. This interaction is very useful for haptic feedback because it also occurs in reverse. An electric charge applied to a piezoelectric device will produce a mechanical change in the material. Recently, several synthetic ceramics have been developed to take advantage of the piezoelectric effect and produce a vibration corresponding to an alternating current applied to the material. Since the ceramics are often brittle, they are frequently enclosed in an insulating protective material before integration into a product. Piezoelectric actuators are more precise than both ERM motors and LRA’s because of their ability to vibrate at a wide range of frequencies and amplitudes that can be independently controlled using the driving AC voltage. Since the vibration does not rely on the resonant frequency of a spring, the frequency may be modified freely without a significant loss of efficiency.

Typical commercially available piezo actuators have a start time of approximately 14ms, which barely outperforms most commercially available LRA motors. More importantly, a piezo actuator can typically produce a much stronger vibration, allowing larger devices to provide richer haptic effects that propagate through a larger portion of the device body. Although Piezo actuators require a much higher voltage to operate, the current consumption (and overall power consumption) is on par with most LRA motors.

The two factors impeding a wider adoption of piezoelectric actuators are cost and power consumption. Unlike ERM or LRA motors, a piezoelectric actuator can produce a more localized vibration for a user interface when the vibrational force is applied to a floating surface. For example, the touchpad on an Apple Macbook laptop provides haptic feedback through vibrations produced by a piezoelectric actuator—since the actuator pushes primarily against the surface of the touchpad that is not firmly mounted to the rest of the device chassis, the force produced by the vibration will be directed primarily into the fingers of the user instead of propagating through the whole device. The technique of using a floating surface can also be applied to touch screens—though most modern cell phones and tablet computers feature a glass capacitive touch surface that is adhered to the device enclosure, new prototypes of floating touchscreens have been developed to demonstrate the possibility of delivering vibrotactile effects that do not propagate through the entire device.

Conclusion

It is best to use an ERM-type motor when:

  • the driving circuit needs to be very simple
  • cost is a primary concern
  • haptic resolution is not extremely important

It is best to use an LRA when:

  • start/stop timing is critical
  • the circuit can accommodate a driver chip
  • amplitude of vibration needs to be independently adjusted

It is best to use a Piezo when:

  • larger space is available to house the actuator
  • frequency and amplitude must be independently adjusted
  • the circuit can accommodate a driver chip and properly produce waveforms

At Somatic Labs, we’re using four linear resonant actuators in Moment, a device that lets you hack your sense of touch.

Enjoy reading this post? Let us know what you think on Twitter @SomaticLabs

Author: Jake Rockland

Jake is currently a junior pursuing a B.S.E. in Computer Engineering at the University of Arizona. A hacker at heart, Jake has experience with firmware development, full stack web development, and iOS development.

3 thoughts on “How do devices provide haptic feedback?”

  1. Jake,
    Nothing technically wrong with what you’ve posted. But I’d like to bring up the issue with IP. InvenSense owns over 1,000 patents on haptic feedback. So while you can drive the motors using the methods you mentioned in your paper, you can’t implement this commercially without potentially paying royalties to InvenSense. You’ll have to do a patent search.

    One way to avoid paying royalties (at least directly) is to use an application specific haptic driver IC like the ones you listed. The pricing of the IC already has the IP royalties accounted for (e.g. TI pays InvenSense some % per part sold)

    1. Hey Brian! Thanks for your response, this is a really good issue to consider. While this post was meant to be a general overview of the different ways haptic feedback can be implemented in hardware devices, it is important to note the IP issue you bring up when designing for commercial products.

Leave a Reply

Your email address will not be published. Required fields are marked *