Smarter than your average pushbutton

June 7, 2006
Designers are turning to MEMS sensors as a way of devising superresponsive operator controls.

Michelle A. Clifford
Applications Engineer
Sensor Product Div.
Freescale Semiconductor
Tempe, Ariz.

The g-cell accelerometer is an example of X-lateral technology that uses a moveable mass with doubly fixed beams. Each pair of g-cell beams form two back-to-back capacitors. As the central mass deflects due to acceleration, the distances between the beams change shifting their capacitance values.

The diagram shows the g-cell sensor connected to an ASIC that performs the capacitive-to-voltage conversion. The ASIC also provides for variable gain control, filtering, and temperature compensation of the outputs.

Motion recognition requires combining the accelerometer sensor with a microcontroller. Simple motions only require an 8-bit controller. The 8-bit controller is replaced with more powerful processors or DSPs as movements become more complex or control functions rise.

The angle on cell phones

Accelerometers embedded in cell phones can make the operation of the phone almost pushbutton free. By monitoring relative positions and movements, the system automatically knows when the user requires phone activation. Coupled with voice-recognition technology for dialing and other controls, the motion-sensing phone would not require pushing any buttons to operate.

When you think of user interfaces or controls, levers, buttons, switches, joysticks, and other manual control mechanisms may come to mind. These controls usually need a specific motion to activate: a lever pushed in or pulled out, a button pressed, or a switch flicked up or down.

However, a new class of inter-faces spawned by miniature accelerometers now give users new ways of making controls more intuitive.

Most accelerometers, like those from Freescale Semiconductors, are capacitive sensors designed using microelectromechanical systems (MEMS) technology. MEMS is an integration of mechanical and electrical systems where microscopic electromechanical structures are created using micromachining.

Bulk micromachining selectively etches parts of a silicon wafer while surface micromachining builds thin-film structures on a silicon wafer. The Freescale accelerometers incorporate two types of structures in their production: a Z-axis form and an X-lateral form. The Z-axis technology is best described as a capacitive trampoline where the tram-poline moves along the axis of acceleration. The X-lateral technology consists of a movable mass and doubly fixed beams. Each pair of g-cell beams form two back-to-back capacitors. The central mass moves due to acceleration. The motion changes the distance between the beams and therefore the capacitance measured between them.

The g-cell output is connected to an application-specific integrated circuit (ASIC) that provides signal conditioning. The ASIC is where the change in the MEMS structure capacitance is converted to a voltage output that is proportional to the sensed acceleration. The ASIC uses switched capacitive technology to create its voltage output from the change in g-cell capacitance. In addition, the ASIC provides filtering, temperature compensation, and g-selection.

The ASIC may be built in with the g-cell, or each device can take the form of its own IC. For example, the MMA7200Q and MMA6200Q from Freescale are two-chip accelerometers with the g-cell in one chip and the ASIC in the other.

Of course, accelerometers measure acceleration and deceleration. But there is an extensive amount of additional information that can be obtained from these measurements. In fact, there are six sensing functions available from accelerometers besides change in velocity: tilt, motion, position, shock, vibration, and free fall.

A motion-recognition system using an accelerometer would gage a combination of all these measurements to detect various motions. Instead of having to push multiple buttons, the operator can use tilt, tap, shake, and other motions to provide access to different device functions.

In the past year, consumer electronic devices have incorporated basic motion detection to initiate functions once accessed through pressing multiple buttons. For example, the small shock from a tap-ping finger can trigger a menu of selections. Scrolling the menu is as easy as tilting the device forward, backward, or side-to-side. The tilt motion signals the scroll bars to move up, down, left, or right. These types of functions are actually simple to implement. More advanced user interfaces are possible by adding three-dimension motion detection.

Motion-recognition systems are found with varying levels of complexity. The simplest implementation includes a one, two, or three-axis accelerometer and a low-cost microcontroller. The microcontroller waits for an acceleration value or sequence of accelerations that match a logic pattern sequence in memory. When it detects a preprogrammed motion or acceleration, the microcontroller executes the corresponding command or control sequence.

The simplest systems need only an 8-bit microcontroller to handle motion-detection algorithms. Processors need more sophisticated algorithms to recognize more-complex motions. These demand more ROM and faster processing capability. A digital signal processor (DSP) might be needed for sufficiently advanced tasks, especially those involving motors. For example, the DSP56F805 is designed with two special pulse-width-modulation (PWM) modules for motor control. Each PWM module can deliver six independent PWM signals depending on its configuration.

It is also possible to have the microcontroller memorize a sequence of accelerations. It can then be programmed to look for acceleration sequences that match what it saw before. When the user performs a similar movement, the controller generates the associated output signal.

An even-more intriguing idea is to use accelerometers for security. The system could detect how a machine is handled and could alert authorities if it detects possible tampering as with a high-impact hammer. Similarly, accelerometers could monitor the position of a control and detect movement in an improper direction.

Another good application is with individuals with limited mobility who lack the strength or range of motion to use common mechanical controls. For example, some people lack the strength or range of motion to properly operate a simple joystick on a wheelchair. An accelerometer-based motion detection system can let software learn the motions such individuals are capable of generating.

One approach is to detect X, Y, and Z-axis displacements within a certain range of values for a selected period of time. This specific value range reduces the chance of misinterpreting random movements as intentional movement. For example, movement would be considered intentional when 10 consecutive measurements are within a particular value range for each axis. This same approach can be used with normal operator controls as well as with those for challenged individuals.

Detection of abnormal occur-rences to trigger protective actions is well within a motion-detection system's forte. One typical application is found in hand-held MP3 players that use hard drives to store digital music files. Accelerometers in the player sense the difference between a fall (acceleration due to gravity) and the random accelerations from carrying and handling. Acceleration from a fall signals the hard drive in the player to park its magnetic heads before the player strikes the floor. The entire sensing of the fall, alerting the drive to park the drive heads, and the actual safe parking of the heads, takes place within the time frame of a 2-ft drop.

Accelerometers used for tilt require sensitivities from 800 to 1,200 mV/g. The range of acceleration is from 1 to 1 g to detect the motion of being held flat (1 g), on its side (0 g), and finally flat again but upside down (1 g). However, with motion detection, the range of sensitivity required fluctuates greatly. Small hand motions would need at least 2 g. Larger motions, such as when a person is moving their hands while talking, can require 3 g or more. An abrupt motion like a person shaking their fist produces 4 g or more of acceleration. The hardware must accommodate all of these motions and distinguish between them in a full-motion-detection system.

To provide a wide dynamic range, some full-motion-detection systems can dynamically scale their sensitivity to motion. Examples of this are the Freescale low-g accelerometers with g-select option, such as the MMA7260Q. G-select provides two logic inputs to the accelerometer driven by a microcontroller that changes the sensitivity and g-range dynamically. The MMA7260Q provides the option to select 1.5, 2, 4, and 6-g acceleration levels.

An accelerometer embedded in a cell phone can detect phone position as the user handles it. During most if its life, the phone would hang vertically in a pocket or lay flat in a purse. When the owner picks up the phone, accelerometers can detect the motion and orientation of the phone to activate it automatically. There's no need to push any buttons; there's enough information in the phone movement to know it is being used. Coupled with voice-recognition technology, cell phone operation now becomes totally pushbutton free.

All in all, accelerometers can add flexibility to levers, buttons, switches, joysticks, and other manual control mechanisms. Combining accelerometers with mechanical or electronic controls can lead to smarter user inter-faces and controls.

Freescale Semiconductor, (800) 521-6274,

Voice your opinion!

To join the conversation, and become an exclusive member of Machine Design, create an account today!