Ever wonder how Hollywood filmmakers get clear aerial shots from a heaving, vibrating helicopter? The conventional method calls for a gutsy camera operator to hang by restraints from an open door of the chopper while shooting the scene with a mechanically stabilized handheld camera. Dampers in the base of the stabilizer cancel vibration while gimbaled arms with counterweights make the camera easy to manage. Still, getting that all-desirable level shot has hinged on operator skill. No more.
Skygyro, a stabilized camera platform from Tyler Camera Systems, Van Nuys, Calif., takes the guesswork out of locating the horizon thanks to a new solid-state gyro. As the pilot changes the helicopter’s attitude, the system keeps the camera level by actuating torque motors coupled to the platform’s pivoting axes. Mechanical springs and isolators at strategic locations damp out vibration from the helicopter drive train. Such a levelling system was impractical before now because, “ring-laser or fiber-optic gyros were too expensive,” says Nelson Tyler, designer of Skygyro.
The Skygyro is just one development to come out of advances in solid-state accelerometer technology. Builders of video game controllers make joysticks sensitive to input forces with surface-micromachined capacitive accelerometers. The sensor-equipped joysticks make the games they’re connected to more realistic. In another application, hundreds of surface-micromachined sensors in the walls of a building can detect damage from earthquakes. Low cost makes such systems feasible.
Silicon micromachined accelerometers are one member in a family of devices called microelectromechanical systems (MEMS). There are two basic types of these silicon accelerometers, piezoresistive and differential capacitive. Piezoresistive accelerometers are made by a process called bulk micromachining. With this method, mechanical features are chemically etched into a silicon wafer.
The etching process produces a tiny silicon mass suspended in a frame by multiple beam features also formed by etching away silicon. Each beam contains an implanted piezoresistor. The resistors are interconnected to form a Wheatstone bridge. At rest, the bridge is electrically balanced and its output voltage is zero. Force from acceleration bends the piezoresistors which causes their resistance to change. This unbalances the bridge, resulting in an output voltage. The greater the acceleration, the greater the voltage.
The piezoresistors also change resistance with temperature. This shows up as thermal drift in the output signal. Compensating circuitry is added to reduce the effect. Even so, piezoresistive accelerometers are more sensitive to temperature changes and have a narrower operating temperature range than differential-capacitive accelerometers.
Capacitive-micromachined accelerometers come in two types. One is made by the same sort of bulk micromachining that produces piezoresistive versions. The other type is fabricated through what is called surface micromachining. Functionally, the two capacitive sensors are the same. A silicon spring and mass assembly is flanked by two fixed plates. The assembly forms a differential capacitor. At rest, the capacitors are electrically balanced, resulting in no output signal. Forces from acceleration move the cantilevered mass toward one fixed plate and away from the other which causes a capacitive imbalance. The greater the acceleration, the greater the imbalance.
Bulk-capacitive accelerometers have a single mass and beam etched from a wafer of crystalline silicon. The silicon plates that make up the other half of the capacitor are bonded to either side of the structure holding the mass, forming a vertical, silicon sandwich.
In contrast, surface-micromachined accelerometers don’t have features etched into the silicon substrate itself. They instead contain an array of masses and beams etched from thin films of polysilicon and silicon oxides deposited on the substrate surface. This lateral approach makes it possible for one substrate to contain accelerometer mechanisms for multiple axes. Another advance of the approach is that accelerometer features are smaller than those of bulk-micromachined devices.
The relatively smaller features require about one-twentieth the substrate needed for a bulk micromachined unit. But smaller size does have its disadvantages.
For capacitive types, smaller features also mean less capacitance. The amplifier circuit that boosts the signal from these smaller capacitors must have a higher gain. For this reason, output signals from surface-capacitive sensors generally contain more noise. Surface types also have a larger threshold of sensitivity compared to bulk types. Their smaller sensing mass experiences a proportionately smaller body force for a given acceleration. Bulk types, with their heavier sensing mass, are better at resolving small acceleration changes.
Another difference between the sensors is the technique used to add support circuitry to the accelerometer mechanism. Surface devices use conventional methods borrowed from IC fabrication. Signal conditioning circuitry, such as an analog-to-digital converter, can be fabricated on the same chip as the sensing element.
Circuitry for bulk devices, on the other hand, must remain separate. This so-called glue logic sits on a separate chip and is electrically connected to the sensing element with tiny wires. Combining the two chips into one package results in a relatively larger and heavier sensor. The upside of this scheme is that the device can more easily be customized with circuitry specific to the application.
Capacitive accelerometers generally find use in low-frequency, low-g applications. Piezoresistive accelerometers are a better choice for sensing relatively larger g levels at higher frequencies. All can measure down to zero frequency, which makes them appropriate as tilt-angle sensors.
Most of the sensors are available in standard DIPs for mounting to printed-circuit boards. Economies of scale in manufacturing make surface types less expensive than bulk types, yet both cost less than servomass accelerometers.
When Prices Drop, Applications Abound
This is the approach taken in the DMU-VGX true vertical gyro, new from CTI. It marries three micromachined angular rate sensors with three surface micromachined capacitive accelerometers. The device is functionally equivalent to a mechanical vertical gyro in that it provides stabilized pitch and roll (artificial horizon) information. Though not meant to substitute for highly accurate ring-laser gyros like those used to help guide nuclear submarines, it is a candidate for less demanding applications such as camera stabilization.
Another firm that is applying the new sensors is TS Engineering Products, Eastlake, Ohio. William Thesling, a designer there, credits the accelerometers with keeping costs down in an Active Tuning Amplitude Controller (ATAC). An ATAC controls the electromagnets that drive vibratory parts feeders and sorters. Accelerometers attached to a feeder bowl provide feedback to a closed-loop control circuit. The system keeps the bowl vibrating close to its resonant frequency. “Before, control was open loop,” explains Thesling. “If a feeder were fully loaded, it would slow down. Also, power fluctuations cause the feed rate to change which can create bottlenecks in automated assembly operations. Feedback control lets parts flow smoothly and more predictably, and you use less power.”
Still, manufacturing personnel will likely have to manually lift boxes of parts into the feeder bowl. BackTalk from Bio Kinetics Corp., Mission, Tex., reminds its wearers to lift with their legs. Worn on the belt, the pager-sized device uses surface-micromachined capacitive accelerometers to sense angular change and acceleration of the torso and spine. BackTalk either beeps or vibrates when a wearer attempts to lift improperly, exceeding limits programmed into the device’s memory. Data from lifting events is recorded for future use and can be downloaded to a PC.
In another human-machine application, the MM-1 from Axon Instruments, Foster City, Calif., quantifies the severity of tremors in patients inflicted with Parkinson’s disease. A triaxial accelerometer from Crossbow Technologies senses movement of a patient’s wrist in three dimensions. A microprocessor collects 15 sec of analog data from the wrist-worn sensor, digitizes the information, then performs a spectral decomposition. About 15 sec later, the MM-1 displays the predominant frequency, the amplitude associated with that frequency, and the total amplitude of acceleration. This gives neurosurgeons a quick and objective measure of patient symptoms before, during, and after functional surgery. It also helps neurologists prescribe appropriate levels of medication.
When cost drives design, as is frequently the case with consumer products, surface-micromachined accelerometers usually win out.