1998 - Draper Laboratory

surface micromachining layers of polysilicon. Manymanufacturers are developing gyros and accelerometers using thistechnology. Their extremely small size combined with thestrength of silicon makes them ideal for very high-accelerationapplications. Between 3,000 and 10,000 devices can be producedon a single 5-in silicon wafer.Draper Laboratory has demonstrated a 4-deg/h bias drift,open-loop silicon tuning-fork gyroscope with folded beamsuspension in which the flexured masses are electrostaticallydriven into resonance with a comb-like structure (see Figure 2).Rotation is sensed capacitively along the axis normal to the planeof vibration. Draper’s first gyroscope is aimed at the automobilemarket and is being marketed through an alliance with BoeingNorth American. Devices with lower drift rates have beendeveloped for more demanding applications, such as autopilotcontrol and smart munitions. Future performance improvementsare expected to bring the performance of these devices to betterthan 0.1-deg/h bias drift.(proof mass, resonating flexure, and support structure) from asingle piece of quartz (see Figure 3). Using such techniques canresult in low-cost, highly reliable accelerometers with ameasurement accuracy of better than 100-µg bias error.Constructing this accelerometer from a single piece of quartzresults in high thermal stability, along with dynamic rangesapproaching those obtainable in the timekeeping industry.Silicon micromechanical resonator accelerometers are also beingdeveloped.Figure 3. Quartz resonant accelerometer.Silicon Micromechanical AccelerometersFigure 2. Micromechanical tuning-fork gyro.Resonating Beam AccelerometersResonant accelerometers (sometimes referred to as vibrating beamaccelerometers) have a principle of operation that is similar to thatof a violin. When the violin string is tightened, its frequency ofoperation goes up. Similarly, when the accelerometer proof massis loaded, one tine is put into tension and the other intocompression. These tines are excited continually at frequencies inthe hundreds of kilohertz range when unloaded. As a result,when “g” loaded, one tine frequency increases while the other tinefrequency decreases. This difference in frequency is a measure ofthe device’s acceleration. This form of accelerometer isessentially an open-loop device, in that the proof mass is notrebalanced to its center position during the application of a force.For accuracy, it relies on the scale-factor stability inherent in thematerial properties of the proof mass supports. Theseaccelerometers can be constructed using several differentfabrication techniques. One method is to etch the entire deviceMicromechanical accelerometers are either the force rebalancetype that use closed-loop capacitive sensing and electrostaticforcing, or the resonator type as described above. Draper’s forcerebalance micromechanical accelerometer is a typical example, inwhich the accelerometer is a monolithic silicon structure (i.e., noassembly of component parts) consisting of a torsional pendulumwith capacitive readout and electrostatic torquer (see Figure 4).This device is about 300 x 600 µm in size. The pendulum issupported by a pair of flexure pivots, and the readout andtorquing electrodes are built into the device beneath the tilt plate.The output of the angle sensor is integrated and then used to drivethe torquer to maintain the tilt plate in a fixed nulled position.The torque required to maintain this balance is proportional tothe input acceleration. Performance around 250-µg bias errorand 250 ppm of scale factor error have been achieved and furtherimprovements are expected.Future Technology ApplicationsSolid-state inertial sensors like those described previously havepotentially significant cost, size, and weight advantages overconventional instruments, which are resulting in a rethinking ofthe options for which such devices can be used in systems. Whilethere are many conventional military applications, there are alsomany newer applications that will emerge with the low cost andvery small size inherent in such sensors, particularly at the lowerperformance end of the spectrum. In nearly every case, whenthese newer solid-state inertial technologies have been evaluatedINS/GPS Technology Trends for Military Systems4