IEEE SENSORS JOURNAL, VOL. 3, NO. 4, AUGUST 2003

383

Interface for MEMS-Based Rotational Accelerometer for HDD Applications With 2.5 rad/s2 Resolution and Digital Output Alberto Gola, Enrico Chiesa, Ernesto Lasalandra, Fabio Pasolini, Michele Tronconi, Tommaso Ungaretti, and Andrea Baschirotto, Senior Member, IEEE

m

Abstract—A 0.6BiCMOS interface for microelectromechanical systems (MEMS) based rotational accelerometers is presented. It is housed in an inexpensive standard SO-24 plastic package with a capacitive rotational accelerometer sensor produced using MEMS technology. This sensitive interface chip includes the analog-to-digital conversion, filtering, and interface functions. The analog-to-digital conversion is realized through a single-bit electromechanical – loop able to detect capacitive unbalancing as low as 50 aF (50 10 18 F). The produced bitstream is then processed by a digital chain and made available through a standard 3.3-V (5-V tolerant) three-wire serial bus. The signal bandwidth is about 800 Hz, the sensitivity is 2.5 rad/s2 , with a full-scale sinewave of 200 rad/s2 and a signal-to-noise ratio peak of 38 dB over 30–800 Hz. Through the serial bus, it is also possible to program device characteristics including gain, offset, filter performance, and phase delay. The complete sensor is used in a feed-forward compensation scheme to cancel external disturbances acting on computer hard-disk drives so as to steadily keep the read-write heads on track: this allows greater track densities and better speed performance.

61

I. INTRODUCTION

Y

EAR by year, the hard-disk drive (HDD) storage capacity is required to increase and the cost/gigabyte to decrease. An important factor to increase the HDD storage capacity is the improvement of the position control of the read-write heads. As it is usually desired to keep the tracking error below one eighth of the track width, and track widths of 1 m are needed to achieve density of 10 Gb/in , leading thus to tracking errors that must be kept below 0.12 m. To obtain low tracking errors in conventional HDDs that are exposed to vibrations and external shocks, it is mandatory to compensate as much as possible all the disturbances acting on the HDD head. To address this problem, a rotational micromachined accelerometer placed on the base casting of the HDD has been used as a disturbance observer and the acceleration signal detected by that device is fed to the HDD through a feedforward control loop to minimize the variance of the tracking error produced by external vibrations Manuscript received June 27, 2002; revised January 27, 2003. The associate editor coordinating the review of this paper and approving it for publication was Dr. Thaddeus A. Roppel. A. Gola, E. Chiesa, E. Lasalandra, F. Pasolini, M. Tronconi, and T. Ungaretti are with the STMicroelectronics, Milan, Italy (e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected]). A. Baschirotto is with the Department of Innovation Engineering, University of Lecce, Lecce, Italy (e-mail: [email protected]). Digital Object Identifier 10.1109/JSEN.2003.815785

Fig. 1. Closed loop control system.

[12]. The conceptual scheme of this concept is given in Fig. 1. The device operates as follows. The minute rotational vibrations are detected and measured in order to generate a feed-forward correction signal for the voice coil drive circuit to keep the head in the correct position. This allows the track-to-track distance to be reduced significantly, increasing the storage density per unit area. In addition, the speed performance of the drive is also improved because less time is wasted restoring the correct head position when vibration has caused the tracking to be temporarily lost. In the following, the capacitive rotational acceleration sensor interface embedded in such a feedforward HDD control system is presented. It uses a microelectromechanical system (MEMS) as rotational sensor. The vibrations of the disk drive cause tiny rotational displacements of the moving part of the MEMS structure. These movements are measured by sensing the small variations in capacitance that they cause. The proposed device is capable to measure capacitance changes as small as 0.05 fF. In addition, it connects directly to a microprocessor through a serial digital interface. Physical position in the drive is noncritical because the sensor measures only the pure rotational component of vibration regardless of the origin. Moreover, the complete system (rotational acceleration sensor and read-out electronics) is housed in a single, conventionally styled package compatible with standard production equipment. The proposed device is much better than the possible alternative solutions based on two linear accelerometers. In these cases, the rotational acceleration is measured from a post-processing of the data coming from the two adopted linear accelerometers. The two-linear accelerometer-based solutions present the following draw-backs. • The measurement suffers from accelerometer misplacement and devices mismatch.

1530-437X/03$17.00 © 2003 IEEE

384

IEEE SENSORS JOURNAL, VOL. 3, NO. 4, AUGUST 2003

• Additional post processing is necessary. • The two linear accelerometers requires two separate packages, more board room, and more external components. Therefore, for all the above reasons, the two linear accelerometers solutions are much less reliable that the here proposed rotational solution. The proposed rotational accelerometer reaches a 38-dB dynamic range, which is mainly limited by the Brownian noise of the sensor. This corresponds to a resolution of about 2.5 rad/s in a 200 rad/s full scale in a 800-Hz bandwidth (which is the maximum frequency of interest for the HDD disturbances). The circuit has been realized in a 0.6- m BiCMOS technology with an area of 9.5 mm . The power consumption is 150 mW from a single 5-V supply. Notice that the presented performance have been achieved nonetheless the system presents a two-chip structure (i.e., sensor and read-out electronics are on different devices). In fact, the large parasitic capacitance of this configuration dramatically reduces the achievable dynamic range. The paper is organized as follows. After this introduction, in Section II, the functionality of the complete HDD system in which the proposed rotational accelerometer is embedded is described. In Section III, the architecture of the rotational accelerometer is described. Section IV deals with the details regarding the analog circuitry, while the digital circuit is described in Section V. Section VI gives the experimental results for the proposed device when it is in stand-alone configuration and when it is embedded in the complete HDD channel. II. HDD SYSTEM FUNCTIONALITY The departure of the read/write head from a perfectly circular track is called runout. The track runout can be decomposed into two categories. • Repetitive runout (RRO): Identical magnitude of runout is obtained at each servo sector of a track. It is mainly due to the fact that the disk platte and the spindle rotation axes are not perfectly concentric. Another source of RRO is the tracking error that occurs during the operation of servo writing which translates into permanent position errors on the disk. • Nonrepetitive runout (NRRO): the amplitude of this runout component is not identical for each revolution of the disk. It is due to vibrations of the mechanical structure of the hard-disk, i.e., suspension resonance (in the band 220–2700 Hz), disk platter resonance (in the band 600–1100 Hz), spindle resonance, and external disturbances. The control system for the head positioning is shown in Fig. 2, where the additional rotational accelerometer proposed in this paper is indicated. The system operates as it is briefly described in the following. is subtracted The digital word of the reference position from the feedback information and it converted into analog in order to drive the voice coil motor (VCM). The relative action is however corrupted by the vibration applied to the HDD and so transfer function the positioning (which is filtered with a corresponding to the mechanical frequency response) results in value. This value is acquired and digitized and the actual

Fig. 2. Feedforward HDD control system.

then supplied to the digital controller to calculate the nominal control output. The additional rotational accelerometer acquires the amount of rotational vibration applied to the HDD and to properly correct the nominal control output signal. This is achieved by summing the rotational accelerometer output to the digital controller output. III. ACCELEROMETER SYSTEM ARCHITECTURE A general process defined thick epitaxial layer for micromotors and accelerometers (THELMA) for the realization of all the inertial sensors (angular, linear accelerometers, and gyroscopes) has developed [13]. It is a kind of epithaxial micromachining process. On a standard silicon substrate, a thermal oxide is grown. On the top of the oxide, a thin polysilicon layer is LPCVD deposed in order to realize buried interconnections. Afterward, a sacrificial oxide layer is deposed on the top of which a thick polysilicon epithaxy is grown (15 m). This epithaxial polysilicon is the structural material that composes the sensor. The structure of the sensor is defined by deep silicon RIE. At the end, the sacrificial oxide is removed by using a vapor phase HF etching. With the process described above, the sensor results to be thicker than sensors obtained with thin film deposition techniques and, therefore, shows an higher sensitivity; moreover, with respect to the structures obtained with the surface micromachining technique, it has greater stiffness to bending along axis and therefore a lower probability to stick on the substrate during sacrificial oxide removal. This process has also a lower cost than processes based on the use of SOI or silicon fusion bonded substrates. In the proposed rotational accelerometer system, the sensor, realized using the above described process, is shown in Fig. 3. Fig. 3(a) shows a microphotograph of the entire rotational sensor, while a particular of it is shown in Fig. 3(b). When a rotational acceleration is applied, the proof mass displaces from its nominal position, causing an imbalance in the capacitive half-bridge. Fig. 4 shows the sensor equivalent electrical circuit: it conand between the single rotor sists of two capacitors and each of the two stators ( and ). Some additional para, , and ) are present between sitic capacitances ( each stator and ground which are also due to the connection between the sensor die and the electronics die. Series resistors and ) and rotor conare present along the stator ( nections. When no force is applied to the rotor, the sensor has and ) of 2 pF between a nominal capacitance value (

GOLA et al.: INTERFACE FOR MEMS-BASED ROTATIONAL ACCELEROMETER

(a)

(b) Fig. 3. (a) Microphotograph of the entire rotational Microphotograph of a detail of the rotational sensor.

sensor. (b)

385

position. Because the output is dependent only on the feedback force, the device is insensitive to first-order to variations in the mechanical spring constant. There are many ways of implementing a force-feedback loop. Among them, – modulation is particularly attractive because it is simple, it provides a digital output, it has a large bandwidth, it is robust to any spread in the mechanical parameters of the MEMS element, and it can be easily implemented in high-density CMOS technologies. The adopted topology within the complete accelerometer is shown in Fig. 5, where the proof mass acts as a second-order filter, shaping quantization noise at frequencies above the mechanical resonant frequency [2], [4]. A compensating network has been introduced to prevent instability. The use of a feedback loop allows to move the natural poles of the proof mass to higher frequency, thus decreasing the phase lag that they introduce on the band of interest that, for this particular device, ranges between 30 and 800 Hz. This relaxes the constraints on the sensor mechanical properties. In terms of noise, the performances of the system are limited by the sensor Brownian noise, that is by a mechanical white noise due to molecular movements of the particles composing the sensor itself. As shown in Fig. 5, the measurement interface can be decomposed into two parts. The first one is the analog front-end which is connected to the rotational acceleration sensor. It produces an oversampled pulse density modulated signal carrying the information on the rotational accelerations which the device is subjected to. The second stage is a digital circuit that downsamples and digitally process the aforesaid bitstream and makes it available to the external through a standard three-wire digital interface. Both of these will be described in the following two sections. IV. ANALOG SECTION DESIGN

Fig. 4.

Sensor equivalent electrical circuit.

each stator and the rotor. When a force is applied, the maximum capacitance variation is 200 aF (corresponding to a displacement of 0.2 nm). This imbalance is measured using charge integration [1]–[3] in response to a single voltage pulse applied to both the sensor capacitors. In this way, the device can be implemented by using fully-differential switched-capacitor techniques, which have demonstrated to be able to perform high dynamic range analog circuits. The sensor and the read-out electronics are realized in different chips and they are connected with bond wire, as shown in Fig. 9. Nonlinearities in the capacitive pickoff and mechanical springs are minimized by using a closed loop configuration, which maintains small deflections. Force balancing of the proof mass is attained by enclosing the proof mass in a negative feedback loop. The feedback loop measures deviations of the proof mass from its nominal position and applies a force to keep the proof mass centered. The accelerometer output is taken as the force needed to null, or zero, the mass

The measurement chain together with the sensor forms an electromechanical – control loop, which senses the input rotational accelerations by measuring the displacements of the moving mass. In order to keep low the mass displacement, the feedback loop controls the mass movements by applying a force in the opposite direction. In this way, the mass typically operates in a range of positions that is very near its static equilibrium position and the accelerometer output signal is the force needed to keep the mass as close as possible to the center position. The moving mass (sensor) acts as the summing node and the integrator of the loop while the proposed device performs the measurement of the displacement by measuring the capacitance imbalance, compares this signal with its mean value (i.e., zero) and decides where to apply the feedback signal. The negative feedback loop is operated as a sampled data system, performing either a measure and cancellation of position sense circuitry static errors (i.e., offsets) and slowly varying dynamic errors noise), or the feedback operation, applying a known (i.e., voltage to right portion of the sensor. In addition, by controlling the mass movements, it performs an embedded analog-to-digital (A/D) conversion of the output measurement. The differential capacitance of the sensor is proportional to the moving mass (rotor) displacement; thus, by sensing the dif-

386

IEEE SENSORS JOURNAL, VOL. 3, NO. 4, AUGUST 2003

Fig. 5. Overall system architecture.

Fig. 6. Analog signal processing chain.

ferential capacitance, the position of the sensor is determined. Then, since the rotor position is known, and the position is related with the input acceleration, the input acceleration can be easily inferred. When a nonzero acceleration is applied, the output bitstream mean value follows the input signal, i.e., the applied acceleration. The noise of the whole measurement chain is set below the 39-dB limit given by the sensor Brownian noise level. The position sensing consists of the following main blocks: • the fully differential switched-capacitors charge amplifiers chain with correlated double sampling (CDS) network; • the comparator and selection logic; • the compensating network and the feedback DAC. The switched capacitors charge amplifiers chain consists of a fully differential charge amplifier followed by two SC amplifiers, as shown in Fig. 6. The structure gives an output sample every 1.56 s (i.e., with a 4.48 MHz clock rate). The system operations are divided into three phases: • Reset phase (RST) in which all the capacitors are precharged to the input and to the output common-mode voltage;

Fig. 7.

Reference voltages generation.

• Correlated double-sampling phase: [not (RST) not (SENS)] in which the first and the second opamps are

GOLA et al.: INTERFACE FOR MEMS-BASED ROTATIONAL ACCELEROMETER

387

Fig. 8. Digital section structure.

operating with no input signal. The output is only due to the offset and to the noise of all the opamps. These are sampled in the CDS capacitor of the 3rd gain stage and all the low frequency components (offset and 1/ noise) are cancelled in the following phase; • Sensing phase (SENS) in which a step is applied to the rotor and this results in a voltage step at the third opamp output. The amplitude of such an output step depends on – and, thus, the input sensor capacitor mismatch on the mass misplacement. For all the opamps, about 400 ns are available to settle. However, they have been designed with different noise and output swing requirements. The first opamp (which is operating in conjunction with an auxiliary opamp for the input common mode feedback [1] due to the unipolar step applied to the differential sensor) has to feature low noise. For this reason, it is the most power demanding (12.5 mW). On the other hand, its output signal is expected to be very small. Thus, a telescopic configuration (with small output swing) has been adopted in order to maximize the ratio gm/totalcurrent and, then, to minimize the opamp noise for a given current level. The other two opamps are embedded in capacitive gain stages implementing a gain of 9 and 3, respectively. Thus, they have to process larger signal amplitude, with less stringent noise performance. A folded-cascode configuration, which features a slightly higher output swing, has been used. In addition, the reduced noise requirement allows to strongly reduce their power consumption (6 mW each). The last amplifier and its switched- capacitor network implements the active CDS technique. The purpose of the CDS network is to reduce the sensitivity of the position sensing circuitry to some nonideal effects as the clock-feedthrough of the integrator switches, the offset and the 1/ noise of the integrating opamps. For all of them, input and output common mode voltand , respectively) are set to . In adages ( dition the output common-mode voltage is controlled by a sampled-data (switched-capacitor) common-mode feedback circuit. Notice that the eventual offset due to mismatched sensor capacitors could not be corrected by the CDS operation. Thus, it is cancelled, in this device, by the factory trimming of capacitors and . Finally, the overall chain gain is undefined due with the first to the uncorrelation of the sensor capacitor . This measurement inaccuracy stage feedback capacitor is reduced in the digital part after a training sequence. The comparator operation is divided in two phases: preamplification and latching. During the latch phase, the latch drives a small logic network that produces the correct feedback signals. These signals close two switches, one of which shorts the while the other shorts the second stator to proper stator to

Fig. 9. Chip photograph.

the rotor. In this way, a feedback force pulse is injected in the loop, keeping the mean rotor displacement close to zero. Since the sensor behaves like a second-order integrator, a compensation block is needed to stabilize the loop. This is achieved by placing a zero in the signal transfer function. In particular, this is realized applying two different feedback forces that are selected accordingly to the comparator output and to its previous value: if the comparator value changes, then the feedback voltage will be set at 3.8 V; otherwise, it will be set at 2.225 V. All the required reference voltages are generated from an internal bandgap reference circuit so to have minimum sensitivity to the operating temperature and supply voltage (as shown in Fig. 7). The values of the adopted voltage reference values are 1.5, 2.5, 2.225, 3, and 3.8 V. Among them, the 2.225 V has to be as clean and stable as possible. For these reasons, it is filtered by an external capacitor. Furthermore, for the generation of the 3.8-V voltage reference, low compliance current mirrors have been used in order to properly operate also with a 4.5-V supply (i.e., for the nominal 5-V supply in presence of a 10% reduction). V. DIGITAL SECTION DESIGN The bitstream produced by the – modulator is processed by a digital chain in order to remove as much as possible the quantization noise and to lower the sampling rate besides correcting the data in terms of offset and gain. The structure of the digital section is shown in Fig. 8. The decimation process is carried out by a programmable second or third-order sinc filter that decimates the signal either by 16 or by 32, accordingly, to the user settings. The decimation

388

IEEE SENSORS JOURNAL, VOL. 3, NO. 4, AUGUST 2003

(a)

(b) Fig. 10. Output bitstream spectrum. (a) Before digital filter; (b) after digital filter.

stage implements the well know filtering transfer function given by (1) and are the decimation factor and the filter order, where respectively. Particular care has been taken when designing the sinc itself in order to reduce as much as possible the computation latency (the computation chain, in fact, has been implemented by using fast parallel adder and subtractors not to require the pipeline to be broken up in multiple steps. The sinc result is then latched at the end of the filter stage to keep the produced data stable at its output).

The sinc stage is followed by a second-order IIR filter used to remove the residual modulation noise that is not adequately attenuated by the decimation stage. The filter has been implemented by a single multiply and accumulate (MAC) unit and it computes the following transfer function: (2) The IIR transfer function is calculated through the use of a MAC unit that decomposes the formula given in (2), as written in (3)

GOLA et al.: INTERFACE FOR MEMS-BASED ROTATIONAL ACCELEROMETER

389

(a)

(b) Fig. 11.

Output sinewave. (a) Before digital filter; (b) after digital filter.

In particular, the order with which the addendums are considered andevaluated has beenchosen so to minimize as much as possible the delay with which the filter output is made available. This delay, in fact, impacts on the overall phase lag introduced by the measurementandprocessingchainand,especiallywhenusingthe accelerometer in feed forward compensationsystems,itshould be reduced as much as possible, since it is a critical parameter for the effectiveness of open loop compensation techniques. All the coefficients of the filter are quantized with 9 bits and are stored inside dedicated registers which are loaded at power on resetfroman embedded flash memory where the coefficients themselves are permanently stored. The user is allowed to change them on the fly to performadaptivefilteringandalsotochangethefilterresponseaccordingly to its application requirements.

When computing the IIR filter transfer function, some more computation steps are carried out in order to shape the truncation error due to finite word length arithmetic. In particular, a second-order shaping of the truncation error is performed and computed by using the MAC unit that performs the IIR filter computation. This unit is exploited as well to perform offset and gain adjustment of the measured and processed acceleration data in order to correct any dc offset and gain loss that are produced by a nonperfect calibration of the mechanical sensor and by the electronic interface. The values employed for offset and gain adjustment, like any other calibration parameter, are stored inside the embedded FLASH memory and are downloaded at power on reset inside their own registers. The data produced by the measurement chain are stored inside a FIFO able to host up

390

IEEE SENSORS JOURNAL, VOL. 3, NO. 4, AUGUST 2003

Fig. 12.

Position error signal as a function of the applied vibration amplitude.

TABLE I PERFORMANCE SUMMARY

to four 8-bit wide acceleration samples and they are accessible through a standard three-wires serial interface. All internal timings and synchronization signals are generated through the use of an embedded all digital PLL. In particular, the device is able to lock both on an HW signal applied on one external pin of the device and on a SW signal that is produced by a reading of the FIFO of the device. The presence of a PLL is particularly useful for synchronizing the data processing with an external device, or a DSP. In this way, any data loss and thus any typically a possible aliasing effect due to a noncorrect data handling are avoided. VI. EXPERIMENTAL RESULTS The proposed electronic read-out device has been realized in a 0.6 m BiCMOS technology. The device is housed in an inexpensive standard SO-24 plastic package with the sensor counterpart. The full package photograph is shown in Fig. 9, where sensor and read-out electronics are indicated. The die size of the full electronics read-out (including read-out and feedback forces) is about 9.5 mm . The device operates from a single 5-V

supply and draws 30 mA (i.e., it consumes 150 mW). In this device the power consumption has not been minimized because the large power consumption of the other blocks in the HDD system make the rotational accelerometer power consumption negligible. The presented device has been characterized by attaching the packaged die to a shaker table excited by a sinusoidal input signal. Data from the test chip were gathered by a computer and analyzed off-line. A precision accelerometer was mounted adjacent to the test chip for calibration. Fig. 10 (a) and (b) show the spectrum of the bitstream before the final digital filter and at the output of the complete system (i.e., after digital filtering), respectively. It corresponds to an acceleration of 150 rad/s at a frequency of about 100 Hz. A SNR of 38 dB is achieved in an 800-Hz bandwidth. This corresponds, for a 200 rad/s full scale to about 2.5 rad/s LSB, which fully satisfies the application requirements. The DR decreases to 27 dB for a 3-kHz bandwidth, and to 20 dB for a 10 kHz bandwidth. Fig. 11 shows the output waveform for the same testing conditions. Also in this case, Fig. 11(a) reports the waveform before digital filtering, while Fig. 11(b) plots the waveform after digital filtering, where it is evident the improvement of the quality of the signal. Table I summarizes the device features. The performance of the proposed device have also been characterized when it is embedded in the full HDD system shown in Fig. 2. In this case, the standard deviation of the position error signal (PES) can be plotted as a function of the amplitude of the vibration applied to the system, as shown in Fig. 12. In this plot, the PES is reported for the case in which the rotational accelerometer is adopted or not. It can be seen that the adoption of the rotational accelerometer gives a significant improvement with respect to the case in which it is not used. This validates both the proposed device and the complete HDD feedforward control system.

GOLA et al.: INTERFACE FOR MEMS-BASED ROTATIONAL ACCELEROMETER

VII. CONCLUSION In this paper a complete fully-integrated rotational accelerometer has been presented. This device has to be applied to a feedforward method for increasing the density in magnetic recording. The output signal of the proposed rotational accelerometer is used to correct the misalignment of the hard-disc drive head during operation. The front-end electronics is BiCMOS technology. It processes the realized in a 0.6signal of the sensor and supplies the feedforward circuitry. Finally, the proposed device has been tested on a commercial HDD, with an adaptive feed forward compensation. The results obtained confirm that the device is well suited to improve the performance of head servo positioning and, in turn, to increase the achievable track density.

ACKNOWLEDGMENT The authors would like to thank D.S. Cini, for his inimitable charisma and unparalleled sharpness of mind which made this work possible, and C. Caminada, for the careful design of the device layout.

REFERENCES [1] C. Lu, M. Lemkin, and B. E. Boser, “A monolithic surface micromachined accelerometer with digital output,” in ISSCC Dig. Tech. Papers, Feb. 1995, pp. 160–161. [2] T. Smith, O. Nys, M. Chevroulet, Y. DeCoulon, and M. Degrauwe, “A 15 b electromechanical sigma-delta converter for acceleration measurements,” in ISSCC Dig. Tech Papers, San Francisco, CA, Feb. 1994, pp. 160–161. [3] L. J. Ristic, R. Gutteridge, B. Dunn, D. Mietus, and P. Bennett, “Surface micromachined polysilicon accelerometer,” in Proc. IEEE SolidState Sens. Actuator Workshop, Hilton Head Island, SC, June 1992, pp. 118–121. [4] W. Henrion, L. DiSanza, M. Ip, S. Terry, and H. Jerman, “Wide dynamic range direct digital accelerometer,” in Proc. IEEE Solid-State Sens. Actuator Workshop, Hilton Head Island, SC, June 1990, pp. 153–157. [5] C. J. Kemp and L. Spangler, “An accelerometer interface circuit,” in CICC Dig. Tech. Papers, Santa Clara, CA, May 1995, pp. 345–348. [6] M. Lemkin and B. E. Boser, “A micromachined fully differential lateral accelerometer,” in CICC Dig. Tech. Papers, May 1996, pp. 315–318. [7] M. Lemkin, B. E. Boser, D. M. Auslander, and J. H. Smith, “A 3-axis force balanced accelerometer using a single proof-mass,” in Proc. Transducers, Chicago, IL, June 1997, pp. 1185–1188. [8] M. Lemkin, M. Ortiz, N. Wongkomet, B. E. Boser, and J. H. Smith, “A 3-axis surface micromachined accelerometer,” in ISSCC Dig. Tech. Papers, Feb. 1997, pp. 202–203. [9] B. E. Boser and R. T. Howe, “Surface micromachined accelerometers,” IEEE J. Solid-State Circuits, vol. 31, pp. 366–375, Mar. 1996. [10] J. H. Smith, S. Montague, J. J. Sniegowski, J. R. Murray, and P. J. McWhorter, “Embedded micromechanical devices for the monolithic integration of MEMS with CMOS,” in Proc. IEDM, Dec. 1995, pp. 609–612. [11] C. Hernden, “Vibration cancellation using rotational accelerometer feedforward in HDDS,” in Data Storage, 2000. [12] D. Y. Abramovitch, “Rejecting rotational disturbances on small disk drives using rotational accelerometers,” in Proc. IFAC World Congr., San Francisco, CA, July 1996. [13] G. Spinola, S. Zerbini, and B. Vigna, “Silicon integrated gyroscope and accelerometers,” in MUSEAS I Workshop, Caserta, Italy, Nov. 8–9, 2001.

391

Alberto Gola was born in Stradella Italy, on September 7, 1959. He received the degree in electronic engineering from the Università degli Studi di Pavia, Pavia, Italy, in 1983. Currently, he is a Design Engineer with STMicroelectronics (at that time SGS Microelectronics) R & D laboratoriess, Milan, Italy, which he joined after his duty in the Army from January 1984 to January 1985. His first experience was in the design of “commodity ICs (op-amps, voltage regulators, power supply supervisors, comparators, etc.). After that, he started working on custom line drivers and receivers (RS232, RS423) requested by Digital Equipment Corporation. In 1990, he was appointed Group Leader and, since then, he has been involved in complex projects (voice coil motor drivers for HDD applications and head driver circuits for inkjet printers). The last successful project (completed in July 1995) was a complete combo driver for inkjet printers, in which, together with the output interfaces, switched mode power supply, thermal management, and pen status control are also present on the chip. From 1985 to 1990, he had the opportunity to design various innovative circuits using the mixed technologies (bipolar CMOS, bipolar CMOS DMOS) that, at that time, were appearing. The result of such “test patterns” design is resumed in papers that have been published or presented at various conferences. From 1996 to July 1997, he was relocated to Irvine, CA, to act as a Local Technical Support for a key customer of STMicroelectronics. In 1997, he started working again in Cornaredo as a Design Group Leader in the Sensors and Microactuators Group, a new group which was built to enter the market of silicon sensors and microactuators. He also took care of the development of electronic interface ICs for inertial sensors (linear and rotational accelerometers). Presently, he is the MEMS Development Unit Deputy and IC Design and Application Group Manager. From 2001 to 2002, he was responsible for seminars at different Italian universities (Universities of Pavia, Pisa, and Lecce) about MEMS. He holds 14 U.S. patents and a number of European patents. His main interest is in analog electronics.

+

+

+

Enrico Chiesa was born in Varese, Italy, in 1973. He received the degree in electronic engineering from University of Pavia in 1999. His thesis was on the development of an analog IC for testing and calibration of MEMS devices. He joined STMicroelectronics, Milan, Italy, in 1999, where he designed some analog circuits for MEMS electronic interfaces. His current interests are in the field of inertial sensors applications.

Ernesto Lasalandra was born in Broni, Italy, on April 1, 1972. He received the engineering degree in electronics from the University of Pavia, Pavia, Italy, in 1997. Since September 1998, he has been with ST Microelectronics, Milan, Italy. His interests are in the design of analog integrated circuits for MEMS.

Fabio Pasolini was born in Pavia, Italy, in 1970. He received the Doctor degree (summa cum laude) in informatics engineering from the University of Pavia in 1994. Since 1995, he has been with STMicroelectronics, Milan, Italy, where he has contributed to the development of many custom integrated circuits. He is currently the Team-Leader for digital IC design and applications of MEMS sensors and he is responsible for the development of smart sensors. His main research interests are in the field of digital integrated circuits, in particular, for low-power and high-speed digital signal processing.

392

IEEE SENSORS JOURNAL, VOL. 3, NO. 4, AUGUST 2003

Michele Tronconi was born in Novara, Italy, in 1975. He received the degree in electronic engineering from the University of Pavia, Pavia, Italy, in 2000. In 1999, he joined the MEMS Business Unit in STMicroelectronics, Milan, Italy, where he is presently a Digital IC Designer. His current interests are in deep-submicron low-power digital circuits design.

Tommaso Ungaretti was born in Pavia, Italy, in 1976. He received the degree in electronic engineering from University of Pavia in 2000. He received the M.S. degree in management of technology from University of Milan, Milan, Italy, in 2001. In 1999, he joined STMicroelectronics, Milan, Italy, where he is presently working as an Analog Designer for the MEMS Business Unit. His technical interests and expertise are in the field of analog circuits for signal processing with both continuous-time and sampled-data techniques. He is currently working on the design of high- sensitivity, low-voltage, low-power integrated sensor interfaces in CMOS and BiCMOS technologies for automotive and industrial applications. He has ublished some papers in conference proceedings and holds patents in the field of capacitive interfaces for integrated micro sensors.

Andrea Baschirotto (SM’01) was born in 1965 in Legnago, Italy. He received the degree in electronic engineering (summa cum laude) and the Ph.D. degree from University of Pavia, Pavia, Italy, in 1989 and 1994, respectively. In 1994, he joined the Department of Electronics, University of Pavia, as Researcher (Assistant Professor). In 1998, he joined University of Lecce, Lecce, Italy, as Associate Professor. Since 1989, he has been collaborating with STMicroelectronics, Milan, Italy, for the design of ASIC. Since 1991, he has been associated with I.N.F.N. (from 1991 to 1998, with the Section of Milan, and, in 1999, with the Section of Lecce) for the design and realization of readout channels for high-energy physics experiments (like L3) and space experiments (like AMS). From 1999 to 2000, he collaborated with MEDICO S.p.A. for the design of a low-power front-end for implantable device (pacemaker) applications. His main research interests are in the design of mixed analog/digital integrated circuits, in particular for low-power and/or high-speed signal processing. He has authored and coauthored more than 40 papers in international journals, more than 50 presentations at international conferences (with published proceedings), two book chapters, and ten industrial patents. In addition, he has coauthored more than 120 papers within research collaborations on high energy physics experiments. Dr. Baschirotto was Guest Editor for the IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS Part II for the special issue on IEEE ISCAS 1998, and he is now serving as Associate Editor for IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS Part II. In addition, he was the Technical Program Committee Chairman for ESSCIRC 2002.