Available online at www.sciencedirect.

com
Procedia
Available online at www.sciencedirect.com

Engineering
Procedia Engineering
Procedia 00 (2011)
Engineering 000–000
25 (2011) 1589 – 1592
www.elsevier.com/locate/procedia

Proc. Eurosensors XXV, September 4-7, 2011, Athens, Greece

Quadrature Compensation for Gyroscopes in electro-

mechanical bandpass ΣΔ-Modulators beyond Full-Scale
Limits using Pattern Recognition
M. Maurera*, T. Northemanna, Y. Manolia,b
a
Fritz Huettinger Chair of Microelectronics, Department of Microsystems Engineering – IMTEK,
University of Freiburg, D-79110 Freiburg, Germany
b
HSG-IMIT – Institute of Micromachining and Information Technology, D-78052 Villingen-Schwenningen, Germany

Abstract

This paper presents a novel quadrature compensation scheme for vibratory rate gyroscopes in closed-loop electro-
mechanical sigma-delta-modulators (ΣΔM). The proposed technique is able to detect and compensate quadrature
errors beyond the full-scale limits of the ΣΔM with a minimal analog hardware effort. The quadrature detection is
based on a pure digital pattern recognition algorithm, the quadrature compensation is done using DC bias voltages.
An appropriate quadrature compensation is reached within a few iterations of the proposed algorithm.

© 2011 Published by Elsevier Ltd.

Keywords: Gyroscope; Quadrature Compensation; Sigma-Delta Modulator; Full-Scale Limits; Pattern Recognition

1. Motivation

Micromachined angular rate sensors are used in automotive applications like rollover detection as well
as in consumer applications like personal navigation. Angular rate sensors consist of two linked spring-
mass systems, where the so called primary resonator is usually driven in resonance using a phase-locked
loop (PLL) [1, 2] or a self-resonant system [3]. The sensor output signal of the secondary resonator
consists of the rate signal caused by the Coriolis effect and a quadrature error signal caused by a
misalignment of the secondary mass due to production imperfections. The quadrature error signal can be

* Corresponding author. Tel.: +49-761-203-7594; fax: +49-761-203-7592.

E-mail address: Michael.maurer@imtek.uni-freiburg.de.

1877-7058 © 2011 Published by Elsevier Ltd.

doi:10.1016/j.proeng.2011.12.393
1590
2 M. Maurer
M. Maurer et al.
et. al. / Procedia
/ Procedia Engineering
Engineering 0025 (2011)
(2011) 1589 – 1592
000–000

Quadrature compensation
Vmax
Quadrature Pattern
DAC
control recognition

V Drive
fs
Vmin
Loop
C/V
Filter 1

Bitstream
Gyroscope
2-Level
Feedback
0
0 10 20 30 40
Electro-mechanical ΣΔM
t/T s

Fig. 1. Electro-mechanical ΣΔ-Modulator with the proposed Fig. 2. Measured output bitstream (lower trace) of the ΣΔM in
quadrature compensation. The pattern recognition block normal operating condition. The bitstream shows a normal ΣΔ
detects overload conditions and extracts the necessary pattern with no visible correlation to the driving voltage (VDrive).
quadrature compensation.

much larger than the rate signal [4, 5] and can usually be compensated by applying DC-bias voltages on
separate quadrature compensation electrodes [4]. The secondary mode can be read out in an open-loop
fashion [6] or with closed-loop concepts which are beneficial due to their higher performance. One
possible closed-loop solution is the electro-mechanical ΣΔM [5, 7]. For ΣΔ readout the large quadrature
error is problematic, since ΣΔM can only handle a certain amount of input signal without overloading and
the output of an overloaded ΣΔM can not be used for rate measurement. The maximum signal is limited
by the full-scale (FS) of the ΣΔM. In a previous published solution the force-feedback of the ΣΔM is
increased to compensate the maximum quadrature error signal [2]. This might decrease the resolution. The
solution presented in this paper is advantageous, since the ΣΔM can be designed for the maximum rate-
signal, resulting in lower power consumption due to smaller feedback voltages and increased sensitivity
with equal dynamic range.

2. System setup and results

2.1. Electromechanical Sigma-Delta Modulator

For quadrature errors larger than FS, the ΣΔM overloads. Therefore the challenge is to detect overload
conditions and to extract the necessary information for the quadrature compensation. Fig. 1 shows a block
diagram for the complete system. The electro-mechanical ΣΔM consists of the gyroscope – for simplicity
only the secondary part of the gyroscope is shown - , a capacitive-to-voltage converter (C/V), the loop
filter, the quantizer, and a feedback block. The C/V converts the capacitive changes of the secondary
detection electrodes to a voltage proportional to the actual position of the secondary mass. The loop-filter
contains the electrical filter part of the ΣΔM. The single-bit quantizer is sampled with a sampling
frequency fs which is typically related to the driving frequency fd of the primary oscillation of the
gyroscope. Here a sampling frequency of fs = 8 × fd is used. The feedback block is generating the two
necessary feedback potentials to compensate the input signals of the gyroscope.
For measurements the sensor system using a 4th-order electro-mechanical bandpass ΣΔM is built up on
PCB [7] and the primary mode is driven in resonance. The used ΣΔM and the design of the loop filter is
described in more detail in [7, 8]. Fig. 2 shows a typical ΣΔ pattern for the non-overloaded ΣΔM without
visible correlation between bitstream and driving voltage.
 M.M.
Maurer et et.
Maurer al. al./
/ Procedia Engineering
Procedia 2500
Engineering (2011) 1589
(20111) – 1592
000–000 1591
3

Vmax Vmax
V Drive

V Drive
Vmin Vmin
1 1
Bitstream

Bitstream
0 0
0 10 20 30 40 0 10 20 30 40
t/T s t/T s

Fig. 3. Measured output bitstream (lower trace) of the ΣΔM in Fig. 4. Measured output bitstream (lower trace) of the ΣΔM in
case of a large positive quadrature error. The bitstream is in case of a large negative quadrature error. The bitstream is in
phase with the driving voltage (VDrive). antiphase with the driving voltage (VDrive).

The behavior of the overloaded system is exemplary analyzed for the used 4th-order electro-mechanical
bandpass ΣΔM. The quadrature error signal is proportional to the position of the primary mass of the
gyroscope and if the primary mass is driven in resonance, phase-shifted by 90° compared to the driving
voltage [4]. For quadrature errors larger than FS the ΣΔM overloads since it can no longer compensate the
quadrature error signal. Due to the 90° phase-shift of the used loop-filter at fd the quantizer input signal
and thus the bitstream is in phase with the driving voltage.
Fig. 3 shows the measured behavior of an overloaded system due to a large positive quadrature error.
As expected the bitstream is aligned to the driving voltage. Due to the phase shifts caused by the sensor
and the electrical filter part the bitstream shows a characteristic behavior, where it changes from ‘0’ to ‘1’
and back at the sample instance after the zero crossing of the driving voltage (VDrive). The bitstream is in
phase with VDrive. In case of a large negative error (Fig. 4) the bitstream shows a similar behavior, but is
inverted due to inverted error signal. This can easily be detected.

2.2. Quadrature detection and compensation algorithm

The upper part of Fig. 1 shows the proposed quadrature detection and compensation block. Apart from
the digital-to-analog converter (DAC) the complete block is build up in the digital domain, so that the
additional analog circuitry is minimized. The pattern recognition block analyses the output bitstream of
the ΣΔM in order to detect overload conditions and if an overload is detected to extract the information
which are necessary to readjust the output value of the quadrature control block. The digital quadrature
control value is then transferred to the analog domain via the DAC and applied to quadrature
compensation electrodes of the sensor.
Fig. 5 shows the proposed quadrature compensation algorithm. The different conditions are detected
using pattern recognition. If a pattern similar or equal to the pattern shown in Fig. 3 and Fig. 4 is detected
it is analyzed, whether a positive or a negative quadrature error is the source of the overload condition and
the output value of the quadrature control block is either increased or decreased. Since overloads can not
only be caused by large quadrature signal, but also for example by a combination of a large quadrature
error and a large rate signal the pattern recognition is set up in a fashion which is able to detect not only
overload events due to a quadrature overload. Therefor the output bitstream is monitored and the distance
between the changes of the bitstream is evaluated. An overload is detected, if the typical pattern with 4
bits ‘1’ followed by 4 bits ‘0’ seen in Fig. 3 and Fig. 4 is present for several cycles of the drive voltage.
1592
4 M. Maurer
M. Maurer et al.
et. al. / Procedia
/ Procedia Engineering
Engineering 0025 (2011)
(2011) 1589 – 1592
000–000

25
No Overload
detected? 20

Q comp [V2 ]
Yes 15
Nominal
operation
Positive Quadrature Negative 10
region of
value
the ΣΔM
5
Threshold
Below

increase decrease
Quadrature Quadrature
control control 0
0 500 1000 1500 2000
Reset t/T s
ΣΔModulator

Fig. 5. Algorithm for the quadrature compensation. If an Fig. 6. Simulation of the proposed quadrature compensation
overload condition is detected the actual direction of the algorithm processing a large quadrature error. The quadrature
quadrature error is extracted according to Fig. 3 and Fig. 4. control value (black) reaches the nominal operation region
(grey) where the ΣΔM is not overloaded within a few steps.

After detecting an overload, the sign of the quadrature error is extracted from the alignement of the
drive voltage’s zero crossings and the changes of the bitstream. To increase the tolerance with respect to
overlaid rate signals a defiance of ±1 sample to the patterns shown in Fig. 3 and Fig. 4 is tolerated. The
compensation value for the quadrature error is adjusted in a successive approximation scheme. After each
overload event a reset is applied to the ΣΔM until the sensor has settled to the new compensation value.
Simulations in Matlab/Simulink show that the proposed compensation scheme can handle quadrature error
signals well beyond FS of the ΣΔM. Fig. 6 shows the applied compensation value for an initial quadrature
error of approximately 10 × FS. The algorithm is able to adjust the quadrature control value to meet the
nominal operation region of the ΣΔM within a few iteration steps. Thus the ΣΔM is no longer overloaded.

References

[1] T. Northemann, M. Maurer, A. Buhmann, L. He, and Y. Manoli, “Excess Loop Delay compensated Electro-Mechanical
Bandpass Sigma-Delta Modulator for Gyroscopes”, Procedia Chemistry, vol. 1, no. 1 , pp. 1183-1186, Sep. 2009
[2] R. Neul, U.-M. Gomez, K. Kehr, W: Bauer, J. Classen, C. Doring, et. al., “Micromachined Angular Rate Sensors for
Automotive Applications”, IEEE Sensros J, vol. 7, no. 2, pp. 302-309, Feb. 2007
[3] T. Northemann, A. Zielger, M. Maurer, and Y. Manoli, “An amplitude regulation for gyroscope drive loops based on
phase-shifting”, Proc. IEEE ISCAS 2010, pp. 3268-3271.
[4] W. Clark, R. Howe, and R. Horowitz, “Surface Micromachined Z-axis Vibratory Rate Gyroscope”, Proc. Solid-State
Sensor and Actuator Workshop, Hilton Head Island, SC, pp. 283-287, June 1996
[5] J. Raman, E. Cretu, P. Rombouts, and L. Weyten, “A Closed-Loop Digitally Controlled MEMS Gyroscope With
Unconstrained Sigma-Delta Force-Feedback”, IEEE Sensors J., vol. 9, no. 3, pp. 297–305, March 2009.
[6] A. Sharma, M. F. Zaman, and F. Ayazi, “A 104-dB Dynamic Range Transimpedance-Based CMOS ASIC for Tuning Fork
Microgyroscopes”, IEEE J. Solid-State Circuits, vol. 42, no. 8, pp. 1790-1802, Aug. 2007
[7] T. Northemann, M. Maurer, S. Rombach, A. Buhmann, and Y. Manoli, “Drive and Sense Interface for Gyroscopes based
on Bandpass Sigma-Delta Modulators”, Proc. IEEE ISCAS 2010, pp. 3264-3267.
[8] A. Buhmann, M. Keller, M. Maurer, M. Ortmanns, and Y. Manoli, “DISCO: a graphical methodology for the design of
Sigma-Delta modulators”, Analog Integrated Circuits and Signal Processing, vol. 60, no. 1, pp. 3-11, Aug. 2009