In the thesis integrated sensor microsystem for magnetic field measurement with implemented self-calibrated sensitivity technique in the temperature range from –40 ℃ to 125 ℃ is addressed. The core of the microsystem, realized in a standard 0.35 μm CMOS (Complementary Metal Oxide Semiconductor) technology, represents a Hall element. One of the main drawbacks of this type of sensor is the drift of sensitivity in the presence of unpredictable and irreversible effects, such as aging, mechanical or piezoelectric forces, humidity and, above all, the influence of temperature, since the Hall element is implemented in a temperature unstable N-well region. In addition, temperature affects the sensitivity of the entire sensor since the characteristics of the electronics which composes the signal processing path are susceptible to the temperature effect. Conventional approaches of sensitivity calibration during the production line could not be applied for the above mentioned distortion influences, but should be handled with an active on the fly calibration technique that adapts based on the operating conditions throughout the lifetime of the integrated sensor.
The self-calibration presented in this paper is based on the adjustment of the sensitivity
with respect to the evaluation of the temperature-dependent reference signal related to the
known and stable reference magnetic field generated by the integrated coil in the integrated
circuit. The coil integration on four metal layers of the CMOS process is realized in the range
of μm dimensions above the Hall element, which contributes to improved coil efficiency, since the external coil at the position of the Hall element would generate a much lower intensity of the magnetic field [1]. The efficiency of the integrated coil could be improved at smaller dimensions since the intensity of the magnetic field is inversely proportional to the diameter of the integrated coil. Despite the improved efficiency, a coil with too small dimensions could generate a negative magnetic field at the location of the Hall element and generally contribute to a rather unevenly distributed magnetic field that directly affects the reference signal used in the process of calibration. With the aim of realizing an optimal geometry of the integrated coil that produces the highest possible magnetic field intensity with uniform distribution and reduction of coil winding resistance, the multi-objective optimization of the 3D model of the integrated coil is performed. The realization of the integrated coil with minimum resistance promises lower thermal losses and consequently contributes to the more accurate calibration of the sensitivity of the overall magnetic sensor signal path. The optimization of the electromagnetic properties of the integrated coil is performed in the Comsol Multiphysics environment, which interchanges 3D model parameters based on the data of the used CMOS technology with the Solidworks environment. The optimized model with 11 turns at four metal layers promises the efficiency of 376 mT/A.
Simulations of the entire electrical circuit with the main structures starting with the
Hall model, the front-end circuit to acquire the signal, the delta-sigma modulator, the DA
converter and the output filter are performed using the simulator HSPICE. The entire signal
path from acquisition to signal processing to output is implemented differentially, which
improves the suppression of common mode distortion. The given approach mitigates not
only the effect of Hall element sen sitivity variations but also that of the entire signal path,
which relaxes the requirements for the design of signal processing stages with temperature
independent gains. Simulation results show improved calibration in the temperature range
from -40 ℃ to 125 ℃ by a factor of 4.2 when multibit quantization is used instead of 1-bit
quantization in combination with the 1-bit DA converter. Over the entire set of process
parameters, the overall calibration performance is slightly degraded, but not significantly, so
the presented approach also mitigates the effect of process variations on the calibration to a
good extent.
The evaluation of the whole microsystem is performed with the measurement of the analog
output signal, but the approach allows further processing of the digital bit-stream signal in
a process of DSP (Digital Signal Processing), which is not the subject of this work. The
calibration of the sensitivity of the integrated magnetic sensor shows some distortions, but it
fulfils the functionality of the presented approach with an accuracy of the calibration of the
sensor sensitivity of 85 ppm/℃. The efficiency of the integrated coil is 131 mT/A, and the
influence of the specific coil turn on the overall magnetic field in region of the Hall element
is in a good correlation with the simulation results.
|