Maximilian Marx

The performance of microelectromechanical system (MEMS) gyroscopes can be significantly enhanced by matching the resonance frequencies of the drive and sense mechanical resonators. However, the precise control of the resonance frequency in a closed loop readout circuit is challenging. This paper presents an automatic detection and control circuit, which performs a background tuning of the MEMS element. The tuning method is based on the observation of noise at the output of a continuous time Σ modulator readout circuit, so that no modification of the sense loop such as additional input force generation is required. Parasitic modulation of the sensor’s zero rate output due to the mode matching loop is canceled by an additional offset compensation circuit. The system is validated on an FPGA connected to a MEMS gyroscope and an ASIC readout circuit. The digital mode matching circuit would consume a power of 21μW on an area of 0.17 mm2 in a 0.35μm CMOS technology.

Electronic readout circuits for microelectromechanical gyroscopes suffer from high power consumption caused by the electrostatic sensor actuation with signals above the standard chip supply voltage. This paper presents an electronic circuit principle and implementation, which exploits the fact that ac actuation signals in the range of the chip supply voltage are sufficient, if they are combined with a high-voltage (HV) dc biasing. Therefore, the developed system performs the generation of the time-varying signal in the low-voltage domain and uses precharged flying capacitors to efficiently add a high dc biasing voltage. The circuit consists of a closed-loop low ripple charge pump, low-voltage driver amplifier, and an HV common frame biasing amplifier. The fully on-chip implementation occupies an area of 0.58 mm² and reduces the power consumption to 141 μW.

This paper presents a needle-shaped neural probe which integrates an incremental $\Delta \Sigma $ -ADC under each electrode in an area of $70\times 70\,\,\mu \text{m}^{\mathrm{ 2}}$ . The modular architecture avoids global routing of noise sensitive neural signals and connects with a digital four-wire interface. The re-configurable 5.6/11.5 mm needle enables parallel readout of all 64/144 integrated recording sites and features a constant width and thickness of $70\times 50\,\,\mu \text{m}^{\mathrm{ 2}}$ from top to bottom allowing to be fully implanted for deep brain monitoring applications. Platinum or iridium-oxide electrodes can be deposited in a custom post-CMOS process. The functionality of the recording system is demonstrated in an in vivo setup. The noise performance of 6.02 $\mu \text{V}_{\text {rms}}$ for LFP signals (1–300 Hz) and 10.46 $\mu \text{V}_{\text {rms}}$ for AP signals (300 Hz–10 kHz) is obtained at a power consumption of 39.14 $\mu \text{W}$ per recording site.

The evolution of tissue-penetrating probes for high-density deep-brain recording of in vivo neural activity is limited by the level of electronic integration on the probe shaft. As the number of electrodes increases, conventional devices need either a large number of interconnects at the base of the probe or allow only a reduced number of electrodes to be read out simultaneously [1,2]. Active probes are used to improve the signal quality and reduce parasitic effects in situ, but still need to route these signals from the electrodes to a base where the readout electronics is located on a large area [3,4]. In this work, we present a modular and scalable architecture of a needle probe, which, instead of routing or prebuffering noise-sensitive analog signals along the shaft, integrates analog-to-digital conversion under each electrode in an area of 70×70μm ² . The design eliminates the need for any additional readout circuitry at the top of the probe and connects with a digital 4-wire interface. The presented reconfigurable 11.5mm probe features a constant width of 70μm and thickness of 50μm from top to bottom for minimal tissue damage with 144 integrated recording sites and can be fully immersed in tissue for deep-brain recording applications.

This paper presents the implementation of a self-test for a gyroscope readout based on electro-mechanical ΔΣ modulation. Commonly, sensor element and readout ASIC are fabricated on separate wafers. Therefore, the ability to separately characterize the sensor element and the ASIC before packaging is desirable in order to reduce unnecessary expense. For the proposed self-test, a charge integrator with collocated detection and feedback is configured to generate an additional feedback path, which is used to operate a purely electrical ΔΣ modulator on wafer level. The self-test is performed in two steps. First, an automatic compensation controller configures the collocated force-feedback and thereby verifies its functionality. In a second step, an offset is added to the settled value of the controller to generate a negative feedback path to the input of the readout interface. The resulting purely electrical ΔΣ modulator is utilized to validate the functionality of every single circuit component of the later operated electro-mechanical ΔΣ modulator. The proposed self-test makes use of the already available hardware and therefore avoids an additional hardware overhead. Numerical simulations are performed to verify the proposed compensation controller concept as well as the self-test itself. The simulation results of the self-test are confirmed by measurements.

This paper presents the implementation of a bandpass filter tuning circuit solely based on noise observation for a continuous-time (CT) delta-sigma modulator gyroscope readout system. By evaluating the noise at the electronic filter input, a filter frequency deviation of less than 50 Hz or approximately 0.5% of the gyroscope's drive resonance frequency is achieved, which minimizes the angular rate noise floor. The automatic tuning works in the background during normal angular rate readout with input signals up to the full scale, over a temperature range of 115 °C, and with interference signals of up to -27.2 dBFS. The power (27 μW) and area (0.06 mm²) overhead to the overall system is less than 2%. The overall CT readout system achieves a noise floor of 0.002 °/s/√Hz and a bias instability of 0.9 °/h while consuming only 1.71 mW.

MEMS gyroscopes are used in closed-loop configuration (CL) to satisfy the demand for high-performance and stable inertial sensors [1]. Due to the higher complexity and power consumption compared to open-loop solutions, these systems have usually been unsuitable for mobile battery-driven devices, e.g., for indoor navigation. Recently, the utilization of CT-ΔΣM for the readout of gyroscopes has shown to be a promising approach for reduced power consumption in a CL system [2]. In general, an accurate matching of the electrical BPF [2] to the drive and sense resonance frequencies of the sensor [3] is a prerequisite for maximizing SNR. For systems with drive frequencies fd of some tens of kHz and with a typical angular rate bandwidth of BW=50Hz, the frequency matching needs to be as precise as BW/fd2 than competing approaches (see Fig. 9.4.6).

This paper reports on the implementation of an interface ASIC for MEMS vibratory gyroscopes exhibiting a wide tuning range from 7 to 31 kHz with a power consumption of 1.6 mW and a dynamic range of 92 dB over a signal band of 40 Hz. The drive loop of the system employs a high voltage stage based on flying capacitors for the excitation of the drive mass allowing a startup time of 44 ms. For the detection of the Coriolis signal, a fourth-order closed-loop electromechanical sense loop deploying continuous-time (CT) circuit techniques and ΔΣ-modulation is implemented. The sense loop deploys a high-Q electronic resonator based on gm-C circuits and a collocated charge-integrator for signal readout and for generating feedback forces. The readout interface with an active circuit area of 2.42 mm2 is fabricated in a 0.35 μm CMOS technology with a HV-module. The ASIC achieves a noise floor of 0.007°/s/√Hz over 40 Hz while the full-scale range is higher than ±1400°/s and the bias stability measures 2.6°/h. The wide tuning capability of the ASIC is demonstrated on the basis of two tested MEMS gyroscopes for different applications with an equal ASIC performance.

A front-end circuit for closed loop continuous-time delta-sigma (CT ΔΣ) micro-electro-mechanical gyroscope readout circuits is implemented. This work presents for the first time a CT collocated feedback, which simultaneously uses the detection capacitors of the sensor for the signal readout and the feedback. This is realized by the modulation of the input common mode of the readout amplifier and relies solely on CT techniques to achieve a low noise floor. Additionally, the concept decreases the number and the complexity of the signals in the high voltage (HV) domain. Therefore, the power and area demands of the developed HV interface for quadrature compensation and mode matching are reduced. The circuit is implemented in a 0.35 μm technology, requires an area of 0.65 mm ² and consumes 770 μW.