Design and application of micro-machined inertial sensor detection platform

With the development of science and technology, many new scientific fields have emerged one after another, among which micron/nanotechnology is a cutting-edge technology that attracts attention in many fields. Since the 1990s, after the successful application of micron/nanotechnology in large-scale integrated circuit production, various microsensors and microelectromechanical systems (MEMS) based on integrated circuit technology and micromachining technology have emerged, with an average annual growth rate of 30%. Micromechanical gyroscopes are an important component of them. At present, all advanced industrial countries in the world attach great importance to the research and development of MMG, and have invested a lot of manpower and material resources. Low-precision products have been launched and are developing towards high-precision.

1 Brief working principle of micromechanical vibration gyroscope

The composition of the gyroscope system is shown in Figure 1. It consists of sensitive elements, drive circuits, detection circuits and force feedback circuits. AC voltages with DC bias but opposite phases are applied to the differential capacitors of the comb electrostatic actuator. Due to the action of the alternating electrostatic drive torque, the mass sheet generates simple harmonic angular vibration around the drive axis Z axis in a plane parallel to the substrate. When there is a spatial angular velocity Ω input in the direction perpendicular to the detection axis (X direction) in the vibration plane, the detection mass piece will vibrate up and down around the detection axis (Y axis) under the action of the Coriolis force. This vibration amplitude is very small and can be detected by the capacitor plate located below the mass piece and deposited on the substrate, and processed by the charge amplifier, phase-sensitive detection circuit and demodulation circuit to obtain a voltage signal proportional to the spatial angular velocity.

In the process of scientific research and processing, an important content is to detect the characteristics of the gyroscope, such as the resonant frequency, bandwidth gain, Q value, etc. in the working state, so the development task of the micro-mechanical inertial sensor detection platform was proposed. According to the working principle of the gyroscope, the entire instrument consists of two parts: the drive signal generation part and the output signal detection part of the meter. The drive signal generation part gives the appropriate drive signal to the inertial sensor to be tested to put the sensor in a working state. The signal detection part requires the detection of tiny capacitance changes. After amplification and demodulation, the analog quantity is converted into a digital quantity and collected into the PC, and the output signal is analyzed to determine the characteristics of the inertial instrument.

2 Micro-capacitance detection technology

In the MMG detection technology, the capacitance sensor is used to detect the vibration angular displacement of the test mass under the action of the Coriolis force to obtain the input angular rate signal. Due to the small size of the gyroscope, in order to obtain a medium accuracy of 10°/h, the capacitance measurement resolution is required to reach (0.01×10-15)~(1×10-18) farads. Therefore, for micromechanical accelerometers and micromechanical gyroscopes, detecting the capacitance change between the test mass and the substrate is a key technology. There are currently three microcapacitance detection schemes used in MMG: switched capacitor circuit, unity gain amplifier circuit and charge amplifier circuit.

2.1 Switched capacitor circuit

The basic principle is to use the charging and discharging of the capacitor to convert the unknown capacitance change into a voltage output. The measurement circuit includes a charge amplifier, a sample-and-hold circuit, and the timing of the control switch, as shown in Figure 2.

During the measurement process, the unknown capacitor (C1, C2) is first charged to a known voltage Vref, and then discharged. The charging and discharging process is controlled by a certain timing and repeated continuously, so that the unknown capacitor is always in a dynamic charging and discharging process. C1 and C2 are discharged continuously, and the current pulses are converted into voltages through the charge amplifier. Then, the output Vc is obtained through the sample and hold. Substituting the formula ΔC=2C0·x/d0, the transfer function of the capacitance detection circuit can be obtained as:

2.2 Unit-gain amplifier circuit

ADXL50 (5g micromechanical accelerometer) jointly developed by AD and UC Berkeley adopts a unit-gain amplifier circuit.

Figure 3 is the equivalent circuit of the unit-gain amplifier. In Figure 3, Cp is the distributed capacitance, Cgs is the pre-stage input capacitance, and Rgs is the input resistance. When the carrier frequency is within the passband of the amplifier, the pre-stage input resistance can be ignored. From Figure 3, it can be obtained that the useful signal output of the pre-stage is:

(Vs-Vout)jω(C0+ΔC)+(-Vs-Vout)jω(C0-ΔC)

The distributed capacitance Cp is about 10pF, and the input capacitance Cgs is about 1~10pF, which are generally larger than the nominal capacitance C0 of the sensor (about 1pF). It can be seen that their existence greatly reduces the sensitivity of capacitance detection. To improve the sensitivity of the circuit, the influence of Cp and Cgs must be eliminated, and the measure usually adopted is equipotential shielding.

2.3 Charge amplifier circuit

The charge amplifier circuit is shown in Figure 4. It uses an inverting input operational amplifier with low input impedance. Cp represents the distributed capacitance, Cf is the standard feedback capacitance, and Rf is used to provide a DC channel for the amplifier to keep the circuit working properly. Rf should be selected so that the time constant RfCf is much larger than the carrier period to avoid output waveform distortion. However, Rf is too large, which will bring inconvenience to future circuit integration. A small resistance resistor can be used to form a T-type network instead of a large resistance resistor.

If the operational amplifier has sufficient open-loop gain and the inverting input terminal is a good virtual ground, then the potential difference between the two input terminals is zero. Therefore, the distributed capacitance Cp of the inverting input terminal to the ground and the input capacitance Cgs of the amplifier will not affect the circuit measurement. Compared with the unit gain amplifier circuit, the charge amplifier circuit has a simpler structure and does not need to consider the equipotential shielding problem; it only needs to convert the influence of stray capacitance into distributed capacitance to the ground, that is, to perform reasonable ground shielding, to obtain better results. Although the

input capacitance and the distributed capacitance of the inverting input terminal to the ground can be ignored in the charge amplifier circuit, the output still has a large attenuation when detecting small capacitance changes. This is caused by the distributed capacitance Cio of the input and output terminals of the amplifier. When the carrier voltage frequency is greater than 1/(2πRfCf) and less than the cut-off frequency of the amplifier, the output voltage Vout should be expressed as:

System composition and working principle of the detection platform

The working principle of the system is shown in Figure 5. After applying appropriate excitation signals to the inertial sensor, the moving plate of the sensor is in a vibrating state, and the capacitance between the upper and lower plates changes periodically. The charge amplifier circuit is used to extract the signal, and after AC amplification and demodulation, it is converted into digital quantity through A/D conversion and collected in the microcomputer. The output response of the sensor is observed, laying the foundation for the next step of using software methods to analyze the time domain and frequency domain characteristics of the micromechanical inertial sensor.

3.1 Excitation signal generator

According to the working principle of the micromechanical wheel vibration gyroscope, a maximum of 4 excitation signals are required. The excitation signal is a sine wave, and each two signals are in opposite phases. In order to measure the frequency characteristics of the gyroscope, the frequency of the excitation signal needs to be continuously changed. At present, the resonant frequency of gyroscopes of different designs is between several hundred hertz and 10 kilohertz, and the excitation signal also needs to be adjusted within this range. In addition, the driving torque of the gyroscope is equal to the product of the AC component and the DC component of the driving signal, so a positive or negative DC bias must be applied to keep the gyroscope in normal working state. The combination of AC phase and DC bias is shown in Table 1.

The frequency of the sine wave generated by the general RC oscillation circuit is adjusted by changing the R and C values, and cannot be adjusted continuously over a large range. Therefore, the design uses a digital method to synthesize the analog waveform, and its principle is shown in Figure 6. The 8254 in Figure 6 is a software programmable counter. It contains 3 independent 16-bit counters, and the maximum counting frequency can reach 8MHz. In the design, a 3MHz clock is input and two counters are used in series, which can increase the frequency control range. The square wave signal generated by 8254 is used as the counting pulse input of the subsequent parallel counter. The parallel counter is composed of two 74LS161 chips to form an 8-bit binary cycle counter. When the 74LS161 counts to the maximum value, it will automatically clear to zero and start counting again. Its output can be used as the address signal of the E2PROM 2817A (that is, the number of sampling points in each sine cycle is 256). The data reading time of 2817A is 150ns. When designing the circuit, both its chip select and read signals are set to be valid to increase the data reading speed. The D/A conversion uses the DAC-08 current output D/A converter. The current output time is 85ns, and the amplifier uses a high-speed and high-precision operational amplifier OP-37. Similarly, the chip select and conversion start signals of the D/A converter are always valid, and its output follows the input changes to increase the conversion speed. The experimental results show that this signal generator can fully generate a sine wave with adjustable frequency within 10kHz. Moreover, using the programmable counter 8254, the frequency of the output sine wave can be adjusted by software methods. If you want to output a non-sinusoidal waveform, you only need to modify the data of the E2PROM to output a periodic waveform of any shape.

3.2 Low-pass tracking filter

The digital signal generator has the advantage of flexible control, but the output signal is not smooth enough and there will be step waves. In situations where the signal requirements are relatively high, filtering is also required. The frequency range of the signal in this design is very large: from a few hundred hertz to 10 kilohertz. In order to further improve the signal quality, the AD633 analog multiplier is used to form a low-pass tracking filter, and its principle is shown in Figure 7.

The cut-off frequency of the passband is controlled by the voltage EC, the output is OUTPUTA, the cut-off frequency is:

OUTPUTB is the direct output of the multiplier, and the cutoff frequency is the same as that of the RC filter:

This filter has a simple structure, no switching capacitors, low noise, and generally uses a digital-to-analog converter to control EC, making it relatively easy to control the passband frequency.

3.3 AC amplifier

After the excitation signal is applied, the micromechanical inertial sensor is in a vibrating state. The sensor has differential microcapacitance changes C0+ΔC and C0-ΔC. The charge amplifier circuit is used to extract ΔC. This voltage signal is still very weak and needs to be further amplified. Therefore, the AC amplifier shown in Figure 8 is used.

The AC amplifier is composed of 4 operational amplifiers with amplification factors of -1, -2, -5, and -10 in cascade, which further amplifies the measured signal and adjusts the amplitude to adapt to the input of the demodulator. The switch in Figure 8 uses the ADG211 analog switch. By controlling the opening and closing of the analog switch, one or several amplifiers can be arbitrarily selected to participate in the work, so as to adjust the amplification factor to integer multiples of positive and negative 1, 2, 5, 10, 20, 50, and 100. For example, if the analog switches S0, S2, S8, and S13 are closed and all other switches are opened, the total amplification factor of the AC amplifier is: (-1)×(-2)×(-10)=-20.

3.4 Data acquisition system

When using a computer bus, there must be an interface between the computer and the peripherals. This system uses a dual-port RAM as a data cache. The signal is first sampled and stored in it, and then transmitted to the host in groups, thereby effectively exerting the efficiency of the master and slave resources, and the design is relatively simple.

3.4.1 System Working Principle

The basic composition principle of the system is shown in Figure 9. It mainly includes dual-port RAM, logic control module, A/D converter group, and computer interface. After the host starts the logic control module through the interface, the CPU resources are open to other requests. The logic control module sends a control signal to start the A/D converter and perform sampling, and stores the conversion results in the dual-port RAM. When the data in the RAM reaches a certain amount, the logic control module sends an interrupt request to the computer. After receiving the request, the host enters the interrupt service program and sends a command to the logic control module to decide whether to continue sampling and read the data in the RAM into the memory.

3.4.2 Hardware Design

This design uses Cypress's CY7C136 (2k×8bit) dual-port RAM. Its two ports have independent control signals, chip select CE, output enable OE and read/write control R/W. This set of control signals allows the two ports to be used like independent memories. When using this device, it should be noted that when two ports access the same unit, the data reading result may be incorrect. There are two ways to solve this problem: one is to monitor the busy signal output. When the busy signal is detected to be valid, the access cycle is lengthened, which is a hardware solution; the other method is to ensure that the two ports do not access a unit at the same time in software, that is, to divide the dual-port RAM into blocks. This system adopts the latter, connecting the busy signal output to the positive pole of the power supply through a pull-up resistor.

In the system, the role of the logic control module is very important, and it is the core of controlling sampling, storage, and computer interface. In order to facilitate the setting of parameters such as sampling rate, this system uses MCS-51 microcontroller in this module. In this way, the sampling rate can be set by programming.

The information exchange with the host includes:

(1) receiving the host control signal to decide whether to start sampling;

(2) sending an interrupt request to the host after the storage area is full.

This system uses the address bus of AT89C51 to select the storage unit of RAM, write it, and store the sampling results in the corresponding unit.

3.4.3 Software Design

The system software includes the host program and the 89C51 program in the logic control module. The key to the software is the microcontroller controlling the A/D converter and the memory part. The software flow is shown in Figure 10.

As for the sampling rate of the system, it is generally achieved by calling a timer interrupt.

The micro-mechanical inertial universal detection system is highly targeted (specially used for micro-mechanical gyroscopes and accelerometers), which can realize automatic testing of sensitive components, automatically scan the frequency to measure the resonant frequency and Q value of the sensor, and can also realize hardware function readjustment to a certain extent, and has achieved good results in actual detection.