Abstract: This paper designs a detection system that can detect the roundness of the inner ring of a bearing. The system uses an optical encoder for angular sampling, digital display, and is connected to a host computer. It overcomes the shortcomings of traditional roundness testing systems, such as the inability to prevent reverse rotation, difficulty in reading, susceptibility to false readings and invalidation, and lack of data processing functions. Keywords : Bearing detection; displacement sensor; single-chip microcomputer; optical encoder; roundness instrument Rolling bearings are highly interchangeable standard components. They have advantages such as low friction, easy start-up, simple lubrication, and easy replacement. They are important support parts for transmitting motion and bearing loads in various machines, and are almost indispensable components in mechanical structures. With the development of industry, higher requirements are placed on the performance, life, and reliability of bearings. The performance, lifespan, and reliability of rolling bearings depend on their design, manufacturing, and testing processes. Testing is a crucial step in improving bearing performance. The inner ring of the bearing is in close contact with the shaft, and it is susceptible to not only dimensional errors but also roundness, roughness, and waviness errors. This paper uses a displacement sensor to measure the roundness of the bearing inner ring, a photoelectric encoder to control the system for angular sampling, and an 8-bit W78E52 microcontroller for the control and data processing unit. Data is transmitted to the host computer via a serial port for centralized data analysis. An external LED displays the data processing results, making readings convenient and thus realizing intelligent and digital rolling bearing roundness detection. 1 System Overall Design The system mainly consists of three parts: a detection section, a signal acquisition and processing section, and an input/output section. The overall architecture is shown in Figure 1. The detection section includes a sensor, a photoelectric encoder, amplification, and filtering. The signal acquisition and processing section is responsible for AD conversion, system control, and storage of sampled data. The input/output section consists of LEDs and a keyboard (as shown in Figure 1). [align=center] Figure 1 System overall block diagram[/align] The bearing inner ring is fixed on a floating probe and two fixed probes. The displacement sensor is connected to the floating probe through a mechanical device. The photoelectric encoder controls the sensor to sample at equal angles, and the sensor signal is amplified and filtered before entering the ADC. The microcontroller processes and stores the signal from the ADC. The bearing inner ring rotates one revolution, and the data acquisition is completed. Finally, the microcontroller finds the maximum and minimum values ​​in the data and calculates the difference, which is displayed through the LED. The relationship between the maximum reading difference F and the roundness error f of the workpiece rotating one revolution is given by the formula where K is the response coefficient, which is obtained from GBT 4380-1984 [1], that is, the difference divided by the response coefficient is the roundness error. 2 Detection section 2.1 Selection of sensor According to the roundness meter standard JB/T 10028 1999, in the instrument error class A, the linear error of the measurement system is not greater than 2% of the full scale, and the sensitivity of the measurement system is not greater than 0.02μm [2]. This system adopts a contact measurement method, therefore, a differential transformer inductive sensor with advantages such as good stability, simple and reliable structure, and strong anti-interference is selected as the displacement sensor. The Zhongyuan Measurement Instrument E-DT-80SB sensor selected in this system has advantages such as high measurement accuracy, high sensitivity, and easy clamping and positioning, and meets the roundness tester standard JB/T 10028 1999. Although its dynamic response frequency is not high, it can fully meet the speed response requirements (sampling points) for roundness measurement. The performance is as follows: Total stroke (mm): 3 Measurement range (mm): ±0.5 Linearity error: ±0.5% Repeatability error (μm): 0.2 2.2 Differential Transformer Displacement Sensor Measurement Circuit The differential transformer sensor outputs AC voltage. If measured with an AC voltmeter, it can only reflect the magnitude of the armature pull, but not the direction of movement. At the same time, its measured value will include zero-point residual voltage. Therefore, in actual measurement, a phase-sensitive detector circuit and a differential rectifier circuit are usually used. The phase-sensitive detector circuit needs to use the primary excitation voltage as a phase reference to determine the polarity of the output voltage. This requires an excitation signal source with constant amplitude and frequency, and compensation for the phase shift of the primary and secondary windings of the differential transformer and the errors caused by temperature and frequency fluctuations. The differential rectifier circuit does not need to consider the phase problem, and the circuit is relatively simple. This paper selects the differential rectifier circuit to perform post-processing on the output signal of the differential transformer (as shown in Figure 2.1[3]). [align=center] Figure 2.1 Full-wave differential rectifier circuit diagram[/align] 2.3 Signal amplification The signal from the sensor is generally weak, usually only a few millivolts to tens of millivolts. The output signal range of this sensor is 0.028mv～100mv, while the A/D converter requires a full-scale input of ±5v. Therefore, it is necessary to amplify to improve resolution and reduce noise, and also to make the maximum value of the conditioned signal equal to the maximum input value of the A/D, so as to improve the conversion accuracy. For this problem, there is only one signal input channel. To prevent the maximum amplified signal from exceeding the ADC's full scale, the amplification factor (i.e., the signal amplification circuit) is set to gain K=50. 2.4 Filtering Circuit: In roundness measurement, various noise signals can render the measurement data unreliable. Therefore, it is necessary to filter the original measurement data to remove unnecessary high-frequency signals and obtain a signal in a specific frequency band. A second-order RC active low-pass filter is used in this system (as shown in Figure 2.2). 2.5 Counting Circuit: Counting methods can be implemented in software or hardware. While pure software counting is simple, it is slow and prone to errors. Using an external counting chip is faster, but the hardware circuit is complex and costly. Combining these two methods, this paper adopts a combined software and hardware approach, using a counter within the microcontroller to implement the counting. The inner ring of the bearing is rotated manually at a speed that is neither too fast nor too slow. The requirements for the resolution, maximum response frequency and maximum allowable speed of the photoelectric encoder are not high. The photoelectric encoder does not bear a large external force, so the requirements for its mechanical performance are not high. Considering the working environment, this paper selects the Koyo rotary encoder TRD-2E A to complete the system design. Its performance specifications are as follows: Item: TRD-2E A Resolution: 1024 Pulse/Revolution Output Signal Form: A·B Two-phase Maximum Response Frequency: 200kHz Maximum Allowable Speed: 5000rpm Starting Torque ≤0.001N·m [align=center] Figure 2.2 Second-order voltage-controlled voltage source low-pass filter circuit[/align] [align=center] Table 1 Uncertain whether it is high or low level[/align] Connect the A terminal of the photoelectric encoder to the D terminal of the D flip-flop and the external interrupt INT1 terminal of the microcontroller. Connect the B terminal of the photoelectric encoder to the CLK terminal of the D flip-flop. The pulse after the D flip-flop, i.e. the direction control pulse (DIR), is connected to the external interrupt INT1 terminal of the microcontroller (as shown in Figure 2.3 [4]). Open the corresponding interrupt and set the gate bit GATE of T1 to 1. At this time, in addition to setting TR1 to 1, the INT1 pin must also be high level in order to start the counter. As shown in Table 1, the inner ring of the bearing rotates forward only when DIR is high and A is positive or negative. Therefore, when the condition of the inner ring rotating forward is met, the sensor reads the value and counts. When the inner ring of the bearing rotates once, it enters the interrupt program, sends the collected data to the PC, and calculates the difference to obtain the roundness. [align=center] Figure 2.3 Wiring diagram of the counting circuit[/align] This circuit does not collect data when the inner ring of the bearing rotates in reverse or does not rotate. This ensures the accuracy of the data and eliminates the inaccuracy caused by the operator's jitter causing the inner ring to rotate in reverse. 3. Selection of A/D converter For the selection of A/D converter, conversion rate and resolution are two important parameters. Its design is as follows [5]: 3.1. Selection of conversion rate In the system, the photoelectric encoder controls the sampling of the ADC. When the photoelectric encoder rotates once, the ADC samples 1024 times. It takes at least 0.8 seconds to manually rotate the photoelectric encoder once, that is, the maximum sampling rate of the photoelectric encoder is 1.25. Therefore, the conversion rate of the ADC should be greater than the sampling rate of the photoelectric encoder of 1.25. 3.2 Resolution Selection The sensor measurement range is ±0.5mm, and the measurement accuracy is 1μm. Through actual measurement, the maximum output signal of the sensor is 1.25V, which is the maximum measurement displacement of the sensor ±0.5mm. Therefore, when the probe moves radially by 1μm, the sensor output signal voltage is u, which is the minimum output signal of the sensor. According to the resolution formula 4.1, the actual ADC resolution of this system is selected as n=10. Taking a larger value of n improves the A/D conversion accuracy, but it is expensive and not economical. In practice, the sensor output signal is too weak and needs to be amplified by an amplifier circuit. The amplified signal sent to the A/D converter reduces the A/D resolution, so n=10 bits is selected. Based on all the calculation results, the actual working conditions, and economic considerations, it is decided to use the MC14433 dual-slope A/D converter manufactured by Motorola based on CMOS manufacturing technology. Its operating performance is as follows: 3 (1/2) bit dual-slope integrating ADC; operating voltage range: dual power supply 4.5-8V; A/D conversion accuracy: 0.05% (11-bit binary number); corresponding to a clock frequency of 50-150kHz; conversion rate: 4-10T/s (greater than the conversion rate of a photoelectric encoder). 4. Conclusion The rolling bearing roundness meter introduced in this paper can accurately measure the roundness of bearings, with advantages of simple circuitry and stable reliability. It uses a photoelectric encoder for equal-angle sampling to prevent errors caused by jitter (reversal), thus improving measurement accuracy. The use of an LED display avoids the problems of difficult readings and high labor intensity associated with traditional bearing inner ring roundness measuring instruments. Sampling data is sent to a PC via a serial port, facilitating centralized analysis of bearing data and compensating for the lack of data processing capabilities in traditional testing instruments. This roundness meter has a simple and reliable structure, high measurement accuracy, good stability, and good economy, and has good application prospects. References [1] GBT 4380-1984 Methods for determining roundness error: two-point and three-point methods, issued by Beijing State Bureau of Standards, 1984 [2] JB/T 10028 1999 Roundness tester, issued by Beijing State Bureau of Industrial Machinery, 1999 [3] Kang Huaguang, Chen Daqin, Zhang Lin, Basic Electronic Technology, Analog Part, Higher Education Press, 2005 [4] Pan Mingdong. Several counting methods for output pulses of photoelectric encoders [J] Electronic Engineer, 2004, (08) 69-71 [5] Guan Bingliang. Digital transformation of TALYROND73 roundness tester [D] Hefei University of Technology, 2006