A large quantity of steel balls automatic precise counting device based on dual sensor auto-correcting method

Abstract

A large quantity of Steel Balls Automatic precise Counting device is introduced in this paper. Inductance sensors are used for ball detection and STC15F2K60S2 microprocessor is used to do the signal processing. When steel balls are sent to the PVC pipe at the same time, some balls will connect together because of the collision. The sensor will send a connected signal instead of multiple signals if it meets the connected balls. This could cause counting mistake of the device. A correction method is used to solve this problem. The device setup angle could be gotten by using dual sensors to get the average moving speed of the steel balls according to the fixed relations between the setup angle and the speed, which is not affected by whether the steel balls overlap or not because it used two voltage Jump edge to judge the start and end moment of timer. After that, the device setup angle is used to get the standard moving time to single sensor according to the known function relationship. The counting numbers are corrected by using the standard duration to compare with each recording low voltage duration passing single sensor. This method solves the counting mistake problem when a lot of steel balls are pushed into the PVC pipe together. The device improves the efficiency and accuracy of the ball counting.

Keywords

Steel ball counting signal aliasing error correction dual sensor structure

1. Introduction

Steel ball is widely used in ball bearings, rolling linear guide and other industrial components. The usual method for steel ball counting is the weighing counting method, which use gravity sensor as a detection element to get the total weight of the steel ball, and then the total weight is divided by the single steel ball weight to achieve ball number. Due to the surface of the steel ball with anti-rust oil and manual reading errors, the ball number is not accurate by using this method. In order to ensure that the number of steel balls provided is sufficient for assembly, steel ball suppliers often provide more steel balls in a certain proportion during the weighing and counting process, which causes unnecessary waste. In the automated production line, the steel balls often need to move from one process to another, and the common methods for detecting the quantity are photoelectric measurement method and image measurement method [1, 2, 3, 4]. These counting methods all have certain limitations. Some methods require the steel balls to be arranged neatly and pass through the sensor in turn [5, 6], which will cause low efficiency of detection. Some are unstable due to the influence of external ambient light. Therefore, it is of great practical importance to design an automatic and accurate counting devices which can quickly detect batch steel balls especially on the automatic production line.

2. Signal analysis when single steel balls and large quantity of steel balls pass through the pipe

Steel ball is a kind of metal ball, its main component is iron. The commonly used sensor for the detection of this material is inductive sensor. PVC pipe with inner diameter 3–4 mm larger than the diameter of steel ball is selected as the steel ball transmission channel, which is installed and fixed at a certain inclined angle. The sensor is mounted on the outside of the PVC pipe and the steel balls are sent into the PVC pipe by the conveyor or manual. In order to accurately detect the steel ball in the pipe, a certain detection distance is needed for the sensor. HuGongLJ18A3-8-J/EZ inductor sensor is selected as the detection sensor, which is an normally open inductive proximity switch. Its supply voltage is DC12V and detection distance is 8 mm. When there is no steel ball passing through the sensor, its output voltage is 5 V. When the steel ball is detected, the output voltage jumps to 0 V. Figure 1 shows the picture of the inductor sensor and the output signal when single steel ball passes through the sensor.

Figure 1.

Picture of the inductor sensor and the signal of single steel ball passing through sensor.

If the steel ball pass through the pipe one by one, the pulse counting method can be used to count the ball number. When a single steel ball passes through, the output signal of the sensor is connected to the microprocessor. The microprocessor sets the falling edge to trigger the counting. For each steel ball passing through, the count number plus 1.

In the actual process of material transportation, due to the need of transmission efficiency, steel balls are not transported one by one. Usually, the steel balls are poured directly into the pipeline in batches. Due to the collision of steel balls with steel balls and the collision of steel balls with inner wall, some steel balls will be squeezed together. The sensor signal will not jump in this case because it has always detects the presence of steel balls. So the microprocessor will regard the two or more steel balls pressed together as only one, which will cause the count number to be less the actual.

Figure 2.

Signal waveform when 10 steel balls passing through the pipe.

Figure 2 shows the signal waveform when 10 steel balls are dumped into the pipe channel together. As can be seen from the figure, the pulse width of the low level of the 6th and 7th, steel balls is twice the width of the single steel ball passing through the sensor because they are squeezed together. According to the previous pulse counting method, the number counted should be 9, which does not match with the actual number 10. Therefore, if the steel balls are poured in batches, when two or more steel balls are close, the sensor output low level signal is connected, which belongs to signal overlap and will make it impossible for the microprocessor to distinguish each ball.

3. The relationship between the sensor low voltage duration and the installation pipe angle

According to the knowledge of mechanics, the larger the inclination angle of the pipe, the larger the gravity component and acceleration of the ball along the pipe, and the shorter the time of ball passing through the sensor. The sensor placed in different positions of the pipe will have different output low voltage duration because of the moving distance between the entrance to sensor. The farther the moving distance, the shorter the duration.

Figure 3.

Pipe installation Angle-low voltage duration and angle-velocity relationship curve.

Figure 3a shows the relationship curve between the low voltage duration of a single steel ball passing through the sensor and its installation angle. It can be seen that, the larger the angle, the shorter the low voltage duration. The diameter of the sensor’s sensing head is 16.5 mm, and the sensing range of sensor is approximately 15 mm diameter. Therefore, the average speed of the ball passing through the sensor is $\frac{15}{t}$ m/s, where the unit of time $t$ is milliseconds, and the curve of average speed-angle can be transformed from Fig. 3a to b. From Fig. 3b, it can be seen that the relationship between the installation angle and the average passing speed could satisfy the quadratic curve. It has the formula $V=a_{1}\theta^{2}+b_{1}\theta^{2}+c_{1}$ , where the $a_{1}$ , $b_{1}$ and $c_{1}$ represent coefficient of conic. Coefficient can be obtained by the method of matrix operation, the collected point pairs $\left({V_{i},\theta_{i}}\right)$ are substituted into the matrix Eq. (1), so that the matrix equation is expressed as $V=X\cdot A$ ï¼Œwhere $Z$ represents the velocity matrix, $X$ represents the angle matrix, $A$ represents the coefficient matrix, then there is $A=(X^{T}X)^{-1}X^{T}Z$

$\displaystyle\left[{{\begin{array}[]{*{20}c}{\begin{array}[]{l}V_{1}\\ V_{2}\\ \vdots\\ \end{array}}\\ {V_{3}}\\ \end{array}}}\right]=\left[{{\begin{array}[]{ccc}1&{\theta_{1}}&{\theta_{1}^{2% }}\\ 1&{\theta_{2}}&{\theta_{2}^{2}}\\ \vdots&\vdots&\vdots\\ 1&{\theta_{i}}&{\theta_{i}^{2}}\\ \end{array}}}\right]\left[{{\begin{array}[]{*{20}c}{c_{1}}\\ {b_{1}}\\ {a_{1}}\\ \end{array}}}\right]$ (1)

After calculating the curve coefficient, the output low voltage duration of the sensor corresponding to each angle can be expressed as:

$\displaystyle t=\frac{15}{a_{1}\theta^{2}+b_{1}\theta+c_{1}}$ (2)

So with the installation distance determined, if the installation angle information is obtained, the corresponding standard time for a single steel ball to pass the sensor can be found by Eq. (2).

Therefore, when the distance between the installation position of the sensor and the entrance position is fixed, if the installation angle information is obtained, the corresponding standard time of a single steel ball passing through the sensor can be obtained through Eq. (2).

4. Precise counting method

4.1 Single sensor standard self-correcting method

It can be seen from Fig. 4 that when Nos 6 and 7 steel balls are overlapped, the output low-voltage duration is twice that of other non-overlapped steel balls. Therefore, the standard output duration can be compared with the actual duration to do the correction. When the pulse width of low voltage is more than 0.5 times of standard duration, the count value will be increased by 1. According to this principle, a single sensor with self-correcting counting method can be used for accurate counting. The recorded value plus 1 for each pulse, and at the same time, microprocessor measure the sensor low-voltage duration when the ball passing through the sensor and record it each time. The recorded low-voltage duration is compared with the standard one, and the minimum value among the recorded duration is taken as the standard value, which is used as a reference for correction.

This method uses a single sensor combined with a microcontroller for counting and low-voltage duration measurement, which is simple and effective. However, there are various limit cases when pouring steel balls in a batch, such as 2 or more balls close together at the beginning, multiple balls close together all the time, etc. If the balls are not next to each other during the whole process, the minimum low level duration as the correction base will result in a large deviation in counting; in addition, if the balls are not next to each other later on, the counting before that is inaccurate and cannot achieve the effect of real-time accurate counting.

As the counting and conveying pipeline has a certain installation angle, the output low level time when a single steel ball passes the sensor at different installation angles is different. Another method is to measure the low voltage duration corresponding to multiple angles in advance and find out the corresponding function equation of angle - standard duration time. When the installation angle changes, the low level standard time is calculated by the function equation according to the angle, and the standard pass time is set manually by software, and the time is used for counting correction.

Figure 4.

The signal of two inductor sensor at the timer start and end moment.

Although this method is simple and efficient, it is tedious to set it once for each change of installation angle and requires the professional engineer to do this job.

4.2 Dual sensor standard time comparison correction method

According to the above analysis, when the single sensor standard time comparison correction method (fully automatic) is selected, there is a high probability of counting error because the moment of the standard pulse appears is uncertain, and it may not even appear; and it is more tedious when the setting is done manually. So if a method can be found to obtain the exact standard pass time at the moment of counting start, then accurate auto-counting can be achieved.

After the signal test, it is found that if two sensors are installed on the pipe through which the steel ball passes, when the distance between the sensors exceeds 60 mm, no matter whether they are squeezed together at the beginning, there is a jump edge at the initial moment of passing through the first and second sensors. Through further observation, it is found that when the pipe angle and the sensor position are fixed, the duration between the two jump edges is also determined.

The microcontroller measurement uses the circumferential method. Firstly, when the steel ball passes sensor A, at this time the microcontroller input pin gets a low level signal from sensor A to start the micro controller timer $T_{2}$ for timing. Secondly, when it passes sensor B, the other input pin of the microcontroller also gets a low level signal from sensor B at this time the timer $T_{2}$ will be turned off. Take out the timer value and then multiply it by the equivalent time. The time between the steel ball passing through two sensors could be gotten and its average speed will be calculated after that. The signal of two inductor sensor at the start and end moment is shown in Fig. 4.

The microcontroller measurement uses the circumferential method. Firstly, when the steel ball passes sensor A, at this time the microcontroller input pin gets a low level signal to start the micro controller timer $T_{2}$ for timing. Secondly, when it passes sensor B, the other input pin of the microcontroller also gets a low level, at this time the timer $T_{2}$ will be turned off. Take out the timer value and then multiply it by the equivalent time, we can get the time between the steel ball passing through two sensors and get its average speed.

Figure 5 shows the average speed-angle relationship curve collected by the microcontroller when the two sensors are installed at a distance L of 300, 265, and 200, respectively. As can be seen from the figure, the relationship between the average velocity and angle satisfies the quadratic curve relationship $V_{2}=a_{2}\theta^{2}+b_{2}\theta^{2}+c_{2}$ . The greater the distance between the two sensors, the more adequate the acceleration, the greater the average velocity. Therefore, as long as sufficient point pair information is collected, the curve coefficients $a_{2}$ , $b_{2}$ and $c_{2}$ can be found. In the case of known curve coefficients, once the duration of the ball passing through two sensors is obtained, the installation angle information of the pipe can be inverse derived.

Figure 5.

Average speed-angle curve at $L=$ 300,265,200.

According to the previous analysis, the duration of dual sensor detection is not affected by whether the steel balls overlap or not because it used two voltage Jump edge to judge the start and end of timer. The first detected duration passing through the two sensors is always determined, and there is a conic relationship between it and the installation angle. Therefore, this characteristic can be used for self correcting.

The basic principle of dual sensor self-correcting method is shown in Fig. 6. In the first step, the low voltage duration of a single steel ball passing through sensor A at different pipe angles is collected, and then the function of time and angle $t=f(\theta)$ is obtained by quadratic curve fitting method. The second step is to collect the duration of a single steel ball passing through the dual sensors under different pipe angles, and obtain the function relationship $\theta=f(t)$ between the angle and duration collected by the dual sensors through the quadratic curve fitting method [8, 9]. The third step is to obtain the sensor self-correcting parameters. At the beginning of counting, collect the duration of the steel ball passing through the dual sensors for the first time, and obtain the pipe angle $\theta$ from the input function $\theta=f(t)$ ; then the pipe angle $\theta$ is substituted into the function to calculate the standard duration of a single steel ball passing through the sensor, which is used to correct the low-voltage duration of the steel ball passing through the sensor A. If the difference between duration and standard duration is small, it is considered that the steel balls do not overlap, and the count is added by 1. If it is n times, the count is added by $n$ .

Figure 6.

Basic principle diagram of dual sensor self correcting method.

The closer the sensor is to the entrance, the longer the steel ball passes through the sensor, and the longer the passing time is conducive to the measurement of low voltage duration by MCU, and the error caused by correction calculation is relatively small. Therefore, sensor A is used as the main counting sensor, because it is close to the entrance. Its signal is connected to the interrupt pin of MCU and the input counter pin of timer T1. MCU collects main sensor pulse signal to count, and at the same time, it measures and records the low-voltage duration, and the recorded data is used to correct the count value. Sensor B is used as the auxiliary sensor. At the beginning of counting, the duration of ball passing through sensor A and B is acquired [10, 11, 12, 13, 14, 15]. The pipe angle is gotten by the formula $\theta=f(t)$ , and then the standard low-voltage duration of a single steel ball passing through sensor A is gotten by the set function $t=f(\theta)$ . This standard duration is used for correcting the ball number.

Figure 7.

Flow chart of steel ball counting with self correcting.

Figure 8.

Physical diagram of the steel ball counting test device.

Figure 7 shows the overall detection process of the steel ball counting detection device. When the microcontroller starts to work, the initialization operation is carried out first, and the microcontroller interrupts to detect whether there is a pulse signal at the access port of sensor A. If there is a pulse signal, add 1 to the count value. At the same time, start the timer T1 to measure the low-level duration of the pulse and record it [6, 8]. and then judge whether the correction job is completed (judge whether the correction flag bit is 1). If not, start the timer T2 At the start time, when the sensor B obtains the low-level signal, stop the timer T2, extract the counter data of the timer T2. This counter data is used to get the standard duration single ball passing though sensor A by the method mentioned above. After that, set correction flag bit to 1.When the correction flag bit is 1, the previous recorded time is compared one by one with the standard duration time, and the compared result is used to correct the ball number. The correcting ball number will be displayed on the screen after correcting job is done.

5. Analysis of experimental results

Figure 9 shows the physical diagram of the steel ball counting test device. The device selects the STC15F2K60S2 type microcontroller with core 51 as the main controller and the 12864 display chip of Nuomeng Optoelectronics. Under different pipe angles, 50 steel balls are poured into the device for 100 times repeated measurement. Three methods are used in this experiment, which are named single sensor pulse counting method, single sensor with self-correcting method and dual sensor with self-correcting method, and the results are shown in Fig. 9.

Figure 9.

Three methods to measure the probability of error graph.

As can be seen from Fig. 10, for the first two methods, at a lower angle, the kinetic energy generated by collision is relatively small, and the mutual extrusion exists, but it is less at a larger angle, so the error rate is small. while between 20 degrees and 70 degrees, with the increase of angle, the extrusion basically increases with the angle increasing, and the error rate increases. On the whole, the measurement effect of single sensor with self correcting method is better than that of single sensor pulse counting method. The double sensor with self correcting method is best. It does not make mistakes when the pipe angle is less than 80 degrees. This is mainly due to the more impact and collision caused by the steeper angle, and the almost free falling motion of the steel ball leads to the larger kinetic energy, unstable speed and fluctuation of the sensor signal. Therefore, the installation pipe angle of more than 80 degrees should be avoided in practical application. The experimental results show that the ball counting result is accurate and reliable by using the dual sensor calibration method, which effectively solves the problem of miscount in batch measurement of steel balls. Compared with the method mentioned in reference [1, 2], it is more reliable and accurate because the image counting method is affected by image interference of light source and anti-rust oil. Another advantage is that it’s cheap.

Figure 10.

Sketch map of bearing assembly automatic production line.

This counting method is applied to count steel balls on bearing assembly automatic production line. Figure 10 is sketch map of bearing assembly automatic production line. The steel ball is poured into the ball container by Steel ball dumping device firstly. Next it would contact the baffle plate and then slide to the bottom of the container. This step can greatly reduce the kinetic energy of the ball. After that, steel balls at the bottom will poured into the pipe. Two sensors are set outside the pipe. Counting device is used to count the steel balls in real time using the dual sensor self-correcting method. Finally, the ball numbers are transferred to the computer by the counting device. The computer control the Steel ball dumping device to add the steel balls into the container according the ball numbers and production needs.

6. Summary

When steel balls are sent to the PVC pipe at the same time, some balls will connect together because of the collision. The sensor will send a connected signal instead of multiple signals if it meets the connected balls. This paper analyses the causes of signal overlap when a large number of steel balls enter the transfer pipe, studies the relationship between the pipe angle and the average speed of the steel ball passing through the sensor, compares the advantages and disadvantages of pulse measurement method and single sensor self correcting method, and designs steel ball precise counting device.

Using the double sensor self correcting method, The device can effectively solve the problems of miscounting caused by signal overlapping when batch steel balls are poured into the transfer pipe. It can omit the tedious manual operation steps. It is successfully applied to count steel balls on bearing assembly automatic production line and get good result. After proper improvement, this counting device can also be extended to other similar products.

Footnotes

Acknowledgments

The work is supported by the Industrial Robot Technology Collaborative Innovation Centre project of Qing Yuan Polytechnicã€Innovative natural science projects of Guangdong Education Department and Intelligent measurement and control technology research team project provided by the Guangdong Education Department. We also acknowledge Student Li Qing and Li Jianping from Qingyuan Polytechnic for setting up the experimental device and doing the test.

References

Wang

Z.S.

Y.Q.

and Ren

, A method of steel balls accurate counting based on image recognition technology, Journal of Changchun University of Science and Technology (Natural Science Edition) 38(3) (2015), 94–96.

Y.Q.

S.X.

Zhao

J.X.

et al., Accurate count and size recognition system of steel balls based on image technology, Journal of Jilin University (Information Science Edition) 37(6) (2019), 631–637.

S.X.

, Research on the counting and size recognition system of steel balls based on image processing, Ph.D. Dissertation, Changchun University of Science and Technology, 2020.

Cui

Chu

Sun

, et al., Simulation and experiments of an intelligent steel ball counter, International Journal of Simulation 33(17) (2016), 1–6ï¼Ž

Zhan

Yang

S.P.

Deng

Y.M.

et al., Automatic counting and arranging device for steel balls [P]. Yunnan: CN201737482U, 2011-02-09.

Zhang

D.D.

, Counting device for steel ball automatic packing system. Chinese Patent, ZL 201420657444.8. 2014.

Y.L.

, Application of optical counting device in the experimental setup of single pendulum, Physics Bulletin 5 (2013), 80–84.

Y.Y.

, Numerical calculation methods, Coal Industry Press, Beijing, 2014, pp. 321–323.

Shi

H.M.

and Yu

Z.J.

, Track gauge measurement method based on least-square curve fitting theory, Journal of The China Railway Society 41(12) (2019), 81–87.

10.

Feng

Y.M.

X.L.

Zhao

S.P.

, Design of high precision testing frequency counting and timing system based on C8051F020, Electronic Measurement Technology 37(4) (2014), 72–75.

11.

C.A.

, Measure for period based on microcontroller, Electronic Technology 12(7) (2015), 19–20.

12.

Ren

H.A.

and Cai

Y.K.

, Pulse signal generation and testing integrated device based on FPGA, Internet of Things Technology 10(10) (2020), 18–21.

13.

W.N.

Luo

X.F.

et al., Design and analysis of hall type steel ball detecting and counting device, Mechanical and Chemical Engineering 26 (2017), 143–144.

14.

Feng

Y.M.

X.L.

and Zhao

S.P.

, Design of high precision testing frequency counting and timing system based on C8051F020, Electronic Measurement Technology 37(4) (2014), 72–75.

15.

Chai

Q.Z.

Chen

D.X.

et al., Design of monitoring and counting system for bee colony based on ultralow-power consumption MCU and photoelectric sensor, Transactions of the Chinese Society of Agricultural Engineering 33(13) (2017), 193–198.