Handheld Vibration Analyzer for Fault Prediction

Handheld Vibration Analyzer for Fault Prediction

Prasad Gaikwad (a), Khushi Naikwadi (a), Mayuresh Lokhande (a), Rutuja Babar (a), Prof. A. R. Nigavekar (b).

(a) UG student, Department of Electronics & Telecommunication Engineering, KITS College of Engineering, Kolhapur, Maharashtra, India

(b) Professor, Department of Electronics & Telecommunication Engineering, KITS College of Engineering, Kolhapur, Maharashtra, India

Abstract Predictive maintenance has become a critical requirement in modern industries to minimize downtime and prevent catastrophic machine failures. Conventional vibration analyzers, while effective, are often bulky and expensive, limiting their accessibility for field engineers and small-scale industries.

This paper presents the design and development of a handheld fault-predicting vibration analyzer built around the Shrike development board and a piezoelectric vibration sensor. The system captures vibration signals from rotating machinery, processes them through the microcontrollers high-resolution ADC, and stores data. Experimental validation demonstrates that the proposed system can reliably detect early fault conditions, making it suitable for predictive maintenance

applications in industrial environments.

Keywords: Shrike development board, vibration sensor

1. Introduction

   In todays world, in industries and automotive systems, unexpected mechanical failures often lead to costly downtime, safety risks, and reduced efficiency. By continuously monitoring vibration signals and digitising them through multiple ADC channels, the analyser provides precise insights into the health of machinery. The integration of an SD card for offline storage ensures that large volumes of data can be preserved and analysed over time, enabling engineers to identify patterns and predict faults before they escalate. This project is significant because it offers a compact, cost-effective, and portable solution that bridges the gap between real-time sensing and actionable maintenance decisions. It empowers organisations to move from reactive repairs to proactive. All the four ADC channels are configured to record a defined number of samples, ensuring accurate representation of system dynamics. The digitised data is stored on an SD card, enabling offline analysis and long-term monitoring of equipment health. By combining real-time acquisition with accessible storage, the analyser provides a reliable means of detecting early signs of mechanical faults, thereby reducing downtime and improving operational efficiency.

2. Literature Survey

   1. The literature on gearbox fault diagnosis mainly focuses on vibration-based condition monitoring, fault detection, and predictive maintenance in railway and automotive systems. Researchers have developed several methods using accelerometer sensors, embedded systems, signal processing techniques, and intelligent monitoring algorithms to detect gearbox faults at an early stage.

   2. According to the paper Gearbox Fault Diagnosis of High-Speed Railway Train by Bin Zhang, A.C.C. Tan, and Jianhui Lin, vibration analysis was used to investigate crack faults in high-speed railway gearboxes. Accelerometer sensors were mounted on the gearbox body to collect vibration and dynamic stress data during train operation. The

      researchers performed modal analysis and finite element analysis to identify resonance and stress concentration points in the gearbox structure. The study concluded that vibration monitoring effectively detected crack faults and improved gearbox reliability in railway applications.

   3. Another important study, Torque Effect on Vibration Behavior of High-Speed Train Gearbox Under Internal and External Excitations by Yue Zhou et al., investigated the effect of torque and wheelrail interaction on train gearbox vibration behavior. Multiple acceleration responses were measured from the gearbox under different torque and rotational speed conditions. Dynamic modeling and modal identification techniques were applied to analyze gearbox vibration characteristics. The study reported that torque significantly affects gear meshing vibration and gearbox modal characteristics, which is important for predictive maintenance of high-speed train gearboxes.

   4. In the paper State-Degradation-Oriented Fault Diagnosis for High-Speed Train Running Gears System by Chao Cheng et al., a state-degradation-oriented fault diagnosis system was proposed for high-speed train running gears. Multiple vibration sensors were used to collect gearbox and bearing vibration signals. The collected data was processed using Wiener state degradation modeling and multi-sensor filtering methods to identify fault progression. The proposed method successfully improved fault detection accuracy and reliability of railway transmission systems.

   5. Researchers have also focused on analyzing vibration and stress characteristics in railway gearboxes. In Vibration and Stress Response of High-Speed Train Gearboxes under Different Excitations, vibration data from high-speed train gearboxes was measured and analyzed under different excitation conditions. The study investigated vibration transmission characteristics and stress response behavior of the gearbox system. The authors concluded that vibration monitoring helps in understanding gearbox dynamic performance and detecting abnormal operating conditions effectively.

   6. In automotive applications, the paper Automotive Gearbox Fault Diagnosis Using Vibration Signal Analysis proposed a vibration-based fault diagnosis system for automobile gearboxes. Accelerometer vibration sensors were mounted on the gearbox casing to collect vibration signals under healthy and faulty conditions. Fast Fourier Transform (FFT) analysis was used to identify characteristic fault frequencies caused by gear wear, broken teeth, and shaft misalignment. The study concluded that vibration signal analysis provides effective early-stage fault detection in automotive transmission systems.

3. Proposed System

   The proposed system is a low-cost vibration-based fault prediction and condition monitoring system built around a Shrike board, a piezoelectric accelerometer. The piezoelectric sensor is mounted directly on the machine to capture vibrations and convert them into electrical signals. These sinewave signals are fed to the Shrike board, where they undergo A-to-D conversion and real-time processing. The system continuously compares the processed vibration levels against predefined threshold values to detect abnormal conditions and potential faults. All collected vibration data and detected fault information are automatically logged onto the PC for further analysis and record-keeping.

   This approach enables early fault detection, shortens unplanned machine downtime, and offers an affordable solution for predictive maintenance in industrial environments.

   PC

Shrike Board (RP2040)

1) BLOCK DIAGRAM

Peizo-electric sensor

1. Methodology

   1. Mathematical Calculations:-

      • 100 kHz per channel

      • That means:

      • 100,000 samples / second / channel

      • samples with respective time

      • 0.05 sec

      • 10,000 × 0.05 = 5,000 samples

      • 0.10 sec

      • 100,000 × 0.10 = 10,000 samples

      • 0.20 sec

      • 100,000 × 0.20 = 20,000 samples

      • Total samples (4 ADC channels)

      • You are using 4 channels (ADC0ADC3)

      • Formula:

      • Total samples = Samples per channel × Number of channels

      • 0.05 sec

      • 5,000 × 4 = 20,000 total samples

      • 0.10 sec

      • 10,000 × 4 = 40,000 total samples

      • 0.20 sec

      • 20,000 × 4 = 80,000 total samples

      • SOFTWARE PART:-

• GPIO Configuration

All inputs use internal pull-down resistors to ensure stable logic levels

ADC Inputs

• ADC Configuration

  1. ADC Channel:- ADC0 ADC1 ADC2 ADC3

  2. ADC operates in round-robin mode

  3. FIFO enabled with DMA request

  4. ADC resolution: 12-bit

  5. FIFO threshold: 1 sample

     Data stored as 16-bit unsigned values

• Memory Allocation

  Total samples are calculated as:

  Samples Rate Time Count Total Samples = Sample Rate×Acquisition Time×Channel Count Dynamic memory allocation (malloc) is used to store the acquired samples. Error handling ensures safe operation if memory allocation fails.

• Data Acquisition Flow

  1. User selects ADC channels and acquisition time

  2. User presses the START button

  3. DMA and ADC are started simultaneously

  4. ADC samples are captured into RAM

  5. DMA signals completion

  6. ADC is stopped

  7. System waits for PC command The LED indicates active acquisition status.

• Algorithm: RP2040 ADC Data Acquisition System

  Step 1 Start the system Initialize USB standard I/O Delay for USB enumeration

  Step 2 Initialize GPIO pins Configure channel select pins Configure time select pins Configure START button Configure LED pin

  Step 3 Initialize ADC Enable ADC GPIO pins (ADC0ADC3) Configure ADC FIFO

  Enable DMA request

  Step 4 Read ADC channel selection pins Generate ADC channel mask Step 5 Read

  acquisition time selection pins Determine acquisition time

  Step 6 Check configuration validity At least one ADC channel selected Valid acquisition time selected

  START button pressed

  Step 7 Generate channel list Count number of selected ADC channels

  Step 8 Calculate ADC clock divider clkdiv = 48 MHz / (Sample Rate × Channel Count)

  Step 9 Configure ADC clock divider Enable round-robin ADC mode

  Step 10 Calculate total samples Total Samples = Sample Rate × Acquisition Time × Channel Count

  Step 11 Allocate memory for ADC buffer If allocation fails Halt system

  Step 12 Configure DMA Source = ADC FIFO Destination = ADC buffer Transfer size = 16-bit Transfer count = Total samples

  Step 13 Turn ON status LED Start DMA Start ADC conversion

  Step 14 Wait until DMA transfer completes

  Step 15 Stop ADC conversion Turn OFF status LED

  Step 16 Send acquisition complete message to PC Enter command wait mode

  Step 17 Receive USB command If command = READ Send ADC data in CSV format

  Step 18 Remain in idle state waiting for further commands

  • Algorithm: PC-Side Python Data Logging

• Step 1 Start Python program Import serial and time libraries

• Step 2 Open USB serial port Wait for communication stabilization

• Step 3 Send READ command to RP2040

• Step 4 Wait for BEGIN_CSV marker from RP2040

• Step 5 Read incoming data line by line Store CSV data

• Step 6 Detect END_CSV marker Stop data reception

• Generate timestamp-based filename

• Create CSV file Write stored data Close file

• Close serial port End program CSV File Screenshot Commands To run Program files:for

• rp2040 SDK :-

  • mkdir build

  • cd build

  • cmake

  • make -j4

  • cp <project name>.uf2 /media/$USER/RPI-RP2/

  • minicom -b 115200 -D /dev/ttyACM0 for check terminal data for python script

  • python3 <project name>.py

1. Results and Discussion

   1. The ADC values represent the digitized signals captured from the reference of stable pot.

   2. The readings show consistent and smooth variation across samples, indicating stable data acquisition by the Shrike board .

• Conclusion:-

  In this project, a handheld vibration monitoring system was successfully developed for vibration data acquisition. The user interface allows selection of channels and data acquisition time, followed by starting the data acquisition process. The RP2040 (Dual-core ARM Cortex-M0+) was utilized as the central processing unit for high-speed data acquisition and analog-to-digital conversion.

  Overall, the project demonstrates a simple, effective, and low-cost solution for vibration analysis that can be applied in industries to monitor machine health, reduce breakdowns, and improve maintenance efficiency.

• Future Scope:-

The developed handheld vibration analyzer provides a strong foundation for predictive maintenance using vibration analysis. However, several enhancements can be implemented to make the system more advanced, intelligent, and industry-ready:

1. Wireless Connectivity and IoT Integration

   • Add Wi-Fi or Bluetooth module to enable real-time data transmission to a cloud platform or mobile application for remote monitoring.

2. Advanced Machine Learning Algorithms

   • Implement AI/ML models (such as SVM, Random Forest, or Deep Neural Networks) for automatic fault classification and more accurate fault prediction.

3. Multi-Sensor Fusion

   • Integrte additional sensors like temperature, acoustic emission, and current sensors to provide comprehensive machine health monitoring through sensor fusion techniques.

4. Real-Time FFT and Advanced Signal Processing

   • Enhance onboard processing capabilities to perform real-time Fast Fourier Transform (FFT), envelope analysis, and kurtosis calculation for better fault diagnosis.

5. Data Logging and Cloud Analytics

   • Implement long-term vibration trend analysis on the cloud to predict Remaining Useful Life (RUL) of machinery components.

6. Mobile Application Development

   • Develop a dedicated Android/iOS application for live data visualization, historical trend analysis, and instant fault alerts.

7. Miniaturization and Improved Battery Life

   • Further reduce the size and weight of the device and optimize power consumption for extended field usage.

8. Industrial Integration

   • Enable compatibility with Industry 4.0 standards for seamless integration into large-scale industrial automation and SCADA systems.

9. Edge Computing Capabilities

   • Upgrade the processing unit to support edge AI for faster on-device decision-making with minimal latency.

• REFERENCES:-

  1. Rp2040 data sheet:-https://datasheets.raspberrypi.com/rp2040/rp2040-datasheet.pdf

  2. https://www.jstage.jst.go.jp/article/sicetr1965/38/12/38_12_112 9/_article/- char/en?utm_source=chatgpt.com

  3. rp2040, pico, datasheet pip.raspberrypi.com

  4. STM32L476RG

     1. https://www.st.com/resource/en/datasheet/stm32l476je.pdf

  5. https://en.wikipedia.org/wiki/STM32

  6. https://vicharak-in.github.io/shrike/

  7. https://www.circuitstate.com/tutorials/getting-started-with-vicharak-shrike-lite-rp2040-slg47910-fpga- development-board/

  8. Hassan, I. U., Panduru, K. K., & Walsh, J. (2024). An in-depth study of vibration sensors for condition monitoring. Sensors, 24(3), 740. https://doi.org/10.3390/s24030740

  9. Uvarajan, K. P. (2024). Vibration analysis of smart structures integrated with embedded piezoelectric sensor networks: A comprehensive review. Journal of Reconfigurable Hardware Architectures and Embedded Systems, 1(1). https://fsrap.com/index.php/JRHAES/article/view/3

  10. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7146328

  11. Development of an embedded piezoelectric transducer for bearing fault detection. (2023). Mechanical Systems and Signal Processing, 188, 109987.

  12. https://doi.org/10.1016/j.ymssp.2022.109987

  13. https://www.sciencedirect.com/science/article/abs/pii/S1350630716301856?utm_source

  14. https://link.springer.com/article/10.1007/s40534-023-00323-3?utm_source

  15. https://www.mdpi.com/1424-8220/20/4/1017?utm_source

  16. https://www.mdpi.com/2076-3417/12/2/712?utm_source