VibrationDAQ

Data Acquisition (DAQ) of the Analog Devices ADcmXL3021 vibration sensor. The VibrationDAQ was developed as part of an internship at Wingtra by Jonas Lauener.

For any technical questions, please open an issue: https://github.com/jolau/VibrationDAQ/issues/new/choose

Copyright (c) 2020, Jonas Lauener & Wingtra AG

Hardware

• Raspberry Pi
• Analog Devices ADcmXL3021 vibration sensor capabilities
  • Corrected data sheet
  • Sensor orientation:
    
  • Mass: 13 g
  • Frequencies up to 110 kHz
  • ±50 g measurement range
  • MFFT mode: Spectral analysis through internal FFT
    
    • 2048 bins per axis with user configurable bin sizes from 0.42 Hz to 53.7 Hz
    • Windowing options: rectangular, Hanning, flat top
  • MTC mode: Time domain capture
    
    • 4096 samples per axis
• Features a status LED

Pinout Raspberry Pi

	pin	pin
RST 1	13
BSY 1	15	16	RST 2
3V3	17	18	BSY 2
MOSI	19	20	EMPTY
MISO	21	22	SYNC
SCLK	23	24	CE 1
GND	25	26	CE 2

The vibration sensor only works on the main SPI pins (SPI0) of the Raspberry Pi - the other SPI ports don't support SPI mode 3.

Custom adapter board

The pins of the ADcmXL3021 vibration sensor can be accessed by using a custom adapter board. The PCB was created by using the software KiCad: KiCad project files

Installation

In case you want to setup your own VibrationDAQ on a Raspberry Pi:

1. Install yaml-cpp (as described below)
2. Clone this repo.
3. cd VibrationDAQ
4. mkdir build & cd build
5. cmake ..
6. sudo make install

Make sure that you have enabled SPI.

You can use it now with vibration_daq_app [full path to config.yaml].

Installation of yaml-cpp

1. wget https://github.com/jbeder/yaml-cpp/archive/yaml-cpp-0.6.3.tar.gz
2. tar -xvf yaml-cpp-0.6.3.tar.gz
3. cd yaml-cpp-yaml-cpp-0.6.3/
4. mkdir build
5. cd build
6. cmake .. -DYAML_BUILD_SHARED_LIBS=ON
7. sudo make install

Enable/Disable auto-start

1. Edit config.yaml path in vibration_daq.service file to own needs
2. sudo systemctl enable [full path to vibration_daq.service file]
   This automatically links the service and enables it. Replace enable with disable to disable service.

Dependencies

Many thanks to the authors of the following libraries:

• yaml-cpp
• loguru
• c-periphery
• date

Usage

Workflow

1. Connect to the Raspberry Pi and adapt the ~/Documents/config.yaml to your requirements. Make sure that auto-start is enabled. If it's the first time you use it, test it by running it manually: vibration_daq_app [full path to config.yaml]
2. Mount the vibration sensor with the double sided tape 3M™ Adhesive Transfer Tape 950. This shouldn't distort the vibration too much.
3. Do your measurement.
4. Download the collected data over SFTP. I recommend to also download the used config file.
5. Open a vibration CSV file in Google Sheets.
   • For FFT measurement:
     • Hide the first two data points as these have usually very high magnitude and don't give meaningful information
     • Plot the data using a column chart.

Status led

If the status led is enabled in config, it will glow when running:

• Constant glow: data acquisition is running normally
• Blink: measurement completed, saving data to a new csv file

Config file

Short primer on the syntax of yaml: https://learnxinyminutes.com/docs/yaml/

Decimation Filter

Setting the right decimation_factor is crucial for getting meaningful data. Choose it according to the maximum vibration frequency you're expecting. Of course, the decimation factor also affects the predefined FIR filters i.e. with a FACTOR_64, the LOW_PASS_10kHz is actually a 156 Hz low pass filter.

decimation_factor	Effective Sample Rate, fS (SPS)	Effective FFT Bin Size, f_MIN (Hz)	Effective Maximum FFT Frequency, f_MAX (Hz)
FACTOR_1	220000	53.71094	110000
FACTOR_2	110000	26.85547	55000
FACTOR_4	55000	13.42773	27500
FACTOR_8	27500	6.713867	13750
FACTOR_16	13750	3.356934	6875
FACTOR_32	6875	1.678467	3437.5
FACTOR_64	3437.5	0.839233	1718.75
FACTOR_128	1718.75	0.419617	859.375

Spectral average count

The spectral_avg_count determines the number of FFT records that the ADcmXL3021 averages when generating the final FFT result. Up to 255 records. Good for getting FFT measurements over longer time periods.

Calculate measurement duration

FFT

4096 samples / (220'000 sample rate / decimation_factor) * spectral_avg_count = record time [s]

Example with FACTOR_64 and 255 spectral averages:

4096 samples / (220000 sample rate / 64) * 255 = 303.84 seconds = 5.06 minutes

MTC

4096 samples / (220'000 sample rate / decimation_factor) = record time [s]

Example config with explanation

storage_directory: "/home/pi/Documents/"
recordings_count: 2 #number of recurring measurements, infinite if == 0 
external_trigger: false # false: triggering over SPI; 
                        # true: triggering over dedicated pin, useful for triggering multiple sensor at exact same time (connect them to same pin)
external_trigger_pin: 4 # only read if external_trigger == true
status_led: true # enable/disable status led, blinks everytime a vibration file is written
status_led_pin: 21  # only read if status_led == true
sensors:
  - name: sensor1 #will be used for logging and filenames
    busy_pin: 22 #BCM pin number
    reset_pin: 27 #BCM pin number
    spi_path: "/dev/spidev0.0"
    recording_mode: MFFT # MTC and MFFT supported
    MFFT_config: &mfftConfig #only read if recording_mode == MFFT
      decimation_factor: FACTOR_2 #supported: [FACTOR_1 = 0, FACTOR_2 = 1, FACTOR_4 = 2, FACTOR_8 = 3, FACTOR_16 = 4, FACTOR_32 = 5, FACTOR_64 = 6, FACTOR_128 = 7]
      fir_filter: CUSTOM #supported: [NO_FILTER, LOW_PASS_1kHz, LOW_PASS_5kHz, LOW_PASS_10kHz, HIGH_PASS_1kHz, HIGH_PASS_5kHz, HIGH_PASS_10kHz, CUSTOM]
      custom_filter_taps: [6, 21, 53, 107, 193, 316, 480, 686, 930, 1203, 1490, 1774, 2034, 2251, 2407, 2489, 2489, 2407, 2251, 2034, 1774, 1490, 1203, 930, 686, 480, 316, 193, 107, 53, 21, 6]
      spectral_avg_count: 2 # value between 1-255
      window_setting: HANNING #supported: [RECTANGULAR, HANNING, FLAT_TOP]
    MTC_config: #only read if recording_mode == MTC
        decimation_factor: FACTOR_2
        fir_filter: CUSTOM
        custom_filter_taps: [6, 21, 53, 107, 193, 316, 480, 686, 930, 1203, 1490, 1774, 2034, 2251, 2407, 2489, 2489, 2407, 2251, 2034, 1774, 1490, 1203, 930, 686, 480, 316, 193, 107, 53, 21, 6]
  - name: sensor2
    busy_pin: 24
    reset_pin: 23
    spi_path: "/dev/spidev0.1"
    recording_mode: MFFT
    MFFT_config: *mfftConfig #copy config from sensor1 above

Example data

The following data was collected on a self-made vibration bench. The bench consists of an unbalanced mass attached to an electrical motor.

• MFFT raw data example
  • Plot of MFFT data:
    
• MTC raw data example