node-red-contrib-vibration-ism330dhcx

A Node-RED node for vibration sensing using the STMicroelectronics ISM330DHCX 6-axis IMU over I2C. Continuously samples the accelerometer, computes time- and frequency-domain vibration metrics on fixed-size windows and emits both per-window values for live charting and an aggregated record for long-term storage.

Installation

cd ~/.node-red
npm install @thehighestbit/node-red-contrib-vibration-ism330dhcx

Configuration

| Property | Description | |----------|-------------| | Data Rate | Accelerometer output data rate (12.5 Hz – 6.66 KHz). Higher rates resolve higher vibration frequencies but produce more data. | | Measurement Range | Full-scale acceleration range (±2g, ±4g, ±8g, ±16g). Pick the smallest range that still covers your expected peaks. | | High Pass Filter | Optional accelerometer high-pass filter to remove DC / gravity bias (None, Slope, ODR/10 … ODR/800). | | Window Size | Number of samples per analysis window (128 – 65536). Larger windows give finer frequency resolution but slower update rate. | | Output Interval | How often the aggregated metrics record (output 1) is emitted, in ms. Defaults to 30000. |

The gyroscope is intentionally powered down — only the accelerometer is used.

Outputs

The node has 5 outputs. Output 1 is intended for database storage; outputs 2–5 are intended for live charts (e.g. Node-RED Dashboard 2.0). All amplitude values (mean, RMS, peak, top-peak magnitudes) are in m/s².

1. Aggregated metrics

Emitted every Output Interval ms. Aggregates all per-window metrics computed since the last emission.

{
    "payload": {
        "ts_start": 1714400000000,
        "ts_end": 1714400030000,
        "window_size": 256,
        "avg_mean": 9.83,
        "max_peak": 24.33,
        "overall_rms": 10.14,
        "avg_crest_factor": 2.41,
        "avg_kurtosis": 3.12,
        "dominant_freq": 124.5,
        "top_peaks": [
            { "f": 124.5, "m": 18.2 },
            { "f": 249.0, "m": 7.4 }
        ]
    }
}

dominant_freq and top_peaks are taken from the highest-RMS window in the interval (a "worst-case spectral snapshot"), since averaging FFT peaks across windows is unreliable when frequencies drift.

2. RMS / Peak (m/s²)

Emitted once per analysis window as two messages with topics RMS and Peak.

3. Crest Factor

Emitted once per analysis window with topic Crest Factor. Ratio of peak to RMS — useful for detecting impulsive faults.

4. Kurtosis

Emitted once per analysis window with topic Kurtosis. Fourth-moment statistic — rises sharply when the signal contains transients.

5. Dominant Frequency (Hz)

Emitted once per analysis window with topic Dominant Frequency. Frequency of the largest FFT magnitude bin (DC removed, Hann-windowed, zero-padded to next power of two).

How it works

1. The accelerometer is configured with the chosen rate, range and (optional) high-pass filter, and put into FIFO continuous mode.
2. Every 10 ms the FIFO is drained and the magnitude √(x² + y² + z²) of each sample is appended to a rolling buffer.
3. Once the buffer reaches Window Size samples, that window is consumed and metrics are computed: mean, RMS, peak, crest factor, kurtosis, and an FFT (DC-removed, Hann-windowed, zero-padded). The five strongest spectral peaks are picked greedily with a minimum separation of three FFT bins so a single true peak cannot fill multiple slots.
4. Per-window metrics go out on outputs 2–5 immediately and are also stashed for aggregation.
5. Every Output Interval ms, the stashed metrics are aggregated and emitted on output 1.