FFT Analysis for Weighing – How to Use the Fourier Transform to Identify Mechanical Noise
Introduction: The Unseen Enemy of Weighing Accuracy
The majority of errors and instability issues in industrial weighing systems are not electronic; they are mechanical, manifesting as vibration and noise. These disturbances are invisible in a simple weight display, which typically shows only the filtered, fluctuating final value. To diagnose these complex issues effectively, high-level weighing technicians employ Frequency Domain Analysis using the Fast Fourier Transform (FFT). FFT converts the weight signal from the standard Time Domain (weight vs. time) into the Frequency Domain (amplitude vs. frequency), allowing for the precise identification of noise sources based on their rhythmic signature.
The Role of the Fast Fourier Transform (FFT)
The FFT is an efficient algorithm that computes the Discrete Fourier Transform (DFT). Its purpose in weighing is to break down a complex, unstable weight signal into its constituent sine waves, revealing the amplitude (severity) and frequency (speed) of every vibration component affecting the scale.
FFT Output Interpretation
- X-Axis (Frequency): Represents the speed of the vibration, measured in Hertz (Hz). This is the key to identifying the physical source (e.g., a pump, fan, or conveyor).
- Y-Axis (Amplitude): Represents the magnitude or severity of the vibration at that specific frequency. A tall peak indicates a strong noise source severely impacting the scale's stability.
In the Frequency Domain, instability is not seen as a fluctuating weight number, but as a specific, identifiable peak on the FFT graph, allowing the engineer to diagnose the physical root cause.
Diagnosing Common Mechanical Noise Sources
By comparing the peaks on the FFT graph to known industrial equipment operating speeds, the technician can pinpoint the source of the noise. This transforms a guessing game into a precise diagnostic exercise.
| FFT Peak Frequency Range | Likely Mechanical Source | Mitigation Technique |
|---|---|---|
| 1 Hz to 5 Hz | Structural Resonance or Natural Frequency of the Weighing Structure (Tank or Platform). | Requires structural damping or adjusting the scale's digital filter cut-off frequency. |
| 10 Hz to 60 Hz | Pumps, Compressors, or Electrical Noise (50 Hz / 60 Hz line frequency). | Requires structural isolation (isolator pads) or verification of ground loops. |
| 80 Hz to 200 Hz | Conveyors, Agitator Motors, or Gearboxes. | Requires balancing the rotating equipment or mechanical separation from the scale. |
Engineering Requirements for Accurate FFT Analysis
Performing a meaningful FFT analysis requires specialized hardware and procedural rigor that distinguishes it from basic signal filtering:
1. High Sampling Rate (ADC Speed)
The weighing indicator must have an Analog-to-Digital Converter (ADC) capable of sampling at least twice the frequency of the highest noise component to be measured (Nyquist-Shannon Sampling Theorem). For example, to accurately diagnose a 100 Hz vibration, the indicator must sample at 200 Hz or higher. Advanced indicators sample at 500 Hz to 1000 Hz for this exact purpose.
2. Data Capture and Windowing
The data must be captured over a stable period, and the analysis must use an appropriate Window Function (e.g., Hanning, Hamming) to minimize spectral leakage, which can otherwise cause energy from a sharp peak to "leak" across the frequency spectrum, obscuring smaller noise components.
3. Digital Filter Design
Once the noise frequencies are identified via FFT, the engineer can precisely tune the indicator's digital bandpass or low-pass filter to reject only those specific noise frequencies. This is far superior to using a strong, generic filter, which significantly increases Residence Time and reduces throughput.
FFT analysis is a high-level diagnostic tool that moves weighing system optimization beyond trial-and-error, enabling the engineer to target and eliminate the exact frequencies of interference for maximum accuracy and speed.
