This measurement column addresses frequently asked questions received by our customer support center and provides answers to those questions.
Many people who use an FFT analyzer to perform frequency analysis have probably wondered where to start and what the procedure should be.
This time, we will introduce the setup procedure for an FFT analyzer.
Measurement process using an FFT analyzer
The most common measurement process is as follows:
- Connect the sensor and configure the sensor input settings.
- Configure various measurement settings to ensure that the input waveform is measured correctly.
- This process discretizes analog signals so that they can be correctly treated as digital signals.
- Divide the data into short time intervals (analysis frames).
- Perform an FFT operation and plot the power spectrum.
- Perform the averaging process and save the analysis results.
Explanation in each part
1. Connect the sensor and configure the sensor input settings.
One of the settings for the FFT analyzer here is CCLD (Constant Current Drive). In the case of microphones or Accelerometer with built-in preamplifiers, this setting supplies current from the FFT analyzer to drive the sensor. This allows the sensor to output the correct voltage signal.
For detailed information on CCLD settings, please refer to the measurement column below.
Engineering Units (EU) and Unit Calibration - Part 1: The Case of Accelerometers -
2. Configure various measurement settings to ensure accurate waveform measurement.
To correctly monitor the output from the sensor, i.e., the input signal to the FFT analyzer, you need to set the input voltage range. Additionally, to treat this input signal as a sound or vibration signal, you need to configure the unit conversion. This is called unit calibration.
For details on unit calibration, please refer to the measurement column below.
Engineering Units (EU) and Unit Calibration - Part 5: Microphones and Sound Level Meters, Part 1 -
Engineering Units (EU) and Unit Calibration - Part 6: Microphones and Sound Level Meters, Part 2 -
3. Discretize analog signals so that they can be correctly handled as digital signals.
Signal processing within an FFT analyzer is performed digitally. Therefore, it is necessary to digitize the analog input signal, a process called discretization (sampling). Specifically, sampling along the time axis is called sampling, and sampling along the amplitude axis is called quantization. When sampling an analog signal, sampling is performed based on a law called the "sampling theorem." The setting of the FFT analyzer here is the frequency range.
For details on sampling, please refer to the measurement column below.
Frequency Analysis from the Basics (7) - "Sampling of Time Signals"
4. Divide into short time intervals (analysis frames).
To perform an FFT calculation using an FFT analyzer, we can only realistically handle a finite amount of data. Therefore, the discretized digital data from the previous step is divided into a finite number of points (also called cutting with a time window) and calculated. However, because the data is cut at arbitrary points, an error called leakage error occurs when performing the FFT calculation. To reduce this error, the beginning and end of the data cut with the time window are weighted to smooth them out (set to 0). This is called a window function. There are various types of time windows, but for continuous sounds or vibrations, the Hanning window is generally used. The settings for the FFT analyzer here are the number of sample points and the window function.
For details on window functions, please refer to the measurement column below.
Frequency Analysis from the Basics (14) - "DFT (FFT) and Time Windows"
5. Perform FFT analysis and plot the power spectrum.
An FFT analyzer allows you to plot various FFT analysis graphs. Generally, the power spectrum graph is used to check the magnitude of each frequency component.
For detailed information on the power spectrum, please refer to the measurement column below.
Fundamentals of Digital Measurement - Part 9: "Various Time Waveforms and Spectra"
6. Perform the averaging process and save the results as analysis results.
Since the power spectrum of each analysis frame changes moment by moment, it is common practice to use a power spectrum that has been averaged over a certain analysis time or number of analysis frames as the analysis result.
For details on the averaging process, please refer to the measurement column below.
Frequency Analysis from the Basics (15) - "Power Spectrum (Part 3)"
summary
This time, we introduced the general procedure for frequency analysis using an FFT analyzer. FFT analyzers have various settings for measurements, but it is important to organize the measurement flow in a systematic way, such as ensuring that the connected sensor is driven, converting voltage units to sound units, and converting analog signals to digital signals. Then, make sure to set the appropriate settings for each step on the FFT analyzer.
(Excerpt from the email newsletter issued on January 20, 2021)
