SLUUDN8 July   2026 TMS320F28P550SJ , TMS320F28P650DK

 

  1.   1
  2.   Abstract
  3.   Trademarks
  4. 1Introduction
  5. 2Library Notice
  6. 3Library Architecture
    1. 3.1 Background Processing
    2. 3.2 Foreground Processing
    3. 3.3 Data Transfer Mechanism
  7. 4Metrology Calculations
  8. 5Configuration and Data Structures
  9. 6Calibration Module
  10. 7User Specific Configurations
    1. 7.1 Debug Mode
      1. 7.1.1 The Function of Debug Mode
      2. 7.1.2 How to Configure DEBUG Mode Parameters
    2. 7.2 ADC Mode
      1. 7.2.1 Using the On-Chip ADC
      2. 7.2.2 Two Point Calibration
        1. 7.2.2.1 Step 1 – Offset Calibration (Zero-Input Point)
        2. 7.2.2.2 Step 2 – Gain Calibration (Full-Scale Point)
      3. 7.2.3 Using an External ADC
    3. 7.3 Switching Between Modes
    4. 7.4 THD Algorithm Selection
      1. 7.4.1 Method 1 – Based on PLL (Default When the HARMONICS_SUPPORT Macro is Not Defined)
      2. 7.4.2 Method 2 – Goertzel DFT (When the HARMONICS_SUPPORT and USE_GOERTZEL_THD Macros are Both Defined)
      3. 7.4.3 Method 3 – 2048-Point Real FFT (When the HARMONICS_SUPPORT Macro is Defined But the USE_GOERTZEL_THD Macro is Not)
    5. 7.5 Topology Selection
    6. 7.6 Enable Features
    7. 7.7 System Parameters
    8. 7.8 Scaling Factors Per Phase
  11. 8Running the Example
    1. 8.1 Accessing the Parameters Per Phase
    2. 8.2 Accessing the Total System Measurements
    3. 8.3 Accessing the Phase Status Flags

7.4.3 Method 3 – 2048-Point Real FFT (When the HARMONICS_SUPPORT Macro is Defined But the USE_GOERTZEL_THD Macro is Not)

Method three uses a full 2048-point real FFT with a Hann window applied before the transform. The FFT window spans approximately 12.8 cycles at 50Hz / 8000Hz (2048 / 8000 × 50 = 12.8). After the FFT, harmonics are grouped according to the IEC 61000-4-7 standard (5Hz bins centered on each harmonic) and THD is computed from the grouped harmonic bins. The results are returned in the vTHDfft and iTHDfft fields.

This method provides full spectral information but requires significantly more RAM (two 2048-element float arrays for the real and imaginary FFT buffers) and longer settling time than the Goertzel DFT method. Use method three only if full spectrum analysis beyond THD is required.