Digital Filters Part 1
33 likes6,540 views
This document provides an overview of digital filters and focuses on finite impulse response (FIR) filters. It defines digital filtering and compares it to analog filtering. It describes different types of digital filters including FIR filters and explains how to design, implement and characterize FIR filters. Key aspects of FIR filters are that they have a finite impulse response, linear phase, and are always stable. Design techniques like windowing methods and Parks-McClellan optimization are covered.
Digital Filters Part 1
- 1. DIGITAL FILTERS Source: ANALOG DEVICES Part1:Finite Impulse Response (FIR) Filters
- 2. Introduction Purpose This module introduce the technology of Finite Impulse Response (FIR) Filters. Outline Introduce the technology of Digital Filters. Introduce the technology of Finite Impulse Response (FIR) Filters. Contents 32 pages Duration 20 Minutes
- 4. Digital vs. Analog Filtering
- 6. Processing Requirements Signal Bandwidth = fa Sampling Frequency fs > 2fa Sampling Period = 1 / fs Filter Computational Time + Overhead < Sampling Period Depends on Number of Taps Speed of DSP Multiplication-Accumulates (MACs) Efficiency of DSP Circular Buffering Zero Overhead Looping
- 7. Types Of Digital Filters Moving Average Finite Impulse Response (FIR) Linear Phase Easy to Design Computationally Intensive Infinite Impulse Response (IIR) Based on Classical Analog Filters Computationally Efficient Lattice Filters (Can be FIR or IIR) Adaptive Filters
- 8. 4-point Moving Average Filter
- 9. Calculating Output Of 4-point Moving Average Filter
- 10. 4-tap Moving Average Filter Step Response
- 11. Moving Average Filter Response To Noise Superimposed On Step Input
- 12. Moving Average Filter Frequency Response
- 13. N-tap Finite Impulse Response (FIR) Filter
- 15. Calculating Outputs Of 4-tap FIR Filter Using A Circular Buffer
- 16. Pseudocode For FIR Filter Program Using A DSP With Circular Buffering 1. Obtain sample from ADC (typically interrupt driven) 2. Move sample into input signal’s circular buffer 3. Update the pointer for the input signal’s circular buffer 4. Zero the accumulator 5. Implement filter (control the loop through each of the coefficients) 6. Fetch the coefficient from the coefficient’s circular buffer 7. Update the pointer for the coefficient’s circular buffer 8. Fetch the sample from the input signal’s circular buffer 9. Update the pointer for the input signal’s circular buffer 10. Multiply the coefficient by the sample 11. Add the product to the accumulator 12. Move the filtered sample to the DAC
- 17. ADSP-21XX FIR Filter Assembly Code (Single Precision)
- 18. Characteristics of FIR Filters Impulse Response has a Finite Duration (N Cycles) Linear Phase, Constant Group Delay (N Must be Odd) No Analog Equivalent Unconditionally Stable Can be Adaptive Computational Advantages when Decimating Output Easy to Understand and Design Windowed-Sinc Method Fourier Series Expansion with Windowing Frequency Sampling Using Inverse FFT - Arbitrary Frequency Response Parks-McClellan Program with Remez Exchange Algorithm
- 19. FIR Filter Impulse Response Determines The Filter Coefficients
- 20. Duality Of Time And Frequency
- 21. FIR Filter Design Using The Windowed-sinc Method
- 22. FIR Filter Design Using Fourier Series Method With Windowing Specify H(f) Expand H(f) in a Fourier series: The Fourier series coefficients are the coefficients of the FIR filter, h(m), and its Impulse Response Truncate the Impulse Response to N points (taps) Apply a suitable Window function to h(m) to smooth the effects of truncation Lacks precise control of cutoff frequency; Highly dependent on Window function
- 23. Frequency Sampling Method for FIR Filters With Arbitrary Frequency Response Specify H(k) as a Finite Number of Spectral Points Spread Uniformly Between 0 and 0.5fs (512 Usually Sufficient) Specify Phase Points (Can Make Equal to Zero) Convert Rectangular Form (Real + Imaginary) Take the Complex Inverse FFT of H(k) Array to Obtain the Impulse Response Truncate the Impulse Response to N Points Apply a suitable Window function to h(m) to smooth the effects of Truncation Test Filter Design and Modify if Necessary CAD Design Techniques More Suitable for Lowpass, Highpass, Bandpass, or Bandstop Filters
- 24. FIR CAD Techniques: Parks Mcclellan Program With Remez Exchange Algorithm
- 25. Parks Mcclellan Equiripple FIR Filter Design: Program Inputs
- 26. FIR Filter Program Outputs
- 27. FIR Design Example: Frequency Response
- 28. FIR Filter Design Example: Step Response
- 29. FIR Design Example: Impulse Response (Filter Coefficients)
- 30. Design Example Using ADSP-2189M: Processor Time for 69-TAP FIR Filter Sampling Frequency fs = 44.1kSPS Sampling Interval = 1 / fs = 22.7μs Number of Filter Taps, N = 69 Number of Required Instructions = N + 5 = 74 Processing Time / Instruction = 13.3ns (75MIPS) (ADSP-2189M) Total Processing Time = 74×13.3ns = 984ns Total Processing Time < Sampling Interval with 22.7μs – 0.984μs = 21.7μs for Other Operations Increase Sampling Frequency to 1MHz Use Slower DSP (3.3MIPS) Add More Filter Taps (Up to N = 1700)
- 31. Designing Highpass Filters using Lowpass Filter Impulse Response Spectral Inversion Technique: Design Lowpass Filter (Linear Phase, N odd) Change the Sign of Each Coefficient in the Impulse Response, h(m) Add 1 to the Coefficient at the Center of Symmetry Spectral Reversal Technique: Design Lowpass Filter Change the Sign of Every Other Coefficient in the Impulse Response, h(m) This Reverses the Frequency Domain left-for-right: 0 becomes 0.5, and 0.5 becomes 0; i.e., if the cutoff frequency of the lowpass filter is 0.2, the cutoff of the resulting highpass filter is 0.3
- 32. Bandpass and Bandstop Filters Designed from Lowpass and Highpass Filters
- 33. Additional Resource For ordering the ADSP-21xx, please click the part list or Call our sales hotline For additional inquires contact our technical service hotline For more product information go to http://www.analog.com/en/embedded-processing-dsp/adsp-21xx/processors/index.html Newark Farnell