IIR filter design (butter, cheby1) in SciPy - Step-by-Step Execution

Concept Flow - IIR filter design (butter, cheby1)

Choose filter type: Butterworth or Chebyshev1

↓

Set filter order and cutoff frequencies

↓

Call scipy.signal function butter() or cheby1()

↓

Get filter coefficients (b, a)

↓

Use coefficients to filter data or analyze frequency response

↓

End

This flow shows how to design an IIR filter by choosing type, setting parameters, getting coefficients, and applying the filter.

Execution Sample

SciPy

from scipy.signal import butter, cheby1
b, a = butter(3, 0.2)
print(b, a)

Design a 3rd order Butterworth lowpass filter with cutoff 0.2 (normalized frequency) and print coefficients.

Execution Table

Step	Action	Input Parameters	Output (b coefficients)	Output (a coefficients)
1	Call butter()	order=3, cutoff=0.2	[0.0181, 0.0543, 0.0543, 0.0181]	[1.0, -1.476, 0.9565, -0.1976]
2	Print coefficients	b, a arrays	Printed b coefficients	Printed a coefficients
3	End	No further action	-	-

💡 Filter coefficients computed and printed; execution ends.

Variable Tracker

Variable	Start	After butter() call	Final
b	None	[0.0181, 0.0543, 0.0543, 0.0181]	[0.0181, 0.0543, 0.0543, 0.0181]
a	None	[1.0, -1.476, 0.9565, -0.1976]	[1.0, -1.476, 0.9565, -0.1976]

Key Moments - 2 Insights

Why do the coefficients b and a have different values?

The b coefficients represent the numerator polynomial and a coefficients the denominator polynomial of the filter transfer function. For a 3rd order filter, b has length 4 (order+1) and a also length 4, but their values differ because they define different parts of the filter.

What does the cutoff frequency 0.2 mean in butter(3, 0.2)?

The cutoff 0.2 is normalized frequency, meaning 0.2 times the Nyquist frequency (half the sampling rate). So the filter passes frequencies below 0.2 * Nyquist and attenuates above.

Visual Quiz - 3 Questions

Look at the execution table, what is the value of the first b coefficient after butter() call?

A 0.0181

B 1.0

C -1.476

D 0.2

💡 Hint

Check the 'Output (b coefficients)' column in step 1 of the execution table.

At which step does the filter coefficients get printed?

A Step 1

B Step 3

C Step 2

D Never printed

💡 Hint

Look at the 'Action' column in the execution table for printing.

If we change the order from 3 to 5, how will the length of b coefficients change?

A Length will be 5

B Length will be 6

C Length will be 3

D Length stays the same

💡 Hint

Recall that for order N, coefficient arrays have length N+1 as shown in variable_tracker.

Concept Snapshot

IIR filter design with scipy.signal:
- Use butter(order, cutoff) or cheby1(order, ripple, cutoff)
- cutoff is normalized frequency (0 to 1, 1 = Nyquist)
- Returns b (numerator) and a (denominator) coefficients
- Use coefficients to filter signals or analyze response
- Order controls filter sharpness and complexity

Full Transcript

This visual execution shows how to design an IIR filter using scipy.signal butter() function. First, you choose the filter type and parameters like order and cutoff frequency. Then you call butter() with these parameters. The function returns two arrays: b and a, which are the filter coefficients. These coefficients define the filter's behavior. The example shows a 3rd order Butterworth lowpass filter with cutoff 0.2 normalized frequency. The coefficients are printed. The variable tracker shows how b and a change from None to their computed values. Key moments clarify why b and a have different roles and what cutoff frequency means. The quiz questions test understanding of coefficients values, steps, and effect of changing order. The snapshot summarizes the main points for quick reference.