Analog Filter Characteristics Analysis

1. Magnitude and Phase Response: The Bode Plot

The complex function $H(j\omega )$ holds all the information about how a system treats sinusoids of different frequencies. We typically analyze it by splitting it into its two components:

• Magnitude Response: $|H(j\omega )|$, which tells us the gain of the system at each frequency.
• Phase Response: $\angle H(j\omega )$, which tells us the phase shift the system introduces at each frequency.

The standard way to visualize these two functions is with a Bode plot, which you've likely encountered in your control systems courses. A Bode plot consists of two graphs:

1. Magnitude response in decibels (dB) vs. frequency on a logarithmic scale.
2. Phase response in degrees or radians vs. frequency on a logarithmic scale.

The decibel scale is a logarithmic measure of gain, defined as:
$G_{dB}=20{\log }_{10}(|H(j\omega )|)$
This scale is useful because it allows us to represent a very wide range of gains and, due to the properties of logarithms, the combined response of cascaded systems can be found by simply adding their individual dB responses. A gain of 1 is 0 dB, a gain of 10 is 20 dB, and a gain of 0.1 is -20 dB. An attenuation by a factor of $\sqrt{2}$ is approximately -3 dB.

Let's ground these concepts in a concrete example. The following video provides an excellent walkthrough of analyzing a simple first-order RC low-pass filter.

As you saw in the video, for the simple RC low-pass filter, the transfer function is:
$H(s)=\frac{1}{RCs+1}$
Substituting $s=j\omega$, we get the frequency response:
$H(j\omega )=\frac{1}{1+j\omega RC}$
From this, we can extract the magnitude and phase responses:

• Magnitude Response: $|H(j\omega )|=\frac{1}{\sqrt{1^2+(\omega RC{)}^2}}=\frac{1}{\sqrt{1+(\omega RC{)}^2}}$
• Phase Response: $\angle H(j\omega )=-\arctan \left(\frac{\omega RC}{1}\right)=-\arctan (\omega RC)$

These two equations define the Bode plot for the filter. Key features of the magnitude plot are:

• Passband: The range of frequencies where the gain is close to its maximum (0 dB for this filter).
• Stopband: The range of frequencies that are significantly attenuated.
• Cutoff Frequency (${\omega }_c$): The boundary between the passband and stopband. For this filter, it is defined at ${\omega }_c=1/RC$. At this frequency, the magnitude is $1/\sqrt{2}$, or approximately -3 dB. This is why it's also called the -3dB point.
• Roll-off: The rate at which attenuation increases in the stopband. For this first-order filter, the gain drops at a rate of -20 dB per decade of frequency.

2. Linking Frequency Response to Poles and Zeros

We can develop a powerful intuition for a filter's frequency response by looking at the locations of its poles and zeros in the s-plane. This geometric perspective connects our previous lessons on the s-plane directly to the Bode plot.

The magnitude $|H(j{\omega }_0)|$ at a specific frequency ${\omega }_0$ can be found geometrically by taking the product of the lengths of vectors from all zeros to the point $j{\omega }_0$ on the imaginary axis, and dividing by the product of the lengths of vectors from all poles to that same point.

This geometric view gives us two key insights:

1. A pole close to the $j\omega$-axis at a frequency ${\omega }_p$ will cause the denominator of the transfer function to become small as $\omega$ approaches ${\omega }_p$. This results in a peak (resonance) in the magnitude response around ${\omega }_p$.
2. A zero on or near the $j\omega$-axis at a frequency ${\omega }_z$ will cause the numerator to become small as $\omega$ approaches ${\omega }_z$. This results in a dip or null (notch) in the magnitude response around ${\omega }_z$.

This powerful concept is the basis for intuitive filter design, which we will explore in the next module.

3. Group Delay: Beyond Phase Shift

While the magnitude response tells us what frequencies get through, the phase response affects the shape of the signal. A non-linear phase response can significantly distort a complex signal, even if the magnitude response is perfectly flat. This is because different frequency components of the signal get delayed by different amounts of time, causing them to arrive out of sync at the output.

To understand this better, let's explore why phase linearity is so important.

As the reading explains, this "envelope delay" is a crucial concept. To quantify it, we introduce two related metrics: Phase Delay and Group Delay.

The following video provides a clear and intuitive explanation of these two concepts, which are central to understanding signal distortion in filters.

To summarize the key definitions from the video:

• Phase Delay (${\tau }_p$): The time delay experienced by an individual sinusoidal frequency (the "carrier"). It is defined as the total phase shift divided by the frequency.
  ${\tau }_p(\omega )=-\frac{\angle H(j\omega )}{\omega }$
• Group Delay (${\tau }_g$): The time delay experienced by the overall envelope of a modulated signal (the "group" of frequencies). It is defined as the negative derivative of the phase response with respect to frequency.
  ${\tau }_g(\omega )=-\frac{d}{d\omega }(\angle H(j\omega ))$

For distortionless transmission, we need all frequency components to be delayed by the same amount of time. This requires a constant group delay. A constant group delay implies a linear phase response ($\angle H(j\omega )=-k\omega$), which is the ideal for preserving a signal's waveshape.

4. Putting It All Together: A Filter's "Personality"

The magnitude, phase, and group delay responses together define the complete "personality" of a filter. Different filter types, which we will study in the next module, represent different trade-offs between these characteristics. For example, some filters have very sharp magnitude roll-off but poor (non-constant) group delay, while others prioritize constant group delay at the expense of a gentler magnitude roll-off.

To see how these characteristics are presented for a standard filter, let's examine the plots for a Butterworth filter, which is known for its "maximally flat" magnitude response in the passband.

Looking at the plots for the 8-pole Butterworth filter, notice that:

• Amplitude: The response is very flat in the passband (from 0 to 1 Hz in this normalized plot) and then rolls off very steeply.
• Group Delay: The delay is not flat. It is relatively constant at low frequencies but rises to a significant peak near the cutoff frequency before falling off. This tells us that frequency components near the edge of the passband will be delayed more than those in the middle, causing phase distortion.
• Step Response: This plot shows the time-domain consequence of the magnitude and phase response. The significant "overshoot" and "ringing" in the step response are direct results of the sharp magnitude cutoff and the non-constant group delay.

This example clearly illustrates the interplay between a filter's characteristics in the frequency domain and its behavior in the time domain.

Conclusion

In this lesson, we have thoroughly analyzed the key characteristics that define an analog filter's performance. Understanding these characteristics is the critical bridge between the abstract mathematics of the s-plane and the practical task of signal filtering.

Key Takeaways:

• A filter's frequency response, $H(j\omega )$, is typically analyzed via its magnitude response, $|H(j\omega )|$, and phase response, $\angle H(j\omega )$, often visualized on a Bode plot.
• The magnitude response shows the filter's gain at each frequency, defining its passband, stopband, and cutoff frequency.
• The phase response describes the time shift applied to each sinusoidal component. A non-linear phase response leads to signal distortion.
• Group Delay, ${\tau }_g(\omega )=-d/d\omega (\angle H(j\omega ))$, is a crucial metric that measures the delay of a signal's envelope. A constant group delay is required for distortionless transmission.
• There is an inherent trade-off between achieving a sharp magnitude cutoff and maintaining a constant group delay.

Preview of the Next Lesson:
We have now completed our analysis of systems in the s-domain. You've seen that realizing an "ideal" filter is impossible and that practical filters involve trade-offs. The next module, "Analog Filter Design Principles," dives into this very problem. We will begin by exploring the historical development of filter approximation theory, learning about the pioneering work of Butterworth, Chebyshev, and Cauer, who developed systematic methods to design filters that approximate an ideal response according to different optimization criteria.