2.3 Passive Filters, Cutoff Frequency, and Attenuator Pads

2.3 Passive Filters, Cutoff Frequency, and Attenuator Pads

In radio frequency systems, signals of interest coexist with unwanted harmonics, mixer spurious products, image frequencies, and adjacent-channel transmissions. Passive filters—constructed from inductors, capacitors, and quartz crystal resonators—provide the essential frequency discrimination required to ensure spectral purity and prevent interference. Complementing filter networks, resistive attenuators stabilize stage impedances, provide inter-stage isolation, and accurately scale signal power levels.

---

Filter Fundamentals and Performance Metrics

A filter is a two-port network designed to transmit signals within one or more designated frequency bands while attenuating signals outside those bands.

Relative transmission (dB)
  0 |----------------.
 -3 |.................*  f_c
    |                  \
    |                   \
-40 |....................*  f_stop (example target)
    |                     \__
    +--------------------------> Frequency
       Passband   Transition   Stopband

The operation of passive filters is characterized by five key parameters:

1. Passband: The frequency range over which signals pass with minimal attenuation (typically defined as attenuation $< 3\text{ dB}$).
2. Stopband: The frequency range over which signals are suppressed to an acceptable floor (often $\ge 40\text{ dB}$ to $60\text{ dB}$).
3. Transition Band: The intermediate frequency range between the passband edge and the stopband edge, characterized by the filter's rolloff rate.
4. Cutoff Frequency: A $-3\text{ dB}$ frequency marks half the reference passband power. Here $f_c$ denotes that frequency. Chebyshev designs commonly specify the ripple-band edge instead; it need not be the $-3\text{ dB}$ frequency. For equal reference impedances and passband voltage gain $H_{\text{ref}}$:

$$\frac{|H(f_c)|}{|H_{\text{ref}}|}=\frac{1}{\sqrt{2}}\approx0.7071,\qquad H=\frac{V_{\text{out}}}{V_{\text{in}}}$$

1. Insertion Loss: The decrease in power delivered to the load when the filter is inserted, with source and load conditions unchanged. It includes dissipation and the effect of impedance mismatch:

$$\text{Insertion Loss (dB)} = 10\log_{10}\left(\frac{P_{\text{load,before}}}{P_{\text{load,after}}}\right)$$

---

Four Fundamental Filter Configurations

• Low-Pass Filters: Standard in HF/VHF transmitter output stages to eliminate harmonic radiation ($2f_0, 3f_0, 4f_0$).
• High-Pass Filters: Connected at receiver antenna inputs to reduce strong MF signals, including stations in Australia's $526.5$–$1606.5\text{ kHz}$ AM broadcasting band.
• Band-Pass Filters: Utilized in transceiver front ends and intermediate frequency (IF) stages to isolate a single amateur band or voice/CW channel.
• Band-Stop / Notch Filters: Used to eliminate single-frequency interference, such as an interfering carrier whistle or nearby repeater transmitter in cross-band operation.

---

Filter Circuit Topologies: L, T, and Pi ($\pi$) Sections

Reactive filters are assembled from elementary building blocks classified by their geometric arrangement:

1. L-Section: An asymmetrical network composed of one series reactive element and one shunt reactive element. Provides a gentle rolloff and acts as a basic impedance transformer.
2. T-Section: Two series elements with a central shunt element. Equal series arms make the section symmetrical.
3. Pi ($\pi$) Section: Two shunt elements with a central series element. Equal shunt arms make it symmetrical; unequal arms can match unequal terminating impedances.

       T-Section Low-Pass Filter            Pi (π) Section Low-Pass Filter
             L1          L2                              L1
        +---[####]--+--[####]---+              +--------[####]--------+
        |           |           |              |                      |
        |          ---          |             --- C1                 --- C2
       In          --- C1      Out            ---                    --- 
        |           |           |              |                      |
        +-----------+-----------+              +----------------------+

In RF transmitter power amplifiers, the Pi ($\pi$) configuration is overwhelmingly preferred for several structural reasons:

• The shunt capacitors at the input and output absorb the stray output capacitance of the amplifier transistor or tube and the stray input capacitance of the output feedline.
• The two shunt capacitors bypass high-frequency harmonics directly to ground at both ports.
• Component values can be adjusted independently to match unequal source and load impedances while filtering.

---

Filter Response Mathematical Approximations

The mathematical polynomial used to synthesize a filter determines its amplitude, phase, and time-domain characteristics. Four classic response types dominate RF engineering:

1. Butterworth (Maximally Flat)

The Butterworth filter response provides the smoothest possible passband transmission with zero amplitude ripple:

$$|H(j\omega)|^2 = \frac{1}{1 + \left(\frac{\omega}{\omega_c}\right)^{2n}}$$

where $n$ is the filter order (number of reactive poles).

• Passband: Ripple-free and maximally flat near DC, then decreasing monotonically to $-3\text{ dB}$ at $f_c$.
• Rolloff: Moderate; asymptotic stopband attenuation increases at $6n\text{ dB per octave}$ ($20n\text{ dB per decade}$).
• Phase Response: Moderately non-linear near the cutoff frequency.

2. Chebyshev (Equal-Ripple)

Chebyshev filters sacrifice passband flatness by introducing a controlled amount of ripple (typically $0.1\text{ dB}$, $0.5\text{ dB}$, or $1.0\text{ dB}$) in exchange for a much steeper transition skirt:

$$|H(j\omega)|^2 = \frac{1}{1 + \epsilon^2 T_n^2\left(\frac{\omega}{\omega_p}\right)}$$

where $T_n$ is the Chebyshev polynomial of order $n$, $\epsilon$ sets ripple amplitude, and $\omega_p$ is the ripple-band edge, generally below the $-3\text{ dB}$ frequency.

• Passband: Features cyclical ripple peaks and troughs.
• Rolloff: Significantly steeper than a Butterworth filter of the same order.
• Phase Response: Poor linearity; causes noticeable group delay distortion near cutoff.

3. Bessel (Maximally Flat Group Delay)

The Bessel (Thomson) filter is synthesized using Bessel polynomials to optimize time-domain transient response rather than frequency-domain amplitude rejection:

• Phase Response: Nearly linear in the passband, with maximally flat group delay near DC ($\tau_g=-d\theta/d\omega$).
• Time Domain: Small overshoot and little ringing preserve pulse shape well, while finite bandwidth still rounds fast edges.
• Rolloff: Very gradual; requires a high order $n$ to achieve substantial out-of-band attenuation.

4. Elliptic / Cauer (Steepest Transition)

Elliptic filters incorporate transmission zeros (notches) at finite frequencies in the stopband:

• Passband and Stopband: Exhibits ripple in both the passband and the stopband.
• Rolloff: The steepest transition rate of any filter topology for a given number of poles.
• Phase Response: Extreme phase distortion and severe ringing on pulse transients.

Comparison Table: Filter Response Profiles

---

Quartz Crystals and Ceramic Resonators

While LC circuits achieve unloaded Q factors of $100$ to $300$, narrow-band intermediate frequency (IF) filters require Q factors in the tens of thousands. This is achieved using piezoelectric quartz crystals.

    Physical Crystal Model                Butterworth-Van Dyke (BVD) Equivalent Circuit
         Quartz Plate                                L_s         C_s         R_s
        +------------+                        +----[####]-------||--------[####]----+
     ===| Metal Elec |===                     |   (Motional)  (Motional)  (Motional)|
        +------------+                        |                                     |
                                              +-----------------||------------------+
                                                                C_p
                                                         (Parallel Shunt)

The Piezoelectric Effect and the BVD Equivalent Circuit

When mechanical pressure is applied across a quartz crystal wafer, an electrical potential appears across its faces. Conversely, applying an alternating voltage induces mechanical vibration. Electrically, this mechanical resonance is modeled by the Butterworth-Van Dyke (BVD) circuit:

• Motional Inductance ($L_s$): Models the vibrating mass of the crystal plate. Typically very large ($0.01\text{ H}$ to several Henries).
• Motional Capacitance ($C_s$): Models the mechanical elasticity (compliance) of the quartz. Extremely small ($0.005\text{ pF}$ to $0.05\text{ pF}$).
• Motional Resistance ($R_s$): Models internal mechanical friction and acoustic losses ($10\ \Omega$ to $50\ \Omega$).
• Parallel Static Capacitance ($C_p$): Formed by the metal plating electrodes and package holder ($2\text{ pF}$ to $7\text{ pF}$).

Because $L_s$ is very large and $R_s$ is low, the unloaded quality factor of a quartz crystal reaches extraordinary levels:

$$Q = \frac{\omega_0 L_s}{R_s} \approx 10,000 \text{ to } 1,000,000$$

Resonant Modes: Series ($f_s$) and Parallel ($f_p$)

A quartz crystal possesses two distinct resonant frequencies spaced closely together:

1. Series Resonant Frequency ($f_s$): Occurs when the motional branch reactances cancel ($X_{Ls} = X_{Cs}$):

$$f_s = \frac{1}{2\pi \sqrt{L_s C_s}}$$

At $f_s$, the impedance of the crystal collapses to its low motional resistance ($R_s \approx 20\ \Omega$). 2. Parallel (Anti-Resonant) Frequency ($f_p$): Occurs when the inductive reactance of the motional branch resonates with the parallel static capacitance $C_p$:

$$f_p = \frac{1}{2\pi \sqrt{L_s \left(\frac{C_s C_p}{C_s + C_p}\right)}}$$

Because $C_p \gg C_s$, $f_p$ is typically only $0.1\%$ to $0.3\%$ higher than $f_s$ (a separation of a few kilohertz on HF). At $f_p$, crystal impedance reaches a sharp peak (hundreds of kilohms).

Crystal Ladder and Lattice Filters

By cascading multiple identical inexpensive microprocessor crystals separated by shunt capacitors to ground, engineers create crystal ladder filters. These filters achieve steep, narrow bandpass responses:

• SSB Voice Filter: Centered at $9.0\text{ MHz}$, bandwidth of $2.4\text{ kHz}$ at $-3\text{ dB}$, with a shape factor below $1.6$.
• CW Filter: Bandwidth of $250\text{ Hz}$ to $500\text{ Hz}$ at $-3\text{ dB}$.

Ceramic filters (utilizing lead zirconate titanate ceramics) operate on similar principles but exhibit lower Q ($500$ to $2000$), providing broader bandwidths suitable for $455\text{ kHz}$ AM and FM receiver IF strips.

---

Resistive Attenuator Pads

Attenuator pads are passive resistive networks designed for a specified signal reduction and reference impedance ($Z_0$, typically $50\ \Omega$). Each port presents $Z_0$ when the opposite port is terminated in $Z_0$.

Primary Functions in RF Engineering

1. Impedance Stabilization: Reduces the influence of reactive or varying loads, bringing the impedance seen by filters, mixers, and oscillators closer to $50\ \Omega$.
2. Level Setting: Attenuates strong test signals to prevent receiver front-end or spectrum analyzer overload.
3. Stage Isolation: Reduces feedback from downstream impedance changes and therefore reduces load pulling of upstream oscillators.

Symmetrical T-Pad and Pi-Pad Architecture

         Symmetrical T-Pad                          Symmetrical Pi (π) Pad
           R1          R2                                      R3
      +---[####]--+--[####]---+                  +------------[####]------------+
      |           |           |                  |                              |
      |          [ ]          |                 [ ]                            [ ]
     In          [R3]        Out                [R1]                           [R2]
      |          [ ]          |                 [ ]                            [ ]
      |           |           |                  |                              |
      +-----------+-----------+                  +------------------------------+

Mathematical Design Formulations

Let $A_{\text{dB}}$ be the desired attenuation, with both source and load resistances equal to $Z_0$ ($50\ \Omega$ here). The linear voltage ratio is:

$$k = \frac{V_{\text{in}}}{V_{\text{out}}} = 10^{\frac{A_{\text{dB}}}{20}}$$

For a symmetrical T-Pad:

$$R_1 = R_2 = Z_0 \left(\frac{k - 1}{k + 1}\right)$$ $$R_3 = Z_0 \left(\frac{2k}{k^2 - 1}\right)$$

For a symmetrical Pi-Pad:

$$R_1 = R_2 = Z_0 \left(\frac{k + 1}{k - 1}\right)$$ $$R_3 = Z_0 \left(\frac{k^2 - 1}{2k}\right)$$

Step-by-Step Worked Attenuator Calculations ($50\ \Omega$ System)

Worked Example for a $10\text{ dB}$ Symmetrical T-Pad ($Z_0 = 50\ \Omega$):

• Voltage ratio: $k = 10^{10/20} = \sqrt{10} \approx 3.1623$.
• Series resistors:

$$R_1 = R_2 = 50 \times \left(\frac{3.1623 - 1}{3.1623 + 1}\right) = 50 \times \left(\frac{2.1623}{4.1623}\right) \approx 25.97\ \Omega \quad (\text{use } 26\ \Omega \text{ or } 27\ \Omega)$$

• Central shunt resistor:

$$R_3 = 50 \times \left(\frac{2 \times 3.1623}{3.1623^2 - 1}\right) = 50 \times \left(\frac{6.3246}{10 - 1}\right) = 50 \times 0.7027 \approx 35.14\ \Omega \quad (\text{use } 35\ \Omega \text{ or } 36\ \Omega)$$

---

Practical Amateur Radio Applications

1. Transmitter Harmonic Suppression Low-Pass Filters: Class-licence section 15 requires attenuation of spurious emissions, relative to power within the necessary bandwidth at the transmitter output, of the lesser of $43+10\log_{10}(p_X)$ and $50\text{ dB}$ below $30\text{ MHz}$, or the lesser of $43+10\log_{10}(P)$ and $70\text{ dB}$ at or above $30\text{ MHz}$. Here $p_X$ is peak-envelope power and $P$ is mean power, in watts. Select the LPF for the required harmonic attenuation and measure compliance at the final output.
2. Medium-Wave Broadcast Band Rejection Filters: A high-pass filter can reduce local AM signals before an HF receiver. A cutoff near $1.8\text{ MHz}$ alone does not establish $50\text{ dB}$ rejection: specify attenuation at the interfering frequencies and check insertion loss across the wanted 160m and 80m bands.
3. Crystal Roofing Filters: A narrow first-IF filter, for example $3\text{ kHz}$ for SSB or $500\text{ Hz}$ for CW, reduces unwanted signals reaching later amplifiers, mixers and ADCs. Protection depends on rejection at each interferer's frequency and the later stages' headroom. It cannot undo overload already produced in the RF front end or first mixer.