15.3 Practical RFI Suppression, Filter Design, and Bonding Techniques

15.3 Practical RFI Suppression, Filter Design, and Bonding Techniques

Eliminating Radio Frequency Interference (RFI) requires a systematic engineering approach combining source filtering, victim hardening, common-mode suppression, and equipotential radio frequency bonding. An amateur station can produce intense local fields. Australian TableC power limits depend on band and emission: ordinary listed bands permit $400\text{ W}$ PEP for C3F/J3E/R3E and $120\text{ W}$ mean for other emissions. Suppressing unwanted harmonic radiation, blocking conducted currents on utility lines, choking feedline shields, and establishing a low-inductance RF ground plane are foundational skills for advanced amateur radio operators.

---

Transmitter Output Low-Pass Filter Design

An HF transmitter needs output filtering adequate for its actual harmonic spectrum. This may be a switched LPF bank or another suitable tuned output network; a single30–32MHz cutoff does not reject every HF second harmonic.

          Seven-reactance shunt-first low-pass ladder
             L1             L2             L3
Input --+---coil---+--------coil---+--------coil---+-- Output
        |         |              |               |
       C1        C2             C3              C4
        |         |              |               |
Return -+---------+--------------+---------------+-- Return
Values are selected for the passband, impedance and required rejection.

Filter Topologies: Chebyshev vs. Cauer (Elliptic)

• Chebyshev Filters: A TypeI Chebyshev response has passband ripple and a monotonic stopband. For an ideal7-pole,0.1dB-ripple filter, attenuation at $1.5f_p$ is about $36.2\text{ dB}$, where $f_p$ is the ripple-band edge. Required rejection must be calculated at each unwanted frequency; order alone does not establish a50dB result.
• Cauer (Elliptic) Filters: Incorporate transmission zeros (notches) in the stopband by placing parallel-resonant LC tanks in the series arms or series-resonant LC tanks in the shunt arms. For a given component count, an elliptic filter provides the steepest cutoff slope of any filter topology, making it ideal for suppressing second harmonics lying immediately above the $28\text{ MHz}$ band (e.g., $56\text{ MHz}$).

High-Power Component Engineering

For a matched $400\text{ W}$ sinusoidal envelope crest in $50\ \Omega$, the carrier-cycle RMS voltage is $\sqrt{400\times50}=141.4\text{ V}$ and the RF peak is $200\text{ V}$. This can illustrate a400W PEP SSB crest; it does not authorize400W continuous CW or determine average heating.

For a capacitor actually subjected to sinusoidal RMS voltage $V_C$:

$$I_C=2\pi fCV_C.$$

A100pF capacitor at30MHz with141.4V RMS across it carries about2.67A RMS. Voltages/currents within the filter need not equal the port values. Mismatch stress depends on reflection phase, network position and whether forward or net accepted power is fixed;3:1 SWR does not simply triple every voltage.

• Capacitors: Select low-loss RF parts, such as suitable silver-mica or Class1 ceramic capacitors, for the actual RF current, voltage, temperature and duty cycle. Package name or a DC voltage rating alone does not establish RF capability.
• Inductors: Use conductors and cores suitable for the RF current/loss. Coil spacing and differing axes reduce mutual coupling; orientation alone does not guarantee zero coupling.

Coaxial Stubs and Cavity Filters

Transmission line segments function as exceptional, high-power filters at VHF and UHF:

• Open Stub, Quarter-Wave at $f_0$: An ideal lossless open-ended line presents a short at $f_0$. Connected as a shunt branch, it rejects that frequency by severe mismatch/reflection; it is not the desired-signal through path.
• Shorted Stub, Quarter-Wave at $f_0$: It presents an open at $f_0$ and a short at $2f_0,4f_0,\dots$ in the ideal model. As a shunt branch it can therefore reduce even harmonics while loading the fundamental only slightly. Real loss, junction parasitics and bandwidth determine performance.
• Cavity Duplexers: High-Q resonators with suitable coupling and interconnecting lines provide the required transmit/receive selectivity. Verify isolation at the actual split, insertion loss and power rating; cavity construction alone does not establish fixed loadedQ or80–100dB isolation.

---

Receiver and Consumer Device Filtering

High-Pass Filters (HPF) on Victim Receivers

When HF energy entering a television or VHF receiver antenna port overloads its front end, a suitable HPF can reject that energy while passing the wanted service.

• Specification: Select the connector impedance, wanted passband, permissible loss and required rejection at the interfering HF frequencies. A45–50MHz cutoff alone does not guarantee40–50dB rejection at30MHz; a Butterworth filter has3dB loss at its cutoff.
• Placement: Put the filter close to the vulnerable antenna connector so intervening wiring cannot bypass it. If an external masthead amplifier is the first overloaded stage, rejection is needed before that stage. An antenna-port HPF cannot cure RF entering through power/audio leads or directly through the enclosure.

AC Mains Delta Line Filters

Differential-mode current flows out on Active and returns on Neutral. Common-mode current flows in the same direction on the power conductors, with a return through protective earth, capacitance or surrounding conductors. A combined mains filter can address both modes:

Active in --+--[coupled winding A]---+------------- Active out
            |                       |
           [Cx]                    [CyA]
            |                       |
Neutral in -+--[coupled winding N]---|----+-------- Neutral out
                                    |    |
                                    |   [CyN]
                                    |    |
PE in ------------------------------+----+-------- Chassis/PE out

The vertical CyA branch crosses the neutral wire WITHOUT connection.
Windings are oriented so normal load-current flux cancels.
Cx: safety ClassX across Active–Neutral.
CyA/CyN: suitable safety ClassY from each power line to PE.
PE remains continuous; this filter adds no Neutral–PE short.

1. Common-Mode Choke: The windings largely cancel normal load-current flux while adding flux for common-mode RF. Leakage inductance, core properties and parasitic capacitance determine the actual frequency response; there is no universal2000ohm guarantee.
2. ClassX Capacitors: Across-line capacitors shunt differential-mode RF. Select the correct mains/surge safety subclass. ClassX suitability does not guarantee that every catastrophic failure will be open-circuit.
3. ClassY Capacitors: Line-to-earth capacitors provide a common-mode RF path. Use the required safety subclass, not an ordinary capacitor or an X-only part. Their capacitance contributes to normal leakage even without a breakdown.

---

Ferrite Chokes and Common-Mode Suppression

Ferrite suppression devices are the primary frontline tool for eliminating common-mode RFI on feedlines, power leads, and peripheral interconnect cables.

Ferrite Material Science: MnZn vs. NiZn

Ferrites are ceramic ferromagnetic materials produced by sintering iron oxide ($\text{Fe}_2\text{O}_3$) with oxides of nickel, zinc, or manganese. Their electromagnetic suppression properties are governed by their crystalline composition:

• MnZn and NiZn: Chemistry affects permeability, resistivity and loss, but the useful band must come from the selected material/core data. MnZn is not restricted to below2MHz: Mix31 is a common HF suppression material. Inductive low-loss operation and deliberately lossy EMI suppression are different applications.

These material descriptions are not guaranteed bandwidth or impedance specifications for every finished winding.

Complex Permeability and the Resistive Dissipation Mechanism

A ferrite material's magnetic behavior is described mathematically by its complex permeability:

$$\mu = \mu' - j \mu''$$

where $\mu'$ represents the inductive (energy storage) component and $\mu''$ represents the lossy (energy dissipation) component. The total series impedance presented by a ferrite bead placed over a conductor is:

$$Z_s = R_s + j X_s = \omega L_0 \mu'' + j \omega L_0 \mu'$$

where $L_0$ is the air-core inductance of the physical turn geometry.

                     Ferrite Complex Permeability and Impedance

     Impedance (Ohms)
          ▲
          │                                 |Z| Total Impedance
          │                              . - - - .
          │                            /     │     \
          │             Rs (Loss)    /       │       \
          │            ............/.........│........\.......
          │           /           /          │         \
          │          /           /           │          \
          │         /           /            │           \
          │   Xs   /           /             │            \  Xs drops
          │  (Inductive)      /              │               rapidly
          │  .....*..........*               │
          │      /          /                │
          └─────*──────────*─────────────────┼────────────────► Frequency
               Low Freq                     f_target (Resistive Peak)
               (Pure Inductance:            (R_s >> X_s: RF Dissipated
                Reflects Energy)             as Thermal Heat)

• Reactive and Resistive Impedance: Below resonance, a choke may be mainly inductive; a resistive component can reduce resonant common-mode current more predictably. The balance varies with material, frequency and winding capacitance. It need not become purely resistive at one universal target frequency.

Turns and Self-Resonance

For the same core and winding geometry, sufficiently below self-resonance, the small-signal approximation is $Z_N\approx N^2Z_1$. Five turns can then give about25times that core’s one-turn impedance; this does not compare it with an unrelated snap-on bead.

Inter-turn capacitance and the core’s frequency-dependent losses alter that scaling. The simple resonance estimate $f_{\text{SRF}}\approx1/(2\pi\sqrt{LC_p})$ helps explain why adding turns shifts the response. Above a winding’s useful band, capacitive bypass can reduce choking impedance. Choose the core/winding from measured data at the needed frequencies and verify suppression and heating in the actual installation;5–8turns on a nominal FT-240-31 is not a universal3.5–30MHz specification.

---

Shielding and Bonding Practice

Faraday Enclosure Shielding and Skin Depth

A conductive metallic enclosure suppresses radiated electromagnetic fields through two primary mechanisms: reflection loss at the metallic boundary and absorption loss as the wave penetrates the conductive wall. The depth at which an electromagnetic wave attenuates to $1/e$ ($36.8\%$) of its surface amplitude is the skin depth ($\delta$):

$$\delta = \sqrt{\frac{\rho}{\pi f \mu}} = \frac{1}{\sqrt{\pi f \mu \sigma}}$$

For high-conductivity metals such as copper ($\sigma \approx 5.8 \times 10^7\text{ S/m}$) and aluminum ($\sigma \approx 3.5 \times 10^7\text{ S/m}$) at radio frequencies:

• At $14\text{ MHz}$, the skin depth of copper is $\delta \approx 17.5\ \mu\text{m}$; for aluminum, $\delta \approx 22.5\ \mu\text{m}$.
• A standard $1.6\text{ mm}$ aluminum equipment chassis represents over $70\text{ skin depths}$, providing hundreds of decibels of theoretical absorption loss.

                      Slot Leakage in Metallic Enclosures

            Poor: Long Continuous Seam       Good: Screws Spaced < λ/20
            Acts as a Slot Antenna!          Maintains Shielding Integrity
            ──────────────────────────       ────┬────┬────┬────┬────┬────
            [       Slot L > λ/20    ]           •    •    •    •    •    •
            ──────────────────────────       ────┴────┴────┴────┴────┴────
                   ▲                                       ▲
                   │ RF Leaks                              │ RF Trapped
                   │ Freely                                │ Inside

RF Bonding vs. $50\text{ Hz}$ Electrical Safety Grounding

One of the most dangerous misconceptions in amateur radio engineering is confusing an electrical safety ground with an RF ground.

• Protective Earth: The green/yellow safety conductor provides a fault-current path and participates in the installation’s shock-protection system. Preserve it. RF bonding does not replace protective earth, and an earth rod is not automatically a low-impedance RF reference.
• RF Bonding: Short, substantial connections between equipment chassis reduce RF voltage differences and provide controlled return paths. They reduce interference risk but do not guarantee zero chassis voltage.

Why a Long Earth Lead Is an Unreliable RF Reference

A conductor’s inductance depends on its dimensions and return path. As an illustrative lumped estimate, $L=5\ \mu\text{H}$ gives

$$X_L=2\pi fL\approx440\ \Omega\quad\text{at14MHz}.$$

A5m run is also a significant fraction of the20m wavelength. Distributed capacitance, return-path geometry and resonances then matter, so the lumped estimate is not its complete input impedance. A quarter-wave shorted transmission line transforms a low load to a high input impedance in the ideal model; a single earth wire does not have a specified $Z_0$ and ideal zero-ohm termination merely because it is5m long. Avoid long leads as the sole RF reference.

Radio chassis ----- short bond ----+
Amplifier chassis - short bond ----+--- local bonding bar
ATU chassis ------- short bond ----+          |
PC chassis -------- short bond ----+          |
                              coordinated connection to the
                              installation earthing/protection system
Protective-earth conductors remain intact throughout.

Practical Bonding

1. Keep interequipment bonds short and direct; a nearby bonding bar is one useful arrangement. Choose conductor dimensions, joints and routing for the installation and relevant frequencies.
2. Wide strap can reduce inductance for a given route, but there is no universal50mm minimum or one-tenth-impedance guarantee. Good short round-wire bonds can also be useful. Protect connections against corrosion and mechanical damage.
3. Coordinate additional electrodes, station bonds and lightning protection with the building’s earthing system. MEN means multiple earthed neutral. An appropriately licensed electrician must determine any required installation-earthing work and conductor sizes under applicable Australian requirements; do not create an extra neutral–earth link or disconnect protective earth to cure RFI.

---

Step-by-Step Worked Calculations

Worked Example 1: Skin Depth and High-Pass Filter Design

1. Calculate the skin depth ($\delta$) of an aluminum chassis enclosure ($\rho = 2.82 \times 10^{-8}\ \Omega\cdot\text{m}$, $\mu = \mu_0 = 4\pi \times 10^{-7}\text{ H/m}$) at $f = 28.500\text{ MHz}$.
2. For an enclosure wall thickness of $t = 1.60\text{ mm}$, calculate the number of skin depths and the theoretical bulk absorption loss.
3. An LC high-pass filter for a television receiver requires a cutoff frequency of $f_c = 50.0\text{ MHz}$ into a $Z_0 = 75\ \Omega$ system. Using a standard 3-pole Butterworth prototype ($g_1 = 1.0, g_2 = 2.0, g_3 = 1.0$), calculate the values of the series input capacitor ($C_1$), shunt inductor ($L_2$), and series output capacitor ($C_3$).

Step 1: Calculate the skin depth of aluminum at 28.5 MHz:

$$\delta = \sqrt{\frac{\rho}{\pi f \mu}} = \sqrt{\frac{2.82 \times 10^{-8}}{\pi \times (28.5 \times 10^6) \times (4\pi \times 10^{-7})}}$$ $$\text{Denominator} = \pi \times (2.85 \times 10^7) \times (1.2566 \times 10^{-6}) = 1.1251 \times 10^2 = 112.51$$ $$\delta = \sqrt{\frac{2.82 \times 10^{-8}}{112.51}} = \sqrt{2.5064 \times 10^{-10}} \approx 1.583 \times 10^{-5}\text{ m} = 15.83\ \mu\text{m}$$

Step 2: Calculate skin depths and absorption loss:

$$\text{Skin Depths} = \frac{t}{\delta} = \frac{1.60 \times 10^{-3}\text{ m}}{1.583 \times 10^{-5}\text{ m}} \approx 101.1\text{ skin depths}$$

Because absorption loss is $A = 8.686\text{ dB}$ per skin depth:

$$A_{\text{absorption}} = 101.1 \times 8.686\text{ dB} \approx 878\text{ dB}$$

The calculated878dB is an ideal bulk-absorption result for this wall material, thickness and frequency. It does not predict a complete enclosure’s measured shielding. Seams, openings, joints and cable entries will normally limit practical HF performance; solid-wall penetration remains a finite, frequency-dependent mechanism.

Step 3: Calculate the 3-pole Butterworth High-Pass Filter values:

For a normalized low-pass prototype converted to high-pass, the component transformation formulas are:

$$C = \frac{1}{2\pi f_c Z_0 g_k}, \quad L = \frac{Z_0}{2\pi f_c g_k}$$

Calculate series capacitors $C_1$ and $C_3$ ($g_1 = g_3 = 1.0$):

$$C_1 = C_3 = \frac{1}{2\pi \times (50.0 \times 10^6\text{ Hz}) \times 75\ \Omega \times 1.0} = \frac{1}{2.3562 \times 10^{10}} \approx 4.244 \times 10^{-11}\text{ F} = 42.4\text{ pF}$$

Calculate shunt inductor $L_2$ ($g_2 = 2.0$):

$$L_2 = \frac{Z_0}{2\pi f_c g_2} = \frac{75\ \Omega}{2\pi \times (50.0 \times 10^6\text{ Hz}) \times 2.0} = \frac{75}{6.2832 \times 10^8} \approx 1.194 \times 10^{-7}\text{ H} = 119.4\text{ nH}$$

The filter consists of a $42.4\text{ pF}$ series input capacitor, a $119.4\text{ nH}$ shunt inductor to ground, and a $42.4\text{ pF}$ series output capacitor.

---

Worked Example 2: Ferrite Toroid Choking Impedance and Turns Calculation

For an illustrative model, suppose one turn through a chosen core measures $Z_1=30+j40\ \Omega$ at7.100MHz. This is given data, not a specification of every FT-240-31 core. Assume the winding remains below self-resonance and $Z_N\approx N^2Z_1$.

1. Find turns for $\lvert Z_N\rvert\ge1500\ \Omega$.
2. A60V RMS common-mode source drives a loop with50ohms total impedance before the choke. Find current before and after adding the choke, current reduction and choke heating. The50ohms is an assumed common-mode loop impedance, not the coax’s differential characteristic impedance.

Step1: Approximate Turns

$$\lvert Z_1\rvert=\sqrt{30^2+40^2}=50\ \Omega, \quad N\ge\sqrt{1500/50}=5.48.$$

Choose6turns. Then

$$Z_6=36(30+j40)=1080+j1440\ \Omega, \quad\lvert Z_6\rvert=1800\ \Omega.$$

Step2: Complete-Loop Current

Without the choke, $I_0=60/50=1.20\text{ A RMS}$. Add the complex impedances before taking the magnitude:

$$Z_{\text{total}}=50+1080+j1440=1130+j1440\ \Omega,$$ $$\lvert Z_{\text{total}}\rvert=1830.44\ \Omega, \quad I_1=60/1830.44=32.78\text{ mA RMS}.$$

Step3: Reduction and Heating

$$A_I=20\log_{10}(1.20/0.03278)=31.27\text{ dB}.$$ $$P_{\text{choke}}=I_1^2\operatorname{Re}(Z_6)=(0.03278)^2(1080)\approx1.16\text{ W}.$$

This model predicts substantial reduction, not guaranteed elimination of interference. Verify the finished choke’s impedance and thermal capability, because winding capacitance, the actual return path and temperature can change the result.

---

Practical RFI Suppression Hardware Comparison Table