🎓 Lesson 8
D5
Filter Transfer Functions, Insertion Loss, and Real-World Limitations
A filter transfer function tells you how much a filter weakens or changes different frequencies in a signal, like turning down bass or treble on a speaker.
🎯 Learning Objectives
- ✓ Calculate insertion loss (in dB) from a given transfer function magnitude at specified frequencies
- ✓ Analyze Bode plots to identify cutoff frequency, roll-off rate, and resonant peaks in passive filter designs
- ✓ Explain how real-world non-idealities (parasitic capacitance, ESR, inductor Q) degrade theoretical transfer function performance
- ✓ Apply transfer function concepts to select appropriate filter topology (low-pass, band-stop) for conducted EMI compliance (e.g., CISPR 11/32 limits)
📖 Why This Matters
In mining and blasting operations, variable-frequency drives (VFDs) for conveyor belts, ventilation fans, and drill rigs generate high-frequency noise that couples into sensitive instrumentation—like seismic monitors or blast timing systems—causing false triggers or data corruption. Understanding filter transfer functions lets engineers predict *exactly* how much noise a filter will suppress *before* installation—avoiding costly rework, downtime, and non-compliance with IEC 61000-4-4 (EFT) or CISPR 11 emission limits.
📘 Core Principles
A transfer function models how a linear, time-invariant (LTI) filter responds to sinusoidal inputs across frequency: its magnitude |H(f)| defines insertion loss (IL), while ∠H(f) defines phase shift. First-order RC low-pass filters exhibit -20 dB/decade roll-off; second-order LC filters achieve -40 dB/decade but introduce resonance if damping is insufficient. Real filters deviate due to parasitics: capacitor equivalent series resistance (ESR) flattens attenuation at high frequencies; inductor self-capacitance creates unintended parallel resonance above f_res = 1/(2π√(L·C_parasitic)); PCB trace inductance degrades high-frequency common-mode rejection. These deviations directly impact whether a filter passes CISPR 32 Class A radiated emissions testing at 30–300 MHz.
📐 Insertion Loss from Transfer Function
Insertion loss quantifies the filter’s attenuation in decibels and is derived directly from the magnitude of the transfer function. For a two-port network, IL(f) = −20 log₁₀|H(f)| dB. This formula applies to both simulated and measured S₂₁ parameters (forward transmission coefficient) and is the primary metric used in EMC validation reports.
💡 Worked Example
Problem: A π-filter (C-L-C) designed for a VFD’s DC bus exhibits |H(f)| = 0.0316 at 1 MHz. Calculate insertion loss at this frequency.
1.
Step 1: Identify |H(f)| = 0.0316
2.
Step 2: Apply IL(f) = −20 log₁₀(0.0316)
3.
Step 3: Compute log₁₀(0.0316) ≈ −1.5 → −20 × (−1.5) = +30 dB
Answer:
The insertion loss is 30 dB, meaning the filter attenuates 1 MHz noise by a factor of 31.6 in voltage (or 1000× in power), well within typical target ranges for VFD harmonic suppression.
🏗️ Real-World Application
At the BHP Olympic Dam copper-uranium mine, blast initiation systems experienced intermittent timing errors during concurrent VFD operation on ore-haul trucks. EMI diagnostics revealed 2–5 MHz common-mode noise coupling via shared grounding. Engineers modeled a 3rd-order LC low-pass filter (10 µH / 10 nF / 10 µH) using SPICE with realistic ESR (50 mΩ) and inductor SRF (15 MHz). The simulated transfer function predicted 42 dB IL at 3 MHz—but measured insertion loss was only 28 dB due to 80 pF inter-winding capacitance in the custom inductor, shifting resonance to 4.1 MHz and creating a gain peak. Revised design used shielded, low-C_parasitic inductors and added ferrite beads—achieving 38 dB IL and passing CISPR 11 Class A conducted emissions testing.