A Novel Phase-Noise Module for the QUCS Circuit Simulator. Part II : Noise Analysis

This paper presents the implementation of a novel, rigorous time-domain phase-noise analysis module in the open-source QUCS simulator, featuring new closed-form expressions for amplitude and phase-amplitude correlations that surpass existing empirical models and commercial EDA capabilities in predicting the performance of noise-perturbed autonomous circuits.

The Gist

Imagine you are trying to tune a radio to a specific station. You want the signal to be crystal clear, but there's always a little bit of static (noise) in the background. In the world of electronics, this "static" is called phase noise, and it's a huge problem for everything from your smartphone to satellite communications.

This paper is about building a brand-new, super-accurate tool inside a free software program called QUCS to predict exactly how much static will appear in electronic circuits, even when those circuits are working very hard (under "large-signal" conditions).

Here is the breakdown of what the authors did, using simple analogies:

1. The Problem: The Old Maps Were Wrong

For a long time, engineers used "old maps" (mathematical models) to predict noise. These maps were based on simplified assumptions, like assuming the circuit behaves like a straight line or a simple machine.

• The Flaw: The authors point out that these old models are like trying to navigate a mountain range using a flat map of a city. They work okay for simple, straight roads (single, free-running oscillators), but they break down completely when you have complex terrain, like a group of hikers holding hands and walking together (coupled oscillators).
• The "Singularity" Glitch: The old maps had a specific spot where they just stopped working—a "black hole" in the math right at the center of the signal. When engineers tried to calculate noise right at the main frequency, the computer would get confused, the numbers would explode, and the results became garbage.

2. The Solution: A New, 3D GPS (The COSC-PMM)

The authors created a new mathematical framework called COSC-PMM. Think of this not as a flat map, but as a high-tech, 3D GPS that understands the actual shape of the terrain.

• No Shortcuts: Unlike the old models that used "phenomenological" tricks (guessing based on what things look like), this new model is built from the ground up using rigorous physics and geometry. It doesn't guess; it calculates the exact path.
• The "Coupled" Superpower: The biggest breakthrough is that this new GPS works perfectly for coupled circuits. Imagine a choir where singers are listening to each other to stay in tune. If one singer stumbles, it affects the whole group. The old tools couldn't predict how the noise would ripple through the choir. This new tool can predict exactly how the noise travels between the singers and how they stabilize each other.

3. How It Works: The "Dancing" Analogy

To understand the math, imagine an oscillator (a circuit that creates a signal) as a dancer spinning on a stage.

• The Perfect Spin (Steady State): In a perfect world, the dancer spins in a perfect circle forever. This is the "Periodic Steady State" (PSS).
• The Shove (Noise): In the real world, someone occasionally gives the dancer a tiny, random shove. This makes them wobble.
• The Old Way: The old models tried to predict the wobble by looking at the dancer from the side and assuming the wobble was a simple back-and-forth motion.
• The New Way: The authors' new method looks at the dancer from every angle at once. It uses a concept called Floquet Theory (which sounds scary but is just a fancy way of analyzing repeating patterns).
  • It identifies the "rhythm" of the dance.
  • It separates the "phase" (when the dancer is in the rotation) from the "amplitude" (how high they jump).
  • It calculates how the random shoves (noise) change the dancer's rhythm over time.

4. The Result: A Brand New Tool

The authors took this complex math and turned it into a software module that anyone can use inside the free QUCS simulator.

• What it does: It takes a circuit design, runs a simulation, and tells you exactly how much "static" (noise) will be in the signal, even for complex groups of circuits working together.
• The Proof: They tested their new tool against the most expensive, commercial software in the world (Keysight-ADS). The results were almost identical (within 0.03% error), proving their new "free" tool is just as accurate as the "paid" one, but without the mathematical glitches that plague the commercial software when dealing with complex, coupled circuits.

Why Does This Matter?

• Free & Open: This tool is free and open-source. It democratizes high-end engineering, allowing students and small companies to do work that previously required expensive licenses.
• Better Design: Because the tool is more accurate (especially for coupled circuits), engineers can design better radios, faster computers, and more reliable satellite systems. They can predict problems before they build the hardware, saving time and money.
• The Future: This is just the second part of a two-part series. The first part built the engine; this part added the noise-analysis dashboard. The authors are now working to make it even faster and more robust.

In a nutshell: The authors built a new, mathematically perfect "noise detector" for electronic circuits. It fixes the broken parts of the old tools, works for complex groups of circuits, and is free for everyone to use. It's like upgrading from a paper map to a real-time, 3D satellite navigation system for electronic engineers.

Technical Summary

Here is a detailed technical summary of the paper "A Novel Phase-Noise Module for the QUCS Circuit Simulator. Part II: Noise Analysis."

1. Problem Statement

The design and optimization of modern communication and remote-sensing systems rely heavily on the accurate prediction of noise in non-linear, autonomous circuits (oscillators and clocks) operating under large-signal conditions.

• Limitations of Current Tools: Existing Electronic Design Automation (EDA) tools, both commercial (e.g., Keysight ADS, Cadence SpectreRF) and open-source, predominantly rely on Linear-Time-Variant (LTV) or Linear-Time-Invariant (LTI) theories, such as the Conversion-Matrix (C-MATRIX) method.
• Theoretical Flaw: These empirical/phenomenological models treat oscillators as internally driven non-linear mixers. When applied to autonomous systems, they fail to fully capture the underlying stochastic response. This results in an artificial singularity at the carrier frequency (zero offset), causing numerical noise accumulation and rendering the calculated phase-noise spectrum unreliable or invalid near the carrier.
• Gap in Open-Source: No credible open-source simulator currently offers a robust, theoretically complete noise analysis tool for large-signal autonomous or coupled circuits.

2. Methodology

The authors implemented a novel noise analysis module within the open-source QUCS-COPEN simulator (a fork of QUCS/QUCS-S). The core of this implementation is the Coupled Oscillator Phase-Macro-Model (COSC-PMM).

• Theoretical Foundation:
  • The model is based on a rigorous, unified time-domain methodology utilizing differential geometry, topology, Floquet theory, and non-linear stochastic integration.
  • It avoids empirical approximations (LTI/LTV) entirely.
  • It treats the oscillator as a system with a stable Periodic Steady State (PSS) and analyzes the noise response via Floquet decomposition of the linearized system around the limit cycle.
• Key Mathematical Concepts:
  • Floquet Modes: The system is decomposed into phase modes (tangent to the limit cycle) and amplitude modes (transverse to the limit cycle).
  • Phase Diffusion: The model explicitly accounts for the unbounded growth of phase noise (phase diffusion constant $c$) and the bounded nature of amplitude noise.
  • Unified Framework: The COSC-PMM generalizes the previous single-oscillator Phase Macro-Model (PMM). It handles both free-running oscillators and coupled oscillator ensembles (e.g., injection-locked oscillators) within a single mathematical framework.
• Algorithm Implementation (Algorithm 1):
  1. Floquet Decomposition: Calculates the Monodromy Matrix (MM) and its adjoint to derive Floquet vectors ($u_i, v_i$) and exponents ($\mu_i$).
  2. Noise Operator Calculation: Computes spectral operators ($\Omega, \Theta, \Pi, \Xi$) involving infinite sums of harmonic contributions, truncated for numerical efficiency.
  3. Spectrum Synthesis: Generates closed-form expressions for:
     • Phase Noise Spectrum ($L(\nu)$)
     • Amplitude Noise Spectrum ($A(\nu)$)
     • Phase-Amplitude Cross-Correlation Spectrum ($R(\nu)$)

3. Key Contributions

• First Complete Open-Source PNOISE Tool: This is the first EDA project (open-source or commercial) to include a phase-noise analysis module that is mathematically complete for coupled autonomous circuits, free from LTV/LTI approximations.
• Novel Closed-Form Expressions: The paper introduces previously unpublished extensions to the COSC-PMM framework, providing unified closed-form expressions for:
  • Amplitude noise response.
  • Phase-amplitude cross-correlation response.
• Resolution of the Carrier Singularity: By avoiding the LTV approach, the algorithm produces a spectrum that is correct at all offsets, including the carrier frequency, eliminating the artificial singularity and numerical instability found in commercial tools.
• Coupled Circuit Capability: The tool successfully models complex topologies, such as injection-locked oscillators (ILOs) and coupled ensembles, capturing non-Lorentzian spectral characteristics that standard tools miss.

4. Results and Validation

The authors validated the QUCS-COPEN PNOISE module against both theoretical expectations and commercial software.

• Test Cases:
  • Free-Running Oscillators: Van-der-Pol (OSC.#1), BJT Colpitts (OSC.#2), and MOSFET Cross-Coupled LC-tank (OSC.#3).
  • Coupled Oscillators: An Injection-Locked Oscillator (COSC.#1) and a 3-ensemble coupled system (COSC.#2).
• Comparison with Keysight ADS:
  • For free-running oscillators, the QUCS-COPEN results were compared against the Keysight-ADS pnmx simulator.
  • Accuracy: The relative deviation ($\Delta L$) at a 1 kHz offset was extremely low, ranging from 0.0016% to 0.028% (approx. 300 ppm).
  • Coupled Systems: For coupled circuits, the QUCS-COPEN results matched the Keysight-ADS reference across various coupling strengths, correctly reproducing the non-Lorentzian spectral shape of the locked oscillator.
• Visual Evidence: The paper presents plots showing the phase noise, amplitude noise, and cross-correlation spectra, demonstrating that the tool correctly captures the Lorentzian shape for free-running oscillators and the distinct, non-Lorentzian behavior of locked oscillators.

5. Significance

• Scientific Impact: The work establishes that previous models (LTI/LTV) are fundamentally incomplete for autonomous systems. The COSC-PMM framework provides a rigorous, non-phenomenological alternative that unifies the analysis of single and coupled oscillators.
• Industrial Impact: Timing circuits are ubiquitous in modern communications. The ability to accurately predict and optimize noise in coupled systems (e.g., phased arrays, synchronized sensor networks) without relying on flawed approximations is critical for high-performance design.
• EDA Advancement: By integrating this advanced stochastic modeling into an open-source environment, the authors democratize access to state-of-the-art noise analysis tools, potentially surpassing the capabilities of current commercial EDA suites in specific large-signal noise scenarios.

In conclusion, this paper presents a breakthrough in circuit simulation theory and implementation, delivering a robust, singularity-free noise analysis tool that fundamentally improves the accuracy of oscillator design and optimization.