Menu
Home
Products
Resources
Blog Contact
Request Quote
Engineer Tools

Phase Noise Budget Spreadsheet Methodology

Phase Noise Budget Spreadsheet Methodology

1. Introduction and Purpose

Phase noise represents the frequency-domain manifestation of short-term instability in oscillators and frequency sources, fundamentally limiting the performance of modern communication, radar, navigation, and test systems. A phase noise budget is a systematic accounting framework that allocates phase noise contributions across an entire signal chain—from primary reference through distribution, processing, and final output—to ensure system-level performance objectives are met. The spreadsheet methodology transforms complex frequency-domain analyses into a tractable, repeatable engineering workflow.

The primary purpose of implementing a structured spreadsheet approach is threefold. First, it enables quantitative prediction of aggregate phase noise performance at critical system nodes, including the final output or receiver local oscillator. Second, it facilitates component selection and design trade-offs by revealing which stages dominate the overall phase noise contribution. Third, it provides documentation and traceability for design reviews, procurement specifications, and troubleshooting activities.

In modern telecommunications infrastructure, phase noise directly impacts error vector magnitude (EVM), adjacent channel leakage ratio (ACLR), and receiver sensitivity in 5G NR and LTE systems. In precision timing applications such as GNSS receivers or network synchronization, phase noise translates to timing jitter that degrades positioning accuracy and network holdover performance. This guide provides practicing engineers with a concrete, actionable methodology to construct, populate, and interpret phase noise budgets using spreadsheet tools, bridging theoretical concepts with practical system design.

2. Technical Background

2.1 Phase Noise Definition and Representation

Phase noise is formally defined as the ratio of noise power in a 1-Hz bandwidth at a specified offset frequency from the carrier to the total carrier power. It is expressed in decibels relative to the carrier per Hertz (dBc/Hz). The single-sideband (SSB) phase noise, denoted as $\mathcal{L}(f_m)$, is the industry-standard specification:

$$ \mathcal{L}(f_m) = 10 \cdot \log_{10}\left(\frac{P_{SSB}(f_m, \Delta f=1\text{Hz})}{P_{carrier}}\right) \quad \text{[dBc/Hz]} $$

where $f_m$ is the offset frequency from the carrier, $P_{SSB}$ is the power in a 1-Hz bandwidth in the sideband, and $P_{carrier}$ is the carrier power. This definition assumes that the total noise power is equally split between the two sidebands (AM and PM), and for well-behaved oscillators, the phase noise component dominates at offset frequencies of interest.

2.2 Origins and Characterization

Phase noise originates from multiple physical mechanisms within electronic components. In crystal oscillators, thermal noise and flicker noise in the sustaining amplifier contribute to the $1/f^3$ and $1/f^2$ regions near the carrier, respectively. In phase-locked loops (PLLs), the voltage-controlled oscillator (VCO) phase noise dominates at far offsets, while the reference and charge pump/phase frequency detector (PFD) dominate near the carrier. Digital frequency dividers add quantization noise that appears as white phase noise plateau.

The IEEE Std 1139-2008 provides standardized definitions and measurement methodologies for frequency stability, including phase noise characterization. Phase noise profiles are typically characterized by piecewise-linear approximations in log-log (dBc/Hz vs. log offset frequency) plots, with distinct regions exhibiting different power-law slopes ($f^0$ for white phase noise, $f^{-1}$ for flicker phase noise, $f^{-2}$ for white frequency noise, etc.).

2.3 Key System Impacts

The relationship between phase noise and system performance metrics is critical for budget allocation. For digital communications, phase noise contributes to EVM through the following approximation for small phase perturbations:

$$ \text{EVM}_{RMS} \approx \sqrt{2 \cdot \int_{f_1}^{f_2} \mathcal{L}(f) \, df} $$

where the integration limits correspond to the signal bandwidth. For radar systems, phase noise limits the clutter cancellation ratio and creates spurious targets in Doppler processing. In coherent optical communications, laser phase noise broadens the signal spectrum and causes constellation rotation, requiring digital signal processing (DSP) compensation.

3. Tool/Methodology Overview

3.1 Spreadsheet Architecture

The phase noise budget spreadsheet consists of several interconnected worksheets (tabs) organized to mirror the signal chain architecture. A typical structure includes:

Table 1: Recommended Spreadsheet Tab Structure

Tab NamePurposeKey Outputs
Signal Chain Component listing and order Block diagram reference
Component Parameters Individual component specifications Phase noise profiles, gains, losses
Budget Calculator Aggregation of contributions Total phase noise at each node
System Requirements Performance targets Margin analysis, compliance status
Reference Data Formulas, constants, lookup tables Standards, conversion factors

The methodology leverages the fact that phase noise contributions add directly in power (linear scale) when the sources are uncorrelated. For correlated sources (such as common reference clock trees), the contributions add vectorially, requiring consideration of phase relationships.

3.2 Fundamental Assumptions and Limitations

The spreadsheet methodology makes several simplifying assumptions that engineers must understand:

  • Linearity: Components operate in their linear regions; compression or saturation is not considered.
  • Superposition: Phase noise contributions from different sources are additive in power when uncorrelated.
  • Steady-state: The analysis assumes steady-state operation; transient behaviors (e.g., PLL acquisition) require separate analysis.
  • Band-limited: Contributions are calculated over discrete offset frequency points; continuous integration is approximated.

4. Step-by-Step Procedure

4.1 Define System Requirements and Architecture

Begin by specifying the system-level phase noise requirements at critical test points. For a 5G NR base station transmitter at 3.5 GHz, typical requirements might be:

  • At 10 kHz offset: -105 dBc/Hz
  • At 100 kHz offset: -125 dBc/Hz
  • At 1 MHz offset: -140 dBc/Hz
Document the signal chain architecture with sufficient detail to identify every component that contributes phase noise: reference oscillators, PLLs, frequency multipliers/dividers, amplifiers, mixers, and distribution networks.

4.2 Create Component Database

Construct a component database with the following fields for each active phase noise contributor:

Table 2: Component Database Structure

FieldDescriptionExample Entry
Component ID Unique identifier PLL1
Type Category (OCXO, PLL, etc.) Integer-N PLL
Input Frequency Carrier frequency in 10 MHz
Output Frequency Carrier frequency out 2400 MHz
Conversion Gain/Loss Power gain in dB -3 dB
Phase Noise Profile dBc/Hz at specified offsets -80 @1kHz, -110 @10kHz, -130 @100kHz
Correlation Factor 0 to 1, with other sources 0 (uncorrelated)

For components with known phase noise models, use parametric equations. For a free-running VCO, the SSB phase noise can be approximated by:

$$ \mathcal{L}_{VCO}(f_m) = 10 \cdot \log_{10}\left(\frac{k_B T}{2P_s} \cdot \left(\frac{f_0}{2Q_L f_m}\right)^2 \cdot F \cdot \frac{1}{1 + (f_m/f_c)}\right) $$

where $k_B$ is Boltzmann's constant, $T$ is temperature in Kelvin, $P_s$ is signal power, $f_0$ is oscillation frequency, $Q_L$ is loaded quality factor, $F$ is noise figure, and $f_c$ is flicker corner frequency.

4.3 Construct the Budget Calculator

The core calculation aggregates phase noise contributions from all components in the signal chain. For each offset frequency of interest, and for each node in the signal chain, compute the total phase noise using the following procedure:

Step 1: Convert all phase noise values from dBc/Hz to linear (power ratio per Hz).

For each component $i$ at offset frequency $f_m$:

$$ \mathcal{L}_{i,linear}(f_m) = 10^{\mathcal{L}_{i,dBc}(f_m)/10} $$

Step 2: Account for frequency conversion effects.

When a signal at frequency $f_1$ is multiplied by factor $N$ to produce frequency $f_2 = N \cdot f_1$, the phase noise degrades by:

$$ \mathcal{L}_{multiplied}(f_m) = \mathcal{L}_{source}(f_m) + 20 \cdot \log_{10}(N) \quad \text{[dBc/Hz]} $$

Conversely, division by factor $M$ improves phase noise by $20 \cdot \log_{10}(M)$.

Step 3: Propagate contributions through the signal chain.

At each node $n$, the total phase noise is the sum of all upstream contributions, each appropriately scaled by gain/loss and frequency conversion:

$$ \mathcal{L}_{total,n}(f_m) = \sum_{i=1}^{K} G_i \cdot \mathcal{L}_{i,linear}(f_m) + \mathcal{L}_{local,n}(f_m) $$

where $G_i$ is the cumulative gain from source $i$ to node $n$ (in linear scale), and $\mathcal{L}_{local,n}$ is the local phase noise contribution at node $n$.

Step 4: Convert back to dBc/Hz for reporting.

$$ \mathcal{L}_{total,n,dBc}(f_m) = 10 \cdot \log_{10}(\mathcal{L}_{total,n,linear}(f_m)) $$

4.4 Implement Margin and Compliance Analysis

Compare calculated total phase noise against system requirements at each node. Compute margin as:

$$ \text{Margin}(f_m) = \mathcal{L}_{required}(f_m) - \mathcal{L}_{calculated}(f_m) \quad \text{[dB]} $$

Positive margin indicates compliance. Industry practice typically requires 3-6 dB margin to account for temperature variations, aging, and manufacturing tolerances.

5. Example Calculations and Data

5.1 System Description

Consider a simplified microwave link transmitter with the following signal chain:

  • 10 MHz OCXO Reference: Phase noise: -110 dBc/Hz @ 10 Hz, -130 dBc/Hz @ 100 Hz, -140 dBc/Hz @ 1 kHz, -150 dBc/Hz @ 10 kHz
  • PLL Multiplier (10 MHz → 100 MHz): Integer-N PLL with $N=10$. PLL noise floor: -155 dBc/Hz. Charge pump current: 1 mA. Loop bandwidth: 10 kHz.
  • Frequency Doubler (100 MHz → 200 MHz): Passive diode doubler with conversion loss: 8 dB.
  • Power Amplifier: Gain: 20 dB, Noise Figure: 6 dB.
System requirement at 200 MHz output: -100 dBc/Hz @ 1 kHz offset, -120 dBc/Hz @ 10 kHz offset.

5.2 Budget Calculation at 1 kHz Offset

Step 1: OCXO contribution at 1 kHz: -140 dBc/Hz. After PLL multiplication by 10, this becomes:

$$ -140 + 20 \cdot \log_{10}(10) = -140 + 20 = -120 \text{ dBc/Hz} $$

Step 2: PLL noise floor contribution: The PLL's own phase noise at 1 kHz is specified as -155 dBc/Hz. Since this is already at the output frequency, no conversion needed.

Step 3: Determine which source dominates at 1 kHz: The OCXO contribution (-120 dBc/Hz) is worse than the PLL noise floor (-155 dBc/Hz), so the OCXO dominates.

Step 4: Frequency doubler effect: The passive doubler introduces 8 dB loss but does not add significant phase noise. The OCXO contribution remains -120 dBc/Hz at 200 MHz.

Step 5: Power amplifier contribution: Amplifiers add phase noise primarily through their noise figure and gain. The amplifier's additive phase noise can be estimated as:

$$ \mathcal{L}_{PA}(f_m) = 10 \cdot \log_{10}\left(\frac{k_B T \cdot F}{2P_{in}}\right) + \text{gain} $$

where $F$ is the noise figure (linear, $F=10^{6/10}=3.98$), $P_{in}$ is input power. For a -10 dBm input signal:

$$ P_{in} = 0.1 \text{ mW} = 10^{-4} \text{ W} $$

$$ \mathcal{L}_{PA}(1\text{kHz}) \approx 10 \cdot \log_{10}\left(\frac{(1.38 \times 10^{-23})(290)(3.98)}{2 \cdot 10^{-4}}\right) = -174 + 17 + 3 = -154 \text{ dBc/Hz} $$

(This calculation uses the approximation that amplifier phase noise is dominated by thermal noise amplified by gain, valid at offset frequencies beyond the 1/f corner.)

Step 6: Total phase noise at 200 MHz output, 1 kHz offset: Convert each contribution to linear and sum.

  • OCXO (after multiplication): -120 dBc/Hz → $10^{-12}$ W/Hz
  • PLL noise floor: -155 dBc/Hz → $10^{-15.5}$ W/Hz
  • Power amplifier: -154 dBc/Hz → $10^{-15.4}$ W/Hz
Total linear: $10^{-12} + 10^{-15.5} + 10^{-15.4} \approx 1.0003 \times 10^{-12}$ W/Hz

Convert to dBc/Hz: $10 \cdot \log_{10}(1.0003 \times 10^{-12}) = -119.998 \approx -120$ dBc/Hz

Step 7: Compliance check: Requirement is -100 dBc/Hz at 1 kHz. Calculated -120 dBc/Hz provides 20 dB margin.

5.3 Budget Calculation at 10 kHz Offset

At 10 kHz offset, the PLL noise floor may dominate. Assume the PLL phase noise at 10 kHz is -160 dBc/Hz.

  • OCXO contribution (after ×10): $-150 + 20 = -130$ dBc/Hz
  • PLL noise floor: -160 dBc/Hz
  • Power amplifier: $-154 + 10 \cdot \log_{10}(10) = -144$ dBc/Hz (approximating 10 dB/decade slope)
The OCXO contribution (-130 dBc/Hz) still dominates. Total remains approximately -130 dBc/Hz, providing 10 dB margin against the -120 dBc/Hz requirement.

6. Common Mistakes and Pitfalls

6.1 Inconsistent Reference Bandwidth

A frequent error is mixing phase noise specifications with different noise bandwidths. The standard definition assumes 1-Hz bandwidth, but some datasheets specify noise in 10-Hz or 100-Hz bandwidths. Always normalize to 1-Hz bandwidth by subtracting $10 \cdot \log_{10}(\text{bandwidth})$ from the quoted value.

6.2 Ignoring Frequency Conversion Properly

When multiplying or dividing frequencies, engineers sometimes apply the $20 \cdot \log_{10}(N)$ factor incorrectly. For frequency multipliers, the phase noise increases by this factor. For frequency dividers (including prescalers in PLLs), phase noise decreases by this factor. For mixers, the phase noise of the local oscillator (LO) directly translates to the output with the same dBc/Hz value (assuming single-sideband mixing).

6.3 Overlooking Correlation Effects

The simple linear addition assumes uncorrelated noise sources. When multiple components share the same reference clock (common in phased arrays or multi-channel transceivers), their phase noise contributions may be correlated. Correlated contributions add as voltage, not power:

For two sources with correlation coefficient $\rho$ (0 ≤ ρ ≤ 1):

$$ \mathcal{L}_{total} = \mathcal{L}_1 + \mathcal{L}_2 + 2\rho\sqrt{\mathcal{L}_1 \cdot \mathcal{L}_2} $$

6.4 Misinterpreting Residual vs. Absolute Phase Noise

Component datasheets may specify residual phase noise (for two-port devices like amplifiers or dividers) which measures output phase noise relative to input, canceling the source contribution. Absolute phase noise includes the source. Mixing these specifications leads to double-counting or omission.

6.5 Neglecting Power Supply and Substrate Noise

Spreadsheets focusing only on the signal path often miss phase noise contributions from power supply ripple and substrate coupling in integrated circuits. A comprehensive budget should include a "power supply noise" category, with contributions estimated from power supply rejection ratio (PSRR) specifications or measurements.

7. Advanced Techniques

7.1 Frequency Domain to Time Domain Conversion

While phase noise is a frequency-domain metric, system impacts often relate to timing jitter. The RMS phase jitter $\phi_{RMS}$ (in radians) can be calculated by integrating the phase noise over the relevant bandwidth:

$$ \phi_{RMS} = \sqrt{2 \cdot \int_{f_1}^{f_2} \mathcal{L}(f) \, df} $$

In a spreadsheet, this integration is approximated numerically using the trapezoidal rule across measurement points. The corresponding RMS time jitter is:

$$ t_{j,RMS} = \frac{\phi_{RMS}}{2\pi f_0} $$

7.2 Non-Linear Mixing Effects

When strong signals are present, non-linear mixing can cause reciprocal mixing, where phase noise from a strong signal degrades sensitivity to weak signals. The degradation in signal-to-noise ratio (SNR) due to reciprocal mixing is:

$$ \Delta SNR = 10 \cdot \log_{10}\left(1 + \frac{P_{strong}}{P_{weak}} \cdot 10^{\mathcal{L}(f_{offset})/10} \cdot B_{IF}\right) $$

where $B_{IF}$ is the intermediate frequency (IF) bandwidth. This effect should be included in receiver budgets operating in crowded spectral environments.

7.3 Monte Carlo Analysis for Yield Prediction

Extend the deterministic budget with statistical analysis to predict manufacturing yield. Assign probability distributions to component phase noise parameters (e.g., Gaussian distribution with mean and standard deviation from datasheet min-typ-max values). Run Monte Carlo simulations (easily implemented in Excel with Data Tables or Python integration) to generate histograms of aggregate phase noise and predict the percentage of units meeting specifications.

7.4 PLL Phase Noise Modeling

For accurate PLL phase noise budgets, use the following comprehensive model that accounts for all noise sources:

$$ \mathcal{L}_{PLL}(f_m) = \mathcal{L}_{ref}(f_m) + 20\log_{10}(N) + \mathcal{L}_{PFD}(f_m) + 20\log_{10}\left(\frac{f_{out}}{K_{PD}}\right) + \mathcal{L}_{VCO}(f_m) \cdot \left|1 - H(f_m)\right|^2 $$

where $H(f_m)$ is the closed-loop transfer function (low-pass for reference noise, high-pass for VCO noise), $K_{PD}$ is the phase detector gain in V/rad, and $\mathcal{L}_{PFD}$ is the PFD/charge pump noise floor.

8. Reference Tables and Formulas

8.1 Phase Noise Conversion Table

Table 3: Phase Noise Power Law Regions and Their Origins

RegionSlope (dB/decade)Physical OriginTypical Components
$f^0$ 0 White phase noise Frequency dividers, digital circuits
$f^{-1}$ -10 Flicker phase noise Amplifier 1/f noise
$f^{-2}$ -20 White frequency noise Thermal noise in resonators
$f^{-3}$ -30 Flicker frequency noise Oscillator flicker frequency noise
$f^{-4}$ -40 Random walk frequency noise Environmental effects, aging

8.2 Key Conversion Formulas

Single-Sideband to Double-Sideband Phase Noise:

$$ \mathcal{L}_{DSB}(f_m) = \mathcal{L}_{SSB}(f_m) + 3 \text{ dB} $$

Phase Noise to Phase Jitter:

$$ \phi_{RMS} = \sqrt{2 \cdot \sum_{i=1}^{n} \mathcal{L}(f_i) \cdot \Delta f_i} \quad \text{(numerical approximation)} $$

Frequency Multiplication:

$$ \mathcal{L}_{out}(f_m) = \mathcal{L}_{in}(f_m) + 20 \cdot \log_{10}(N) $$

Amplifier Additive Phase Noise:

$$ \mathcal{L}_{amp}(f_m) \approx 10 \cdot \log_{10}\left(\frac{k_B T F}{2 P_{in}}\right) \cdot \frac{1}{1 + (f_m/f_c)} $$

8.3 Typical Component Phase Noise Specifications

Table 4: Representative Phase Noise for Common Components at 1 GHz Carrier

Component100 Hz1 kHz10 kHz100 kHz1 MHz
OCXO (10 MHz) -100 -130 -150 -160 -165
TCXO (10 MHz) -80 -110 -130 -140 -145
Integer-N PLL -90 -105 -120 -130 -140
Fractional-N PLL -85 -100 -115 -125 -135
SAW Oscillator -70 -95 -115 -125 -130
Frequency Doubler +6 +6 +6 +6 +6
Frequency Divider (÷2) -6 -6 -6 -6 -6

8.4 Spreadsheet Implementation Tips

When implementing the budget in Excel or similar tools:

  • Use named ranges for key parameters to improve formula readability.
  • Implement the frequency conversion calculations in separate columns with clear labels.
  • Use conditional formatting to highlight margin violations (red for <0 dB, yellow for 0-3 dB, green for >3 dB).
  • Create charts showing phase noise profiles at each node overlaid with requirements.
  • Include a sensitivity analysis showing how variations in component specifications affect total performance.
  • Document all assumptions and data sources in a dedicated documentation tab.
This comprehensive methodology provides a robust framework for predicting and optimizing phase noise performance in complex systems. By following these steps and leveraging the provided formulas and examples, engineers can systematically allocate phase noise budgets, select appropriate components, and ensure compliance with system requirements while maintaining appropriate design margins.