RF DCO Varactor Bank Segmentation for Wider Tuning Resolution

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The challenge in modern RF CMOS design is the difficulty in achieving high fidelity analog circuits due to conflicting requirements of area, power, and performance, exacerbated by the inability of SPICE models to accurately simulate the RF operational environment, leading to costly design iterations and poor characterization of varactors in digitally controlled oscillators (DCOs), which affects the frequency tuning and modulation resolution of RF transceivers.

Innovation Solution

A novel method is introduced to extend the frequency tuning range and improve modulation resolution of RF DCOs by dividing the varactor bank into MSB, LSB, and fractional sigma-delta banks, using a pre-distortion scheme to digitally calibrate mismatches and a harmonic calibration technique to determine average mismatch values, allowing for efficient characterization and compensation of mismatches in the DCO.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If a single unified tuning bank is divided into MSB and LSB banks for frequency tuning, then the frequency tuning range is extended and modulation resolution is improved, but device complexity increases due to additional banks and calibration requirements

Engineering Contradiction:
Improvefrequency tuning precisionVSAvoidtuning bank structure
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The unified tuning bank is segmented into MSB (Most Significant Bit) and LSB (Least Significant Bit) banks, allowing independent control and calibration of each segment. This segmentation enables precise frequency tuning by combining coarse adjustments from MSB banks with fine adjustments from LSB banks, resolving the contradiction between tuning precision and device complexity through modular architecture.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

A pre-distortion scheme is implemented to pre-calibrate mismatch errors between MSB and LSB banks before actual frequency tuning operations. By performing calibration in advance and storing correction values in lookup tables, the system eliminates the need for complex real-time calibration circuitry, thereby improving tuning precision while managing device complexity.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If pre-distortion scheme is used to compensate mismatches, then modulation resolution is enhanced, but device complexity increases due to additional calibration circuits and lookup tables

Engineering Contradiction:
Improvemodulation resolutionVSAvoidcalibration circuitry
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

Mismatch compensation values are captured during calibration and stored in lookup tables (LUTs), creating a digital copy of the correction data. During normal operation, these pre-stored values are retrieved and applied without requiring complex real-time calibration circuits, thereby enhancing modulation resolution while minimizing additional hardware complexity.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The pre-distortion calibration is performed once during device initialization or manufacturing, and the resulting correction values are stored for reuse. This preliminary action eliminates the need for continuous complex calibration operations, achieving high modulation resolution with minimal ongoing computational or hardware overhead.

Inventive Principle:
Principle #10Preliminary action

3Area of stationary object

If varactor banks are made smaller to reduce area, then area requirement is reduced, but measurement precision deteriorates due to difficulty in characterizing small varactors

Engineering Contradiction:
Improvevaractor bank areaVSAvoidvaractor characterization accuracy
Core Design Contradiction:
Area of stationary objectVSMeasurement precision

Solution Approach 1:

The system performs self-calibration by measuring its own mismatch errors during initialization and storing correction values in lookup tables. This self-service approach allows accurate characterization of small varactors without requiring external precision measurement equipment, enabling area reduction while maintaining measurement precision through automated self-testing.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

Varactor mismatch characteristics are pre-measured and stored during manufacturing or initialization, enabling accurate compensation for small varactor imperfections. This preliminary characterization allows the use of smaller varactors with reduced area while maintaining measurement precision through pre-captured correction data that compensates for manufacturing variations.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS7439817B2Frequency tuning range extension and modulation resolution enhancement of a digitally controlled oscillator
Publication Date: 2008.10.21 TEXAS INSTRUMENTS INC
  • US7439817B2 patent drawing
  • US7439817B2 patent drawing
  • US7439817B2 patent drawing

AI summary

A novel apparatus and method of extending the frequency tuning range and improving the modulation resolution of an RF digitally controlled oscillator (DCO). In addition to the coarse PVT MIM varactor bank, the DCO uses a single unified bank of varactors that is further subdivided divided into an MSB bank, LSB bank and sigma-delta (SD-LSB) bank. Any ratio mismatches between MSBs and LSBs are digitally calibrated out using a DCO step-size pre-distortion scheme wherein the LSB steps are adjusted to account for the ratio mismatch between the MSB/LSB step sizes. A harmonic characterization technique is used to estimate the mismatches in the minimal size CMOS tuning varactors of a digitally controlled RF oscillator (DCO), wherein the nominal ratio mismatch between the MSB and LSB devices is estimated using hybrid stochastic gradient DCO gain estimation algorithms. The nominal ratio mismatch and the mismatches in the MSB and LSB banks are used to determine the average MSB/LSB mismatch. The average mismatch value is then used to correct the LSB steps.