Thermal Sensor Curvature Correction Using Sigma-Delta Reference Sampling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current temperature sensing circuits in Systems on a Chip (SOCs) face inaccuracies due to non-linearity in the reference voltage generated, which affects the accuracy of temperature readings, and existing designs fail to adequately correct for this non-linearity.
Innovation Solution
A temperature sensing circuit utilizing a sigma-delta modulated analog to digital converter that includes a switched capacitor circuit and a quantization circuit to selectively sample and integrate voltages, canceling out non-linear curvature in the voltage complementary to absolute temperature, thereby producing a temperature-independent reference voltage with negligible non-linearity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a voltage complementary to absolute temperature (Vctat) is generated using a bipolar junction transistor, then the reference voltage can be produced, but non-linear curvature is introduced due to process-dependent slope variations
Solution Approach 1:
The patent segments the Vctat generation into multiple bipolar junction transistors with different current density ratios. By dividing the single-transistor approach into multiple transistors (e.g., Q1-Q4), each contributing a portion of the complementary voltage, the circuit achieves better linearity. The segmented approach allows the non-linearities of individual transistors to cancel each other out when their outputs are combined, resolving the contradiction between achieving reliable Vctat and maintaining linearity.
Solution Approach 2:
The patent applies local quality by giving each bipolar junction transistor a specific current density ratio tailored to its position in the circuit. Each transistor is biased at a different current density (e.g., Q1 at ratio p1, Q2 at ratio p2, etc.), creating localized optimizations that when combined produce an overall linear Vctat. This local differentiation allows precise control over the non-linear characteristics of each segment to achieve global linearity.
2Reliability
If operational amplifiers are used to add compensation voltage for curvature correction, then non-linearity can be reduced, but offset errors are introduced
Solution Approach 1:
The patent extracts and eliminates the operational amplifier from the compensation voltage addition process. Instead of using an op-amp to sum the Vctat components and compensation voltage, the circuit directly combines the outputs of multiple bipolar junction transistors through passive summing nodes. This removal of the active op-amp component eliminates the source of offset errors while maintaining the ability to achieve linearity through the transistor segmentation approach.
Solution Approach 2:
The patent introduces passive summing nodes and current mirror circuits as intermediaries to combine the Vctat components without using an operational amplifier. These intermediary elements serve as mediators that add the voltages from multiple transistors while introducing no active offset errors. The current mirrors and summing nodes provide a clean, passive addition mechanism that preserves measurement precision while achieving the desired linearity correction.
3Ease of manufacture
If a single-trim CMOS bandgap reference is used, then manufacturing simplicity is improved, but non-linearity correction remains insufficient
Solution Approach 1:
The patent employs a composite approach by combining multiple bipolar junction transistors with different current density ratios in a unified Vctat generation structure. This composite transistor architecture, integrated within a CMOS process, creates a reference voltage that benefits from both the manufacturing simplicity of standard CMOS and the enhanced temperature independence achieved through the multi-transistor composite structure. The composite design allows standard fabrication processes to produce a superior reference voltage with better linearity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution achieves highly accurate temperature readings by effectively canceling out non-linear errors in the reference voltage, improving the precision of temperature sensing in SOC applications.
Implementation Method 1
A temperature sensing circuit utilizing a sigma-delta modulated analog to digital converter that includes a switched capacitor circuit and a quantization circuit
Implementation Method 2
a switched capacitor circuit configured to selectively sample voltages produced by the voltage generation circuitry and provide the sampled voltages to inputs of an integrator
Implementation Method 3
cooperate with the integrator under control of the bitstream to: when a most recent bit of the bitstream is a logic zero, cause integration of a difference between a base-emitter voltage of the first bipolar junction transistor and a base-emitter voltage of the second bipolar junction transistor
Implementation Method 4
A voltage proportional to absolute temperature Vptat can be produced as the difference between the base-emitter junction voltages of two bipolar junction transistors biased at different current densities
Data Source
AI summary
A method includes generating a voltage proportional to absolute temperature, generating an uncorrected voltage complementary to absolute temperature, and generating a correction voltage. The method further includes selectively sampling the voltage proportional to absolute temperature, the uncorrected voltage complementary to absolute temperature, and the correction voltage, providing those sampled voltages to inputs of an integrator, and then quantizing outputs of the integrator to produce a bitstream. The method continues with causing the integrator to integrate the voltage proportional to absolute temperature or causing the integrator to add the correction voltage to the uncorrected voltage complementary to absolute temperature to produce a corrected voltage complementary to absolute temperature and then integrate the corrected voltage complementary to absolute temperature, depending upon a most recent bit of the bitstream. The bitstream is filtered and decimated to produce a voltage indicative of a temperature of a chip on which the method is performed.