Randomized Thermometric Encoding for High-Speed DAC Mismatch
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed digital-to-analog converters (DACs) face dynamic range reduction due to static mismatch, such as timing offsets, which are exacerbated by fabrication variations and temperature gradients, particularly in giga-samples-per-second range DACs, leading to unwanted spurious content affecting the bandwidth of interest.
Innovation Solution
The implementation of a randomizing thermometric encoder that distributes mismatch among multiple codes by randomizing the selection of current sources, varying packet width, and swapping pairs of values to mitigate DC linearity errors and switch timing offsets, thereby improving DAC performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If high-speed DAC operation is implemented, then conversion speed is improved, but dynamic range is reduced due to static mismatch
Solution Approach 1:
The patent applies dynamics by making the encoder output sequence variable and randomized rather than fixed. The randomizing encoder dynamically shuffles the thermometric codes so that the same digital input value can produce different output sequences at different times, which randomizes the impact of static mismatch and timing offsets, thereby maintaining high conversion speed while improving dynamic range.
Solution Approach 2:
The patent changes the parameter of code assignment by using randomization to vary which current sources are activated for a given digital input. Instead of a deterministic mapping, the system randomly selects among equivalent thermometric codes, changing the temporal distribution of mismatch effects and reducing spurious content in the frequency band of interest.
2Device complexity
If conventional thermometric encoding is used, then encoding simplicity is maintained, but spurious content is generated due to static mismatch
Solution Approach 1:
The patent introduces dynamics into the encoding process by implementing a randomizing encoder that varies the output code sequence over time. This dynamic randomization approach maintains relatively simple encoder architecture while effectively reducing spurious content generated by static mismatch and timing offsets through statistical averaging.
Solution Approach 2:
The patent converts the harmful effect of static mismatch into a beneficial statistical averaging effect. By randomizing the code assignment, the deterministic spurious tones generated by mismatch are transformed into random noise that can be filtered more effectively, and the mismatch errors are distributed and averaged out over time, turning a harmful deterministic effect into a beneficial statistical phenomenon.
3Stability of the object's composition
If fixed code assignment is used, then encoding consistency is maintained, but mismatch effects are concentrated
Solution Approach 1:
The patent replaces fixed code assignment with dynamic randomization, where the encoder output sequence varies over time even for the same digital input. This dynamic approach distributes mismatch effects across multiple different code patterns, preventing concentration of errors and improving measurement precision through statistical averaging while maintaining overall system stability.
Data Source
AI summary
Techniques for compensating high-speed digital-to-analog converters (DACs) for static mismatch are described. In ideal circumstances, the current sources of a DAC are identical to each other, leading to a frequency response presenting a relatively flat noise spectrum. In the presence of mismatch, however, the response creates unwanted spurious content, which can negatively affect the DAC's dynamic range. The techniques described herein involve randomized thermometric encoders. First, the direction in which a packet contracts or expands, depending on the value to be encoded, can be randomized. Second, pairs of values in a packet (and/or pairs of values outside the packet) can be swapped with one another in a randomized fashion. Third, the decision of whether to apply randomization or not can itself be randomized. By applying one or more of the randomization techniques described herein, the negative effects of switch timing offset and errors in DC linearity can be mitigated.