On-Chip ADC Linearity Testing With a Moving Histogram

The patent divides the full ADC code range into multiple segments or windows. Instead of processing all codes simultaneously, the histogram is computed for a limited window of codes at a time, then shifted to process adjacent code ranges. This segmentation reduces the computational burden per processing cycle while maintaining comprehensive coverage of the entire ADC transfer function for accurate DNL and INL measurements.

The patent employs a dynamic sliding window approach where the histogram processing window moves across the ADC code range. The window position is dynamically adjusted to process different code segments sequentially. This dynamic approach allows the system to maintain a small, fixed-size histogram buffer while still evaluating linearity parameters across the full code range, significantly reducing memory requirements and processing time compared to static all-code processing.

The patent segments the ADC code range into smaller windows, allowing the histogram to store counts for only a limited number of codes at a time rather than all possible codes. This segmentation enables the use of small, fixed-size memory buffers while still achieving comprehensive linearity assessment through multiple sequential measurements across different code ranges.

The sliding window mechanism dynamically repositions the histogram buffer across different code ranges. By moving the window through the full code space and accumulating results, the system achieves complete linearity characterization using minimal memory. The dynamic approach allows reuse of the same small buffer for different code segments, eliminating the need for large static storage.

The patent processes ADC codes in segmented windows rather than transferring and processing all codes at once. Each window contains a manageable subset of codes that can be processed locally with minimal data transfer. This segmentation reduces the volume of data that needs to be transferred between the ADC, processor, and memory, thereby reducing data transfer time while maintaining measurement accuracy through comprehensive coverage of all code ranges.

The dynamic sliding window approach enables incremental processing of code ranges. Instead of requiring complete data collection and transfer before processing, the system can dynamically shift the processing window and begin computations on each segment as data becomes available. This reduces the peak data transfer requirements and allows overlapping of data acquisition with processing operations, minimizing total test time.