Multiphase Power Converter Phase Shedding Using Load Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional multiphase DC-DC power converters face instability and inefficiency due to inappropriate phase shedding strategies, particularly when dealing with rapid output current variations and high slew rates, which can lead to suboptimal performance and potential system instability.
Innovation Solution
A novel multiphase power converter system that employs a machine-learning process to dynamically determine the number of active phases based on real-time input and output parameters, using a predictor to adjust phase statuses and optimize operation, ensuring efficient energy supply even under varying load conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If auto-phasing is used to reduce the number of active phases based on output current thresholds, then power savings are improved, but system stability deteriorates due to inappropriate phase dropping during rapid current variations
Solution Approach 1:
The predictor determines future output current values in advance before they actually occur, allowing the phase shedding decision to be based on predicted rather than historical data. This preliminary action prevents inappropriate phase dropping during rapid current transitions while still enabling power savings during steady-state low-current conditions.
Solution Approach 2:
The system dynamically adapts the number of active phases based on predicted future load conditions rather than fixed thresholds or historical data. The regulator continuously adjusts phase configuration in response to predicted current requirements, enabling both power savings and stability through dynamic optimization.
2Reliability
If re-balancing time delay is implemented to prevent phase dropping for short durations, then phase stability is improved, but the responsiveness to load changes deteriorates
Solution Approach 1:
By predicting future output current values before they occur, the system eliminates the need for re-balancing time delays. Phase shedding decisions are based on anticipated load requirements, allowing immediate response to genuine load changes while filtering out transient fluctuations that would otherwise require delay-based protection.
3Speed
If the number of active phases is increased to meet large slew rate requirements, then output current performance is improved, but power consumption increases
Solution Approach 1:
The system dynamically determines the optimal number of active phases based on predicted future load requirements. During transient conditions requiring high slew rates, the predictor anticipates the need for additional phases and activates them in advance. During steady-state low-current operation, the system reduces the number of active phases to minimize power consumption, achieving both performance and efficiency.
Data Source
AI summary
A multiphase power converter comprises a regulator, a value-supply system arranged for collecting at least one operating point of the power converter, and a predictor for determining updated phase statuses, for activating or deactivating each of the phases (111, 112, 113, . . . ) during a further operation of the power converter. The updated phase statuses are determined using a process based on the at least one collected operating point and predictor parameters obtained from a machine-learning process.