Sensorless Overcurrent Prediction in Isolated Bidirectional DC-DC Converters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional isolated bidirectional DC-DC converters face challenges in accurately detecting overcurrents due to fluctuating inductor current peak values, leading to potential damage to semiconductor switches, as current sensors struggle to sense the peak values within the required sampling period.
Innovation Solution
A sensorless prediction method models the inductor current waveform using input values such as primary side voltage peak, switching frequencies, internal inductance, and phase differences to calculate average current values and predict overcurrents without the need for current sensors, employing PI control to update and refine the current values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current sensors are installed to detect inductor current, then overcurrent detection capability is improved, but measurement precision deteriorates due to fluctuating peak values that cannot be sensed within the sampling period
Solution Approach 1:
The patent performs preliminary action by calculating and predicting the peak inductor current value using a prediction equation that incorporates switching frequency, duty cycle, and voltage parameters before the actual current flow occurs. This allows the system to anticipate overcurrent conditions and prepare protective actions in advance, rather than attempting to measure the rapidly fluctuating peak current during the sampling period.
Solution Approach 2:
The patent introduces an intermediary approach by using a prediction calculation model as a mediator between the control system and the actual current measurement. Instead of directly measuring the difficult-to-capture peak current, the system uses an intermediate prediction equation that computes the expected peak current based on known parameters, thereby indirectly obtaining accurate current information without direct sensing challenges.
2Measurement precision
If sampling frequency is increased to capture peak current values, then measurement precision is improved, but device complexity and cost increase due to higher requirements for current sensors and processing speed
Solution Approach 1:
The patent replaces the mechanical/electrical measurement system (current sensors and high-speed sampling circuits) with a computational prediction system. Instead of using physical sensors to directly measure the peak current, the system substitutes a mathematical prediction model that calculates the expected peak current based on electrical parameters, thereby eliminating the need for high-frequency sensing hardware and complex processing circuits.
Solution Approach 2:
The patent creates a computational copy or model of the actual current waveform through prediction equations. Rather than physically measuring the current, the system generates a predicted current profile that mirrors the expected behavior, allowing overcurrent detection without requiring direct physical measurement of the rapidly changing current peaks.
3Object-affected harmful factors
If conventional current detection methods are used, then overcurrent protection is provided, but reliability deteriorates due to failure to detect peak values, causing semiconductor switch damage
Solution Approach 1:
The patent applies preliminary anti-action by calculating the predicted peak current and comparing it against safe operating limits before the actual current peak occurs. If the predicted value exceeds the threshold, the system takes preventive action (such as adjusting duty cycle or triggering protection) in advance, thereby counteracting the potential harmful effect before it can damage the semiconductor switches.
Solution Approach 2:
The patent implements feedback by continuously monitoring operating parameters (switching frequency, duty cycle, voltages) and using them in the prediction equation to calculate the expected peak current. This feedback loop allows the system to adapt the overcurrent protection threshold dynamically based on actual operating conditions, improving both protection accuracy and switch safety.
Data Source
AI summary
Proposed is a method for accurately predicting an overcurrent flowing inside an isolated bidirectional DC-DC converter even without using a current sensor on primary and secondary sides of a transformer. In the converter according to the present disclosure, an average value of the inductor current is calculated after deriving inflection point current values by respectively modeling a current waveform for an inductor current of the transformer. A secondary side output current average value is calculated by comparing the calculated average value of the inductor current with a secondary side capacitor current average value of the converter at no load. Next, an error between the secondary side output current average value and an actually measured secondary side output current is calculated, and the inflection point current values of the current waveform are updated using a gain for reducing the error through PI control, whereby the overcurrent may be predicted.