ADR Shutdown Timing for AC Undervoltage Ride-Through
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As compute systems grow in size and power consumption, conventional approaches to maintaining a 2 ms hold-up time for asynchronous dynamic random access memory (ADR) refresh become inadequate, requiring at least 6 ms to prevent data loss during shutdown, and immediate reaction to AC power loss can lead to unnecessary shutdowns if power is restored.
Innovation Solution
A dynamic timing technique that detects AC undervoltage conditions and dynamically determines the hold-up time based on the present load, allowing for a monitoring period before initiating shutdown processes like ADR, ensuring data integrity and preventing premature system shutdown.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system immediately reacts to AC power loss by initiating shutdown, then data loss is prevented, but unnecessary shutdowns occur when power is temporarily undervoltage and then restored
Solution Approach 1:
The system performs preliminary monitoring of AC undervoltage conditions during a determined monitoring period before initiating shutdown. This preliminary action allows the system to distinguish between temporary brown-out events (where power is restored) and permanent power failures, preventing unnecessary shutdowns while maintaining data integrity through the ADR refresh process when needed.
2Productivity
If the system waits for extended monitoring period before shutdown, then unnecessary shutdowns are prevented, but data loss risk increases during the waiting period
Solution Approach 1:
The system performs the ADR refresh operation in advance during the monitoring period, transferring data from volatile memory to non-volatile memory before shutdown is initiated. This preliminary data protection action ensures that if shutdown occurs, data integrity is maintained, while the monitoring period itself prevents premature shutdown during temporary brown-outs.
Solution Approach 2:
The monitoring period is dynamically determined based on system state and power supply characteristics. This dynamic adjustment allows the system to optimize between preventing unnecessary shutdowns and ensuring data protection, adapting the waiting period to current operational conditions rather than using a fixed timeout value.
3Reliability
If the hold-up time is increased from 2 ms to 6 ms to support ADR refresh, then data protection is improved, but power supply requirements and system complexity increase
Solution Approach 1:
The system performs ADR refresh as a preliminary action before shutdown is initiated, ensuring data is transferred to non-volatile memory while power is still available. This approach allows adequate hold-up time for the refresh operation without requiring the power supply to maintain voltage throughout the entire monitoring period, reducing power supply complexity.
Solution Approach 2:
The system uses non-volatile memory as an intermediary storage medium to temporarily hold data during the ADR refresh process. This intermediary allows data to be protected during shutdown without requiring extended hold-up time from the power supply, as the refresh operation completes before power is lost.
4Reliability
If the system performs ADR refresh during brown-out events, then data corruption is prevented, but the system may not have sufficient hold-up time to complete the refresh
Solution Approach 1:
The system detects AC undervoltage conditions and initiates ADR refresh as a preliminary action while power is still sufficient to complete the operation. The monitoring period is calculated to ensure the refresh can complete before power is lost, creating a time buffer that guarantees data protection without requiring excessive hold-up time.
Solution Approach 2:
The system continuously monitors power supply status and uses this feedback to determine whether to initiate shutdown or continue operation. This feedback mechanism allows the system to adapt to actual power conditions, initiating ADR refresh only when there is sufficient time to complete the operation before power failure, optimizing the use of available hold-up time.
Data Source
AI summary
A technique for managing undervoltage in a compute system is disclosed. The technique includes a method that further includes: detecting an AC undervoltage condition in the compute system; and upon detecting the AC undervoltage condition: dynamically determining a holdup time as a function of the present load; determining a monitoring period as a function of the dynamically determined holdup time; waiting for the determined monitoring period to expire; and upon expiration of the determined monitoring period, perform a shutdown process if the AC undervoltage condition persists.