Method and apparatus for power distribution using a multiphase voltage regulator with phase redundancy and fault tolerant operation
A fault-tolerant multiphase voltage regulator includes a plurality of power stages, each of which is configured to deliver a phase current to a processor, and a controller. The controller is configured to: control the plurality of power stages to regulate an output voltage provided to the processor; detect and disable a faulty power stage; generate a throttling signal to indicate if one or more of the power stages is faulty and disabled; and communicate the throttling signal to the processor over a physical line running between the processor and the controller. A corresponding method of operating a fault-tolerant power distribution system is also described.
1. A fault-tolerant multiphase voltage regulator, comprising:
a plurality of power stages, each of which is configured to deliver a phase current to a processor; and
a controller configured to:
control the plurality of power stages to regulate an output voltage provided to the processor;
detect and disable a faulty power stage;
continue to regulate the output voltage provided to the processor even if one or more power stages of the plurality of power stages is detected as being faulty and disabled;
generate a throttling signal to indicate if one or more of the power stages is faulty and disabled; and
communicate the throttling signal to the processor over a physical line running between the processor and the controller.
2. The fault-tolerant multiphase voltage regulator of claim 1 , wherein the controller is configured to communicate with a system manager independent of communicating with the processor, to indicate to the system manager if one or more power stages of the plurality of power stages is detected as being faulty and disabled.
3. The fault-tolerant multiphase voltage regulator of claim 1 , wherein the controller is configured to generate the throttling signal also in response to an overtemperature condition, so that the throttling signal is generated and communicated to the processor over the same physical line used by the controller to indicate to the processor if any of the following conditions occur:
a power stage is detected as being faulty and disabled; and
an overtemperature condition.
4. The fault-tolerant multiphase voltage regulator of claim 1 , wherein the controller is configured to generate the throttling signal also in response to an overvoltage condition, an undervoltage condition or an overcurrent condition, so that the throttling signal is generated and communicated to the processor over the same physical line used by the controller to indicate to the processor if any of the following conditions occur:
one or more of the plurality of power stages is detected as being faulty and disabled;
the overvoltage condition;
the undervoltage condition; and
the overcurrent condition.
5. The fault-tolerant multiphase voltage regulator of claim 1 , wherein the controller is configured to generate the throttling signal also in response to an overtemperature condition, an overvoltage condition, an undervoltage condition or an overcurrent condition, so that the throttling signal is generated and communicated to the processor over the same physical line used by the controller to indicate to the processor if any of the following conditions occur:
one or more of the plurality of power stages is detected as being faulty and disabled;
the overtemperature condition;
the overvoltage condition;
the undervoltage condition; and
the overcurrent condition.
6. The fault-tolerant multiphase voltage regulator of claim 1 , wherein the controller is configured to place the multiphase voltage regulator in a self-test mode in which the processor is operated at a known computational load and the controller operates each power stage independently to determine if any of the power stages is faulty under the known computational load.
7. A method of operating a fault-tolerant multiphase voltage regulator having a plurality of power stages, each of which is configured to deliver a phase current to a processor, the method comprising:
controlling the plurality of power stages to regulate an output voltage provided to the processor;
detecting and disabling a faulty power stage;
continuing to regulate the output voltage provided to the processor even if one or more power stages of the plurality of power stages is detected as being faulty and disabled;
generating a throttling signal to indicate if one or more of the power stages is faulty and disabled; and
communicating the throttling signal to the processor over a physical line running between the processor and the controller.
8. The method of claim 7 , further comprising:
communicating with a system manager independent of communicating with the processor, to indicate to the system manager if one or more power stages of the plurality of power stages is detected as being faulty and disabled.
9. The method of claim 7 , further comprising:
generating the throttling signal also in response to an overtemperature condition, so that the throttling signal is generated and communicated to the processor over the same physical line used to indicate to the processor if any of the following conditions occur:
a power stage is detected as being faulty and disabled; and
an overtemperature condition.
10. The method of claim 7 , further comprising:
generating the throttling signal also in response to an overvoltage condition, an undervoltage condition or an overcurrent condition, so that the throttling signal is generated and communicated to the processor over the same physical line used to indicate to the processor if any of the following conditions occur:
one or more of the plurality of power stages is detected as being faulty and disabled;
the overvoltage condition;
the undervoltage condition; and
the overcurrent condition.
11. The method of claim 7 , further comprising:
generating the throttling signal also in response to an overtemperature condition, an overvoltage condition, an undervoltage condition or an overcurrent condition, so that the throttling signal is generated and communicated to the processor over the same physical line used to indicate to the processor if any of the following conditions occur:
one or more of the plurality of power stages is detected as being faulty and disabled;
the overtemperature condition;
the overvoltage condition;
the undervoltage condition; and
the overcurrent condition.
12. The method of claim 7 , further comprising:
placing the multiphase voltage regulator in a self-test mode in which the processor is operated at a known computational load and each power stage is operated independently to determine if any of the power stages is faulty under the known computational load.
13. A fault-tolerant power distribution system, comprising:
a processor;
a multiphase voltage regulator comprising a plurality of power stages, each of which is configured to deliver a phase current to the processor, and a controller configured to:
control the plurality of power stages to regulate an output voltage provided to the processor;
detect and disable a faulty power stage;
continue to regulate the output voltage provided to the processor even if one or more power stages of the plurality of power stages is detected as being faulty and disabled;
generate a throttling signal to indicate if one or more of the power stages is faulty and disabled; and
communicate the throttling signal to the processor over a physical line running between the processor and the controller; and
a system manager in independent communication with the multiphase voltage regulator and the processor.
14. The fault-tolerant power distribution system of claim 13 , wherein the system manager is configured to reduce a computational load of the processor if the controller of the multiphase voltage regulator indicates one or more power stages of the plurality of power stages is detected as being faulty and disabled.
15. The fault-tolerant power distribution system of claim 14 , wherein the system manager is configured to increase the computational load of the processor after the multiphase voltage regulator reports a faulty power stage, shuts down and repowers up again.
16. The fault-tolerant power distribution system of claim 14 , wherein the system manager is configured to direct the multiphase voltage regulator to perform a self-test mode to confirm whether any of the power stages are faulty.
17. The fault-tolerant power distribution system of claim 16 , wherein the system manager is configured to increase the computational load of the processor if the multiphase voltage regulator indicates the self-test mode confirmed none of the power stages are faulty.
18. The fault-tolerant power distribution system of claim 16 , wherein the system manager is configured to shut down the processor and the multiphase voltage regulator if the multiphase voltage regulator indicates the self-test mode confirmed that one or more of the power stages is faulty.
19. A method of operating a fault-tolerant power distribution system having a processor, a multiphase voltage regulator comprising a plurality of power stages, each of which is configured to deliver a phase current to the processor, and a system manager in independent communication with the multiphase voltage regulator and the processor, the method comprising:
controlling the plurality of power stages at the multiphase voltage regulator to regulate an output voltage provided to the processor;
detecting and disabling a faulty power stage at the multiphase voltage regulator;
continuing to regulate the output voltage provided to the processor even if one or more power stages of the plurality of power stages is detected as being faulty and disabled;
generating a throttling signal at the multiphase voltage regulator to indicate if one or more of the power stages is faulty and disabled;
communicating the throttling signal from the multiphase voltage regulator to the processor over a physical line running between the processor and the controller; and
managing a computational load of the processor by the system manager based on feedback from the processor and the multiphase voltage regulator.
20. The method of claim 19 , wherein managing the computational load of the processor by the system manager comprises:
reducing the computational load of the processor if the multiphase voltage regulator indicates one or more power stages of the plurality of power stages is detected as being faulty and disabled.
21. The method of claim 20 , further comprising:
increasing, via the system manager, the computational load of the processor after the multiphase voltage regulator reports a faulty power stage, shuts down and repowers up again.
22. The method of claim 20 , further comprising:
directing, via the system manager, the multiphase voltage regulator to perform a self-test mode to confirm whether any of the power stages are faulty.
23. The method of claim 22 , further comprising:
increasing, via the system manager, the computational load of the processor if the multiphase voltage regulator indicates the self-test mode confirmed none of the power stages are faulty.
24. The method of claim 22 , further comprising:
shutting down the processor and the multiphase voltage regulator, via the system manager, if the multiphase voltage regulator indicates the self-test mode confirmed that one or more of the power stages is faulty.