PMBusTakes Command of Data Center Power Issues

Apr 1, 2008 12:00 PM
By Brian Griffith, Server Power Delivery Architect, Intel, DuPont, Wash.

Thermal Management

PMBus has commands that are useful for server system thermal management. These include fan monitoring, fan control and temperature sensing. In some cases, fans inside power supplies often cool other components in the system. In order to optimize the power consumed by these fans, the system needs a means to control the fans based on the temperature of the system components they are cooling.

FAN_COMMAND_1 is one of four commands used to control up to four fans in a power supply. In other systems, fans in the system cool the power supply. In these cases, the system needs to know the power supply's temperature to allow for optimized power-supply cooling and lower system fan power. READ_TEMPERTURE_1 is one of three commands used to read temperatures inside the power supply. In both sets of commands, PMBus can be used to help reduce power consumed by system fans or power-supply fans.

Diagnostics

PMBus also has fault-related commands to standardize the diagnosis of power-supply problems in the system. Power supplies are among the highest-failure-rate components in the data center. The more information available on these power supplies, the better a user can determine the root cause of a failure and verify that it is an actual power-supply problem. PMBus also has warning-related commands to head off power-supply problems before the system is affected by a fault.

The STATUS_WORD command is used by the system as the first level of diagnostic information containing a high-level summary of all fault and warning conditions. This command is the first to be used by the system to quickly read the status of the power supply. If more detailed information is needed by the system, the second level of status commands are used such as STATUS_INPUT and STATUS_TEMPERATURE. The sensors used in the power supplies affected by the various status commands are:

• Input voltage

  This provides warnings if the input voltage drops too low and provides a fault if the input voltage drops lower than the minimum operating point.

• Output voltage

  This provides a warning and a fault if the output voltage is sensed to be out of operating range.

• Temperature

  This provides a warning if the temperature is about to exceed its maximum operating point and provides a fault if an excessive temperature causes the power supply to shut down.

• Fan speed

  This provides a warning if a fan is wearing out and running too slow and provides a fault if the slow-running fan has caused the power supply to shut down.

• Output current

  This provides a warning if the system loading is at the maximum rating of a power supply and provides a fault if the system has overloaded the power supply causing it to shut down.

New Features and Higher Accuracy

As servers add new power-management capabilities, PMBus will play an important role in these new capabilities. The monitoring of real power consumption at the rack, system and subsystem levels is beginning to be used to help make data centers more efficient, increase data center densities and calculate system energy consumption for charge back. PMBus will help make these new features standard capabilities of all server power supplies and allow for improvements in capabilities such as accuracy.

References

1. Advance Configuration and Power Interface specification, www.acpi.info.

2. System management bus specification, http://smbus.org/specs.

Click here for the enhanced PDF version of this article

Acceptable Use Policy blog comments powered by Disqus