Browse Prior Art Database

Fan Control Algorithm to Proactively Increase Fan Speeds in Electronics Cooling Based on Rate of Temperature Change

IP.com Disclosure Number: IPCOM000218904D
Publication Date: 2012-Jun-08
Document File: 3 page(s) / 54K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method of cooling hardware systems that predicts temperature alerts before they actually occur, allowing the system fan speed control algorithm to proactively increase the fan speeds. This behavior reduces both over-temperature conditions as well as fan overshoot and increases the overall cooling efficiency of the system.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 45% of the total text.

Page 01 of 3

Fan Control Algorithm to Proactively Increase Fan Speeds in Electronics Cooling Based on Rate of Temperature Change

Electronics hardware often relies on fans for system cooling. Software controls are typically used to change the fan speed in response to temperature alerts generated by various system components such as Central Processing Units (CPUs), Dual In-Line Memory Modules (DIMMs), secondary chips, and option cards. The controls must provide adequate cooling for the worst case conditions (e.g., the system ramping suddenly from idle to maximum utilization), and as a result, it does not maximize the power efficiency.

Two issues make this traditional behavior inefficient. First, often the fan speeds are unable to respond quickly enough to prevent the temperature of the component from rising too rapidly; the component temperature may not be adequately controlled to prevent an over-temperature situation from occurring. Second, the fans often "overshoot", meaning that the fans ramp to a value well above the required fan state to adequately cool the heated components. Since component temperatures are transient in nature (related to system usage, ambient temperature, etc.), the system fan speed control behavior typically becomes highly oscillatory. Additionally, the fan speed intervals may have to be large enough that they are acoustically undesirable.

This proposed method of cooling predicts temperature alerts before they actually occur, allowing the system fan speed control algorithm to proactively increase the fan speeds. This behavior reduces both over-temperature conditions as well as fan overshoot and increases the overall cooling efficiency of the system.

Temperature data from the critical components are collected at a particular rate (e.g., once per second). The current value is then be compared to the previous value over a specified time period (delta temp/delta time); this could be referred to as the slope (i.e., rate of change) of the line created by the temperature-time curve. If the rate of change is greater than a desired value (e.g., 3degC per 10 seconds), then the fan speed control algorithm causes the fans to increase by one fan speed step. By proactively raising the fan speed in anticipation of a thermal event, this reduces the amount of steps (and time) necessary to elapse once the temperature alert is asserted, thus minimizing the over-temperature and overshoot occurrences. This function can be applied to each component individually.

Algorithm for Sensor-Based Fan Speed Control

STEP 1


Enter fan table. Refer to this starting point as n. Proceed to STEP 2

STEP 2


Sample all sensor values for a period of 2 minutes(E3h) every 1 second IF T

(E8h at any time,

THEN set T

               transient, CPU1 = Tsensor, CPU1, set Ttransient, CPU2 = Tsensor, CPU2, increase fan speed n+1, set iterateCPU1=0, wait 10 seconds(E5h), and go to STEP 3

sensor, CPU1 > Tlimit, CPU1

1


Page 02 of 3

ELSE IF T

sensor, CPU2 > Tlimit, CPU2

(E9h) at any time, T...