Lenovo ThinkPad R50p Hard Drive Active Protection system white paper - Page 4
unloading, increased, drive, shock, tolerance, Prediction, Algorithm - ibm drivers
View all Lenovo ThinkPad R50p manuals
Add to My Manuals
Save this manual to your list of manuals |
Page 4 highlights
Head unloading for increased hard disk drive shock tolerance As a result of extensive testing with the computer lid both open and closed, IBM has found that in typical drop accidents either from the knee or a desk to a concrete floor, the internal hard disk drive is protected by the durable ThinkPad shell so that the received shock level to the hard disk drive generally does not exceed 800G or 800 times the force of gravity. Since 800G is the maximum rating of shock tolerance of the hard disk drive with the head unloaded, the hard disk drive can be better protected during most of these common accidents in typical usage if the disk head is unloaded prior to receiving the shock impact. Head unloading can take 300 to 500ms, and the average duration after detecting a free-fall situation is 300ms. Therefore, it is too late to start head unloading after detecting the free-fall situation. This makes it necessary to concentrate on the behavior of the machine prior to a drop. The initial behavior acts as a prediction of the impending accident. In general, the duration of the predicting event is much longer than the actual free-fall. This enables the active protection system to respond to the impending accident prior to receiving the shock. The machine posture or system orientation is constantly checked by monitoring acceleration and using a prediction algorithm to statistically analyze recent behavior in typical usage. Head unloading is initiated by degrees of tilt, acceleration, or shock. The degrees at which head unloading occurs are adjusted and fine-tuned every time a proper prediction is applied for the current motion. The disk drive head remains unloaded until detecting a stable environment for approximately one to four seconds depending upon the amount of excessive acceleration at the beginning of the current shock or posture change. Use of Prediction Algorithm The active protection system employs a heuristic learning algorithm to track system orientation. The Shock Manager, a system thread created by the Shockprf.sys of a kernel mode device driver, analyzes the variations in acceleration and collects system orientation data into the Shock History Database. The collected data is then used to tune sensitivity and predict excessive shocks. This tuning is important in minimizing the disk performance penalty caused by prediction failures. Once the Shock Manager detects a certain variation which may be equivalent to one usually seen just before receiving an excessive shock, the Shock Manager acts immediately to stop the hard disk drive. The sensitivity to predict a shock varies depending on usage conditions. In general, the more stable the ThinkPad computer stays, the more sensitively the Shock Manager behaves. The Shock Manager focuses its attention on the current acceleration variation and the weighted average in the recent past. The log is used to minimize intrusive operation during normal usage of the computer. If the hard disk drive head is unloaded after detecting a potentially harmful situation, the user can shorten the current head-unloading period up to 1 second on the condition that a calm state, stable enough to remove the risk of a drop, continues for at least one second and at the same time mouse movement activity for more than 200ms is detected thereby informing the active protection system that this is not a drop situation. The design of the active protection system allows for certain shocks or vibrations that fall within a span of accepted or normal motion and does not repeatedly stop the hard drive when this level of motion is detected. You can select a checkbox in the properties window to temporarily disable hard drive protection while repetitive motion or vibration is detected. This feature is particularly useful when 2 IBM Active Protection System Whitepaper