Recently someone interfaced to the sensors in an Apple laptop (slashdot discussion yesterday) and it appears that there are two gyros and one accelerometer. The gyros sense rotation of the laptop side to side and front to back (don't care about spinning the laptop in a circle). The accelerometer is in the Z (down) direction as the laptop sits on a table. I suspect that when a laptop falls the rotation takes more than a few hundred mS as the unit slides off the surface (hand, table, armrest, etc). There should be plenty of time to stop the heads if you sense a change in angle of more than 10 degrees or so in a short period of time. Discussions about it seem to indicate that it's very agressive, and will park the heads when the user picks the laptop up from a table, some indicate that if they hit the table next to the laptop hard then it parks the heads. They change the sensitivity so they get better performance since it seems to be parking the heads when they shouldn't. But I suspect the roll rate is more easily detected first rather than 0g. Further, laptop hard drives designed for this should be able to park their heads significantly faster than 500mS. All you have to do is rail the servo with a transister. The head is tiny and light enough, and the servo powerful enough (it has to be for fast access times) that it should be much faster than 1/2 a second. I imagine you could test this with an old bad HD by hooking an appropiately amplified square wave to the servo mechanism and slamming the head from side to side - increase the frequency until it no longer touches both limits. I'd be surprised if it couldn't manage greater than 10Hz. Might be easier than that, though. The maximum seek time is the time it takes to travel from the edge of the platter to the center and rotate the disk one full revolution. Average seek time is typically computed as 1/2 the maximum time. Most drives are rated for around 10mS for the average seek time (12mS is common). Parking the head should not be much longer than twice that. In most cases, the parking head time includes spinning down the platters. For this purpose parking the head meerely means driving the head to the inner, unused, tracks and keeping it there in case it falls to the platter during a hard hit or during spin down. -Adam On Sun, 06 Mar 2005 04:36:11 +1300, Russell McMahon wrote: > I see IBM advertising that their Thinkpads have "drop sensing" which > parks the drive when the laptop senses that it has been dropped. May > have had this for years for all I know. > > I assume that sensing is done with several accelerometers - an XYZ > cluster would be needed to work in all orientations. When the vector > sum of acceleration falls to near zero it's time to shut down. > > They say that head parking time is 500 mS. For a straight clean drop > that's a 1.25m fall :-(. Too far. To get distance down to 600 mm you > need about 350 mS parking. > > Has anyone encountered this system yet? Any idea of how it works in > practice and how effective it is? It should really scream when dropped > and before impact to instil user confidence :-). > > RM > > -- > http://www.piclist.com PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist