2D vs. 3D Accelerometer Technology
I like our accelerometer chip that powers our Active Protection System. I wrote earlier about how you can use it to play games. While this is one of our most widely understood technologies, I still get questions on it all of the time. One common question goes something like this:
I noticed that Lenovo’s Active Protection System works in 2 dimensions. I’ve been looking at Brand X and they have a 3 dimensional accelerometer. For the company that basically invented this use of the technology, you all are behind the curve.
Alternately…
Brand X has a 3D technology. Lenovo only has 2D technology. YOU GUYS SUCK!
It’s a valid question. After all, it seems that anything 3D would be more advanced than something 2D. So it stands to reason that we should use the more advanced technology. Yet we don’t, and on the spec sheet, we look sleepy, even though from a technical perspective, we don’t see any advantage to using a 3D chip.
Theoretically there is only one possible situation where having a 3D accelerometer chip would work to protect a hard disk drive where a 2D accelerometer would not. I call it the Star Trek scenario.
The way our Active Protection System (APS) works is that it constantly detects movement and then an algorithm uses that movement to predict the risk that the system will undergo a shock that might damage the hard drive. In the Star Trek scenario, there is a test that simulates a table that is instantly vaporized from underneath the notebook–just like Captain Kirk would do. After the table is vaporized, the notebook then falls to the ground without any X or Y axis pre-information to tell our APS to work. This is nearly an impossible situation in the real world as there is always SOME movement ahead of time to allow our APS to react appropriately.
That reason alone wouldn’t be enough to keep us from using a 3D sensor. Another reason we use our current 2D sensor is that it is more accurate than most commonly used 3D sensors at protecting the hard drive. This is because it detects smaller accelerations than the 3D sensors. Ideally you want a sensor that is more sensitive to small changes than to big changes. This way it can react to smaller drops and bumps (accelerations).
For those performance gurus, a 3D sensor requires more processor cycles to process the data, but in reality the overhead is so small, it is a rounding error.
Customers have also made some interesting suggestions for improving our APS technology. One I particularly like is to add reporting capabilities to the software driver. That way the IT department could determine when a system comes in for repair how much a user abuses their system.
Another request is to have the system beep when the APS engages. The idea behind this would be to modify user behavior through negative reinforcement. Personally, a setting like that would drive me crazy and I’d end up throwing the system down in disgust just to get it to shut up.
Long term, the APS becomes irrelevant. When we all are using solid state memory to store our data instead of moving platters, reliability becomes infinitely better.
Lenovo Meet the Modder Dean Liou
Lenovo Meet the modder- Chris Blarsky Dairy 2
Lenovo Meet the modder- Chris Blarsky Dairy 1
Lenovo H320 desktop
March 27th, 2007 1:32 pm
I’d like if if there was a (configurable/optional) alert when APS engages. That way, if your system freezes or gets real sluggish, you’d know if it was due to APS stopping disk access. We’re talking about Windows, after all. Anything that can help us understand why it’s acting funny would be useful.
March 27th, 2007 2:30 pm
Rafe,
There is an option that says “When hard drive is stopped, prompt me to manually suspend the Active Protection System.” I don’t think that is quite what you’re looking for though – I know on older ThinkPads or older APS software there was an option where a window would pop-up and require you to hit OK everytime APS intervened.
I envision a small Windows tool-tip popup when APS intervenes, maybe some kind of change to the HDD indicator LED? I’m just talking off my head here, but I like the idea
March 27th, 2007 3:13 pm
Matt – how about a post soon specifcially on those solid state devices…?
March 27th, 2007 3:24 pm
A nice idea is to use the “new” technology of Vista OS that is called “ReadyBoost” on the main board for Notebooks that will increase the performance and reduce the use time of the HDD. In that way the use of APS would be less notice by the user. When Lenovo will install on the notebook the SSD disks?
March 27th, 2007 8:52 pm
What bothers me more about APS is not the 2D sensor, but the fact that the drivers seem to rely on the operating system and the main CPU. (I may be wrong about this, but that’s the impression I got through some testing.)
If this is the case, doesn’t that mean that APS may fail to trigger in time if I’m running a CPU-intensive application such as Photoshop or a game?
March 27th, 2007 10:19 pm
I don’t really use the APS sensor, its a nice idea but in my usage it just activates and annoys me when my ThinkPad hasn’t been dropped, I am constantly downloading something (off newsgroups) and drag my system around while its running, I know this isn’t good for it but its an acceptable risk to me, i’ve dropped several thinkpads from several high heights and have never lost data as a result; to me protection is better than prevention (APS is prevention imho) and ThinkPads are the best when it comes to that due to their solid build quality and now the shock mounts for the HDD. I’d idealy like some sort of sprung suspension system to encase the hard drive but, as you say, this is somewhat of a moot point with SSD on the cusp of taking over in the next couple of years.
I hope that ThinkPads (particularly X series) are available with the SSD drives soon as I see sandisk is now shipping to the OEMs and is estimating only a 200 dollar premium for the 32GB SSD.
March 28th, 2007 7:31 pm
Matt,
THe other vendor, even though it has 3D APS, does not have any sort of software driver/interface to let you know it is working. The APS on ThinkPad’s is great because it lets you know when it is pausing the drive, and gives you a visual representation of the system. I had been sold on the other vendor saying, if you have 2D, we have 3D, 3 is more than 2 so you should think this is better. It is also a discussion of physics, I have never seen a laptop fall completely flat. They are all weighted so that some rotation will occur. I have 2 vendors laptops, and while one has 3D… I don’t know that it does except in the spec sheet when ordered.
March 28th, 2007 11:27 pm
Angus, Great idea. I’ll do one soon.
Panagiotis, I can’t comment on unannounced products, but I’ll give you some more information hopefully that can help.
D — You’re right. The drivers for any of the accelerometers out there are OS based and therefore there is a very short time when booting where the drive does not have the protection of the APS. To minimize this danger period, we load the APS driver very early in the boot sequence. To answer your question, even if you are running the CPU 100%, you still have APS protection. This is one of the tests we perform in the development cycle for the driver.
Snife – a spring system for the HDD would have its own problems, namely the drive bouncing around and causing head/platter slap as the spring bottoms out. If you pull out a drive from a recent ThinkPad (a ‘60 or newer), you’ll see that inside the system we surround the HDD with rubber rails that provide additional cushioning and do much the same thing as you’re asking.
J S – Disagreement is great for discussion, but anytime someone wants to agree, I won’t say no. Thanks.
March 29th, 2007 5:00 am
Matt, why not release info about what each sensor does exactly so the driver isn’t so tied to Windows?
April 4th, 2007 12:17 pm
One solution to the APS notification issue might be to have a little APS light on the row of status LEDs, kind of like how modern cars have a traction control light on the dash that flashes when it’s working.
April 9th, 2007 10:40 pm
Z – We have intellectual property tied to how the senor data gets interpreted and reacted upon.
May 3rd, 2007 11:13 am
Why not have a dual colored LED for the HDD light? So normal color for HDD access but red when APS is active. (there can’t be any HDD access with the HDD is parked anyway) No extra real estate needed and has the negative reinforcement a beep has without being so annoying. (under the assuption people will think a red flashing light is bad so should try and avoid it)
September 10th, 2007 9:21 am
I suggest following thing: make APS generate interrupts when movements exceed configured thresholds. Why? Because now embedded controller must be polled to check position. It wakes CPU thus shortening battery life.
Other manufacturers of laptops with accelerometers (one with fruit in logo, and second making mainly printers) made them generate interrupt on sudden movement.
February 4th, 2008 1:53 pm
You pointed out that the 2D senor does not catch the case where the laptop is dropped straight down, without tilting. It was noted that this is unlikely to occur in real life. But a 2D sensor also does not catch the case where the laptop is jarred in a way that it rotates around on its bottom without tilting the bottom off the horizontal plane. That case *is* likely to occur in real life.
June 24th, 2008 6:00 pm
X YandZ: that would only work if the axis of rotation coincides with the exact position of the sensor, right?
I think Lenovo needs to develop a 4D sensor. Not just because 4 is a bigger number, but because if our laptops were temporally-aware they could protect themselves from other things being dropped on them. Currently, how would a 2D or 3D sensor detect an incoming projectile?
I want a TardisPad.
September 25th, 2008 3:16 am
I’m not convinced: a 2D accelerometer can’t gauge the magnitude of the acceleration (except that the reading is a lower bound), so has to have a smaller threshold for parking the heads; whereas a 3D accelerometer should be able to tell you when the acceleration is about 9.8m/s², so should give fewer false positives for a given level of protection.
October 21st, 2008 7:34 pm
Matt gave us only two real reasons for using a 2D sensor instead of a 3D one:
1. 2D is good enough for real-world APS.
I agree with that, but Matt goes on to say,
2.”That reason alone wouldn’t be enough to keep us from using a 3D sensor. Another reason we use our current 2D sensor is that it is more accurate than most commonly used 3D sensors at protecting the hard drive.”
But this reason is not honest: ThinkPads apparently use the ADXL320 accelerometer unit from Analog Devices. But there is also the 3-axis ADXL330 accelerometer, which is, according to the manufacturer, even more sensitive than the 2-axis ADXL320 (due to lower range, 5g vs. 3g), and its noise is nearly the same. But, it is 50% more expensive ($2).
So, unsubstantiated cost is IMHO the only reason behind 2D vs 3D!
October 21st, 2008 11:07 pm
Cost is always the number one factor of any manufacturer whom wants to make money. If you have unlimited funding you can be NASA, go out and spend 10 million USD back in the early space program, and invent a pen that write in weightlessness and underwater. Or if you are cash strapped, then go with the Soviet version, a simple charcoal pencil.
June 7th, 2009 1:39 pm
Please start supporting APS in Linux
February 3rd, 2010 5:36 pm
Is there any chance that Lenovo is going to release a driver that supports Windows’ 7 Sensor and Location API?