08-30-2018, 08:31 PM
(This post was last modified: 08-30-2018, 08:32 PM by Sebastian Staacks.)
Actually, it is the other way round:
The accelerometer is a sensor which measures forces acting on a mass. You can imagine it as a small test mass, which is kept in place by a few springs (here is a nice short animation of a more realistic model https://www.youtube.com/watch?v=4kfzqZpttTA). When you accelerate the system, the mass will move against these springs because of inertia. But it will also move against the springs because of the gravitational force. Therefore those accelerometers always measure the acceleration of the device and the earth's acceleration as well. (When referring to the animation: Imagine, that you would tilt the sensor, so that gravity pulls the inner part to one side).
This is why every device has an accelerometer. Originally, people were not that much interested in the actual acceleration, but in the direction of gravity, so the device can recognize when it is rotated, so it can rotate the content of the screen as well.
Also, this is why the sign of earth's acceleration seems to be wrong (depending on how you define this). If the phone is resting on a table, gravity will pull the mass "down". But to the only way to pull down the mass due to inertia is by accelerating the phone up. So "downward" gravity has the same sign as "upward acceleration", which is somewhat unintuitive.
Now for the acceleration "without g":
This one is actually a bit tricky and is usually composed from the accelerometer with g and the gyroscope. The trick usually is, to assume that the measurement starts with a resting device, so you can just subtract the measurement at the beginning (the 9.81 m/s²). After that, you need gyroscope to keep track of any rotation, because if the phone rotates, you need to subtract these 9.81 m/s² from a different direction. On top of this, the sensors tend to drift a bit, so there usually are some algorithms that try to remove drift with additional assumptions.
Luckily for us, phyphox does not need to do this itself, but instead the "acceleration without g" is provided by the system (on both, Android and iOS). The downside of this is, that we cannot really know what the maker of a phone is doing there to generate this data and some phones even try to do these calculations without a gyroscope, which does not really work...
So, in fact, during a free fall, you would expect a constant 9.81 m/s² for the version "without g" and just zero for the version "with g". I have to admit, that I have no idea right now, how you could get 20 m/s²... Can you share data from such a drop?
(By the way, I just moved this thread to the general section.)
The accelerometer is a sensor which measures forces acting on a mass. You can imagine it as a small test mass, which is kept in place by a few springs (here is a nice short animation of a more realistic model https://www.youtube.com/watch?v=4kfzqZpttTA). When you accelerate the system, the mass will move against these springs because of inertia. But it will also move against the springs because of the gravitational force. Therefore those accelerometers always measure the acceleration of the device and the earth's acceleration as well. (When referring to the animation: Imagine, that you would tilt the sensor, so that gravity pulls the inner part to one side).
This is why every device has an accelerometer. Originally, people were not that much interested in the actual acceleration, but in the direction of gravity, so the device can recognize when it is rotated, so it can rotate the content of the screen as well.
Also, this is why the sign of earth's acceleration seems to be wrong (depending on how you define this). If the phone is resting on a table, gravity will pull the mass "down". But to the only way to pull down the mass due to inertia is by accelerating the phone up. So "downward" gravity has the same sign as "upward acceleration", which is somewhat unintuitive.
Now for the acceleration "without g":
This one is actually a bit tricky and is usually composed from the accelerometer with g and the gyroscope. The trick usually is, to assume that the measurement starts with a resting device, so you can just subtract the measurement at the beginning (the 9.81 m/s²). After that, you need gyroscope to keep track of any rotation, because if the phone rotates, you need to subtract these 9.81 m/s² from a different direction. On top of this, the sensors tend to drift a bit, so there usually are some algorithms that try to remove drift with additional assumptions.
Luckily for us, phyphox does not need to do this itself, but instead the "acceleration without g" is provided by the system (on both, Android and iOS). The downside of this is, that we cannot really know what the maker of a phone is doing there to generate this data and some phones even try to do these calculations without a gyroscope, which does not really work...
So, in fact, during a free fall, you would expect a constant 9.81 m/s² for the version "without g" and just zero for the version "with g". I have to admit, that I have no idea right now, how you could get 20 m/s²... Can you share data from such a drop?
(By the way, I just moved this thread to the general section.)