What an accelerometer feels
Lay the board flat on a table and it reports 1 g, not zero. That surprises almost everybody, and it is the fact the rest of this book is built on: an accelerometer does not feel gravity pulling. It feels whatever is pushing back.
A weight on springs
Inside the chip, cut out of silicon, is a small mass hung in a frame on springs. When the frame is pushed, the mass lags behind and the springs bend. The chip measures how far, as a change in capacitance between the mass and fixed plates beside it, and turns it into a number.
That number is the push the frame is getting, divided by its mass. It is reported in g, where 1 g is the push it takes to hold something up against the Earth: 9.8 metres per second, every second.
Four situations make the idea clear. Pick each one and watch the springs:
On the table, the table pushes up and the springs sag under the mass. The reading is 1 g, pointing up. Dropped, nothing pushes at all, the mass floats in its frame, and every axis reads close to 0. In a lift, the push grows as the car starts upward and shrinks as it slows at the top.
The lift numbers in the figure are an illustration of the size of the effect, not a measurement. The other two are what any accelerometer on Earth reads.
Why that is useful
It sounds like a defect: the one thing the chip is for is measuring motion, and standing still it reports a whole g. In practice it is the most useful thing about it.
That 1 g always points straight up, whatever the board is doing, as long as it is not being shaken. So a board that is still can work out which way up it is, and by how much it is tilted, from nothing else. That is how a phone knows to rotate its screen and how a camera levels its horizon.
And a board that is being moved reads something other than 1 g in total. The difference is the motion. A knock, a shake or a drop all show up that way, and counting shakes is built on it.
What this chip is
The part on the TK115 is a Silan SC7A20H, 2 mm square, in the middle of the board between the two mounting holes. It measures three directions at once, up to ±16 g in its widest range, up to 4434 times a second at its fastest, and it draws 16 µA doing it at 100 readings a second. It talks to your board over I²C, the two-wire bus most sensors share.
It will not tell you where the board is, or how fast it is going. It measures pushes, and only pushes.
When it does not work
That is correct. A still board feels the table pushing it up with exactly enough force to stop it falling, and that push is 1 g. Zero on all three axes would mean the board is falling freely, which is the one state in which nothing pushes on it.
Also normal. The chip is calibrated at the factory, but soldering it down stresses the package slightly and every axis can read a few tens of milligrams off. The data sheet allows up to 120 mg on a mounted chip. From counts to g covers what that costs and how to take it out.
No. It reports acceleration, the change in speed, not speed. A board on a train at a steady 100 km/h reads exactly what it reads on your desk. Speed would have to be worked out by adding up acceleration over time, and the small errors in every reading add up with it, so after a few seconds the answer is useless.
The chip measures the push on three axes at once, and how the 1 g is shared between them is how it knows which way up it is.
One g, shared three ways →Edit this page — content/books/3-axis-accelerometer/what-it-feels.mdx
Questions about this product
See what other owners have asked, and read their solutions.
3-Axis Accelerometer
Loading discussions…
Discuss this article
Ask about this page. The answer stays here, on the page it belongs to, for whoever hits the same wall next.