Tip:
Highlight text to annotate it
X
Hello everyone, today we are going to talk about the basics of an accelerometer.
There are two main types of acelerometers
Those that track the acceleration due to gravity, where the accelerometer readings
give the acceleration field in 3 axes. These sensors need to track motions
only as fast as we say.. move our phone.
and are mainly used for orientation. So it’s measurements are on the order
of 1g with milliseconds of response time.
Another application is impact detection. Free fall sensors in laptops need to dock drives
before the laptop hits the ground (0g) .
Automotive accelerometers need to respond quickly, with enough
accuracy and precision to reliably tell the difference between a sudden stop and a collision. This means
that these sensors require higher bandwidth and resolution that an orientation accelerometer.
With that said, the techniques for their construction and operatin principles
are the same. We are only going to focus on simple one axis accelerometers.
Something closer to a phone’s accelerometer than an automotive one.
Let’s start with the proof mass. The proof mass translates the acceleration into force.
The larger the mass, the larger the force. We want this to be
as large as possible without increasing the cost.
Next, we need to convert Force into displacement by using a spring system.
The same working principle as a spring scale.
Except that we need to make sure that the spring mass system can rotate
With the spring mass system in place, we need to measure capacitance. The most common method
uses capacitance. To make the math simple, let’s just look at small displacements.
The difference between the accelerometer capacitance and
its value at zero acceleration is proportional to to displacement which in turn is
proportional to acceleration. We can measure the difference of the two capacitors using a transimpedance amplifier.
The negative feedback forces node 2 to virtual ground
This means that the current flowing in the feedback network is due to Csense and Cref. This translates
into a voltage at Vout that is again proportional to the acceleration.
This is all we need for a single axis accelerometer. So let's think about what sensitivity we can get
out of a spring mass system. Cost limits us on the area of the plates
but we can minimizethe distance between them as long as the device has room to move.
We could also increase the capacitance by adding high dielectric films or patterning the plates.
Increasing m and decreasing k is also an option
as long as we are careful. As we apply an acceleration that changes in frequency.
Our accelerometer responds at the same frequency. At frequencies above it's resonance.
The accelerometer no longer responds. This means that the requirements for response time will put a limit on the parameters m and k.
We also need to worry about noise. The first source of noise is due to
the spring mass system. It is produced by random vibrations in the proof mass.
It also gets worse with softer springs and bigger masses.
We can also increase sensitivity through our electronics by decreasing the feedback
capacitance or increasing the input voltage. This limits the dynamic range of “a” as the
signal can’t be more than the supply. This only helps when the circuit noise is more
dominant that the device’s thermal noise, which is the noise due to the spring mass system.
Another consideration for the circuit is it's own noise.
To first order, the circuit is not sensitive to parasitic capacitance.
However any parasitic capacitance at the summing junction decreases the feedback attenuation, Beta.
The feedback node sees any parasitic capacitance in parallel
with the signal capacitance. This increases the circuit noise and reduces the bandwidth of the opamp.
as higher gains means lower bandwidth.
That’s all for today, but consider the following classes for more detail on
accelerometers, gyroscopes, microphones, compasses, chemical sensors
and many other MEMS devices. Thanks for watching!