I would like to get your opinion on the feasibility and practicability of my project idea.
For the qualification of different damping materials, I have thought about a test setup that is as cost-effective as possible, but which should still provide meaningful measured values.
In the following I will describe my idea of the experimental setup:
A hollow metal ball, filled with the measurement technology and foam, is dropped from a height of 3 meters onto the damping material to be tested. During this process, the measurement technique records the acceleration values.
The metrology itself consists of an Arduino, power supply, some sort of data storage or transmission, and accelerometers (which are attached directly to the inside of the metal sphere). Since I am aiming for a sample rate of 10kHz, I believe based on my research that multiple accelerometers will need to be used. This would allow their limited sample rate to be added up to the desired sample rate of 10kHz.
Accelerometers (all sensors, really) have to be bandwidth limited to half the output data rate, otherwise you get aliasing artifacts. So, if you want to collect data with 5 kHz maximum frequency content, you need an accelerometer capable of a 10 kHz output data rate.
You might be able to get 5 kHz for both the acc and gyro components from the LSM6DSL after some filtering. But that is 60 kBytes of data per second (12 bytes x 5000) to process. Certainly not in the realm of Arduino UNO or even maybe Cortex M0+.
If all sensors (happen to) sample at the same time, they will report the same values - no improvement
A major bottleneck will be data transmission. With I2C 2 or 3 bytes (including address) will have to be transferred for each sample and a maximum of 5k samples/second at 100kHz. This rate can not be increased by the use of multiple sensors.
Similar considerations apply to the storage device. It must not compete with the sensors on the same bus and must have a sufficiently large data buffer.
How many samples do you expect for one experiment?
I would like to add that I do not have that much experience in arduino and programming or writing codes. But I am willing to learn. But my ideas was to check if my project it worth to make it "real".
I did some research and found this accelometer:
It has a huge measurement range (+-200g which should be sufficient) and a bandwith of 200Hz to 3,2kHz.
jremington Thanks. That means with this accelemoter I mentioned above the max. sample rate would be 3,2kHz/2 = 1600Hz?
@hzrnbgy Thanks. I would like to add that I built a prototype where I used the GY-521 which has only a measurement range of +-16g. This is way to low. I dropped the arduino + Gy-521 from ~ 40cm above the pillow on a soft pillow and the results showed that there is (of course) this 16+ limit. This tells me, that a need a higher measurement range.
@DrDiettrich Thanks. Let's say money doesn't matter, would it be an improvement if I would use multiple systems (Arduino, power supply, acc. sensor, data storage) and merge the measurement data afterwards with excel?
The measurement itself is rather a qualitative measurement than a quantitative one. Not sure if I got your question correct, but I would like to test "many" samples from time to time. My intention was to start with some EPS with different densities to create some kind of baseline or reference.
I can give only very general help. Most important will be the reproducibility of your experiments. E.g. how do the values differ for small variations in the angle of the impact? Is there a variation or decrease in subsequent drops onto the same material?
That's a really good point.
My idea was either to use the resultant g ( sqrt(x²+y²+z²) ) to get rid of variations in the angle of the impact or another build of the setup to have only the z-axis for example to take into account.
After the initial impact there could be some kind of rebound. But this event can be seen due to the data collected. I will use a new sample every drop test.