I'm using Matlab for all my analyzes of vibration data from both SlamSticks and other data. I would prefer to have a common work flow between methods.

As I collect multiple other data during the test I do distinguish the different tests with different folders; hence reducing the usability of the batch tools.

It would really help me if you could supply Matlab code that read the IDE file directly into Matlab. Is this something you have, or plan to distribute?

Best Regards

Henrik

• Straight integration and double integration
• Calculating a PSD, then integrating that
• Calculating a shock response spectrum pseudo velocity which tries to define severity (admittedly this is better for shock, not vibration data)

Here is the video of the analysis:
https://info.endaq.com/hubfs/velocity-integration.mp4

Double Integration

First is the x axis

Then the y axis

Now the z axis

PSD

The RMS of each is calculated:

• X Axis: 0.4g
• Y Axis: 1.0g
• Z Axis: 0.7g
  

Now calculating the PSD in velocity, here is the RMS:

• X Axis: 0.15 in/s
• Y Axis: 0.70 in/s
• Z Axis: 0.38 in/s
  

Pseudo Velocity Shock Response Spectrum

Related article on data ("guess the vibe").

Time Data

First here is the time series data, the resultant is calculated with a root sum of squares = (x^2+y^2+z^2)^0.5

Here it is zoomed in, still very messy and hard to understand!

Power Spectral Density

These initial PSDs are "messy" because of the small bin width.

But they can be converted to 1/3 octave spacing (each frequency point is x2^(1/3) compared to the previous). In this cleaner format, the root sum of the squares can also be determined for a resultant PSD.

And now this can be converted to velocity easily in the frequency domain, attached is a PDF from VibrationResearch which goes into more detail

Some screenshots of interest:

Acceleration data of the two events are shown. The most significant moment for the trash truck was when it was placed down on the ground whereas the recycling barrel saw the highest shock as it was emptied.

Using a script to calculate the resultant acceleration then plot it, we can find the peak event shown here first for the recycling truck.

And now for the trash barrel:

Now I also found it interesting how the orientation changed as the barrel was picked up. So I made another script to **calculate tilt from the change in DC offset **which required some smoothing first. Shown first is the recycling truck, then the trash.

Now we also can convert Quaternion to Eulers to see the orientation change with another part of the same script shown here (first recycling, next trash):

Finally, I wanted to do some more formal shock analysis and calculate the shock response spectrum of the recycling barrel being emptied which is shown here. All in all not a super harsh shock event as the peak velocity is well below 100 in/s.

I just tested the vibrationdata Toolbox and it seems very interesting. Unfortunately some points are still tough for me, for example how to plot two curves on same window (as shown in the "flick vs punch" example).

Do you have a more detailed Tutorial on this software?

Regards

https://www.youtube.com/watch?v=ioSyeKusvE8

I attached the data file and below are some screen shots of interesting elements of the data. I'll embed a video analyzing all this data in a few days when we publish the full test video!

article link: https://www.popsci.com/technology/article/2011-01/nfl-test-helmets-impact-sensing-accelerometers-concussion-analysis/

Original Video

https://youtu.be/zvAQot_Mpro?rel=0

Answer Video (detailed analysis)

https://youtu.be/0p4NhZDOq5A?rel=0

Guess the Vibe during Sprinting

The main area of focus was the portion that he was sprinting we captured on the camera. Believe it or not he hit a ~75g peak for one of his strides! The rate of his strides was apparently at around 3.5 Hz with a second harmonic at 7 Hz.

Vibration of Regular Stride or Walk

I was also interested to see what the data looked like while just walking at a consistent clip. Here the peak acceleration levels are around 1.5g and have a very steady frequency of 2.4 Hz (with the second harmonic at 4.7).

Overview of Entire Walk

The walk started with the following stages:

• Turning the sensor on after putting on his leash
• Followed immediately by an excited shake from Barney
• Then I got sidetracked on my way out the door and had a conversation with someone
• Then I got outside and walked in one direction for a while, and around a building and back
• Once back inside I took him over to the front and let him off the leash to go sprint into the kitchen

Light

I think the light data is pretty interesting to show clearly inside vs outside.

Orientation

Then the orientation sensor is also pretty interesting to see it recognize I turned around. We need to do some work to help visualize this data...

Temperature, Pressure, Humidity

I am using Slam stick C and endaq software. Is there any way I can view frequency vs time plot? I want to get an median vibration frequency for the time period.

This is a long shot but the device I am testing give off varying acceleration in X, Y, Z direction. Is there a way to get the average acceleration of all three sub channels?

The Test

We took an enDAQ piezoresistive sensor and strapped it to a punching bag. Then we had a bunch of our employees punch the bag! Stiven, who did this testing will be generating a more formal report that we'll share soon. But I wanted to quickly illustrate the difference between acceleration and energy. So I took one of these punching datasets and compared it to one where I flicked/tapped the sensor quickly.

Acceleration Comparison

You can see from the plot that the acceleration levels of the flick are quite high, but they are over quickly.

Energy Comparison

Using the VibrationData Toolbox I calculated the energy response spectrum of each. Here is some more information on how I did that calculation. And here is the energy comparison!

Attached is a paper from Tom Irvine on how to calculate the energy response spectrum for more information.

1. While we run the motor and place the sensor on top of the Gearbox the vibration, it will capture due to all system attached to the motor not just by Gearbox and Bearing. So, how to find out the important signal?

2. What all parameters, features of the signal I should consider? What will be a good way to do it?

3. How to find that some fault has occurred?

4. How to distinguish between types of fault?

I want to use Python for signal analysis and trying to do it in real time.
Please guide me.
Thank you.
Gourav Kumar

Shock Response Spectral Envelopes (All Axes) for Typical Shock Events, Railroad
Freight Car Environment (Ahlbeck and Doyle 1974)

Greyboe, an industry-leading composite fiberglass stock manufacturer, has heard for years by users that their stocks exhibit the least recoil. To objectively prove this is true, Greyboe sought to perform some testing with accelerometers to directly measure the recoil compared to competitors’ stocks.

We mounted a few Slam Sticks to 3 different stocks and recorded the response during firing.

The Data

The Greyboe amplitudes were noticeably lower which suggests the reduced recoil that shooters feel with these stocks.

A team led by Christopher Carr (full citation below) performed some testing with a Slam Stick to quantify the "weightless" or reduced gravity conditions experienced in space or on planetary surfaces, e.g. the Moon or Mars. In their paper, they present a method for collecting, analyzing, and classifying the altered gravity environments experienced during a parabolic flight. Specifically, they used Slam Stick during a Boeing 727-200F flight with 20 parabolas.

Their data and analysis code are freely available:

• Report: https://arxiv.org/abs/1712.05737
• Data: https://osf.io/nk2w4/, a segment is attached
• Code: https://github.com/CarrCE/zerog

Results

The following figure shoes the section of the aircraft the Slam Stick was located in, the orientation, the raw DC data, and then the filtered data.

The following is a power spectral density of g level that was used to help identify the filter frequency cutoff to use.

Citation

Carr CE, Bryan NC, Saboda K, Bhattaru SA, Ruvkun G, Zuber MT. Acceleration Profiles and Processing Methods for Parabolic Flight. arXiv:1712.05737; code: https://github.com/CarrCE/zerog; data: https://osf.io/nk2w4/

Raw Data

Spectrogram

The spectrogram is especially cool because you can see different periods of my drive that I'm on the highway and/or accelerating and revving the engine.

The Test

Tom prepared a report on the pedestrian bridge's natural frequency and posted it to his vibration blog. Using the vibration data captured and an amazing vibration analysis MATLAB GUI package he wrote (and offers for free!) he was able to determine the bridge has a fundamental frequency of about 2.2 Hz with 0.16% damping.

The Data

Real Life Examples

Performing some MIL-STD-810 drop testing to evaluate a product, the standard is attached. The particular tests included in this post were dropped from 2 feet. Acceleration levels can be calculated from drop heights if the pulse width is known, the attached paper provides more details. A shock event with a 2 to 3ms pulse width, dropped from 2 feet should have between 170g and 280g peak acceleration.

The Data

The results taken from the piezoresistive accelerometer provided very clean results and was consistent with the peak levels one should expect with the known drop height and pulse width.

I wrote a detailed blog that goes through the test and data for more information. But to summarize, I put a few Slam Sticks on shipments around the country to get some interesting data that could be useful to folks trying to understand how to best design their system to survive shipment.

UPS Ground

One of the shipments was UPS ground in a small carboard box.


There were also a lot of significant shock events as the package was "thrown around." Here's one that was nearly 100g.

UPS Freight

One of the shipments was UPS ground in a small carboard box.

We once had a small development effort to evaluate the potential for harvesting energy from a truck bed to power a monitoring system. There appeared to be some decent energy available but the project didn't go too far due to funding. But it did let me get some pretty cool data!

I mounted a Slam Stick to the bottom of the truck bed and recorded a trip it went on which included highway travel.

The Data

Zooming in, there was pretty much always some healthy vibration.

There were also the occasional bumps from engine braking and potholes.

The Spectrogram

I found this spectrogram especially cool because there was some band of frequency content that was between 200 Hz and 400 Hz. The shape of that suggests it is directly proportional to the truck's speed and/or the engine.

The PSD and FFT

The PSD and FFT data suggest that a vibration energy harvester could potentially work although it would be generating microwatts of power in ideal conditions.

A customer of ours was developing a robot and needed to understand the vibration environment it was going to experience to simulate that in their lab. They also wanted to better design the isolators that support the electronics to work best in the actual environment.

The Data

The Spectrogram

The frequency content was surprisingly consistent regardless of the robot's activity which was definitely helpful in designing an appropriate vibration isolator!

The PSD

The PSD data was used to then drive their shaker system to simulate the environment.

The Test

• Slam Stick taped down
• Frequency dwells: 10, 25, 50, 100, 200, 300, 400, 500, 1000, 2000 Hz
• Amplitude dwells: first sweep at 1.05g, 10g, and 15g (the 10 Hz signal is 1/15 the other frequencies)

The Data

The Spectrogram

Full dataset

The full dataset is especially interesting because it has some large shock events as well as some consistent vibration periods.

Shock event

As the drill head hits various different material it experiences large shock events that their electronics would need to survive through.

Vibration

During times the drill is just humming away, the vibration environment is rich with not only frequency content from the drill's rotation speed but also some other structural information at higher frequencies.

Spectrogram

The spectrogram is always cool, and this test is no different!

I always found the spectrogram particularly cool with the different harmonics in the signal.

But a look at the raw data helps show the pretty crazy vibration too!

This data may be useful to someone to understand what type of vibration a passenger feels on a plan. Below is an image of the raw data as well as a spectrogram which looks pretty cool!