ENERGY HARVESTING PRODUCTS
Slam Stick™ Vibration Recorder
The First Step for Developing Any Vibration Energy Harvesting System
Simple and Easy to Use
Industrial Health Monitoring Network Sensors
The SlamStick™ is a high speed ultra portable rechargeable data logger capable of measuring acceleration in all three axes. The device uses a USB Port for on the fly configuring, charging and downloading of data. Simple configuration software allows users to tailor the device to their specific needs.
The configuration options include an adjustable delay time before each measurement, different measurement durations, and a trigger based on a pre-determined acceleration level. Free analysis software allows for complete vibration characterization enabling the user to determine the frequency content of their given vibration.
The SlamStick™ can be used for many different applications including energy harvesting vibration characterization, modal surveys and equipment monitoring. Simply configure the device for your application, mount the device on your vibrating structure, hit the button and download the data to your computer.
Easy to Use!
Attach the Slam Stick to the vibration source using any means that will ensure a high quality vibration energy transfer with minimal damping. Ensure that the mounting method also allows for easy removal of the Slam Stick after the test. Ensure that the axes orientation of the Slam Stick is suitable for the desired measurement (see figure at right). It is suggested to take a picture of the device for later confirmation of mounting position. Suggested mounting options include bee’s wax, thin double sided tape and duct tape.
To begin the vibration measurement press and release the white button on the Slam Stick. The delay before measurements are taken as well as the sampling frequency and the total test time will be chosen by the customer during the ordering process. A flashing LED will indicate when the testing is in progress. At the completion of the test the LED will stop flashing.
To download the data, plug the device into a USB port and copy the data.dat file to a desired location (don't delete files from the Slam Stick). The data is in a binary format - use one of the software downloads above to view data interactively or convert to a plain text (.CSV) format for use with other software.
To perform a reset hold down the white button while unplugging the USB connection. This will reset the device and it will be ready for more testing - Don't delete files from the Slam Stick.
Leave the device plugged into the USB for recharging. A green LED will signify that the device is charging. A normal charge will take 1 to 2 hours. Generally 3 to 4 recordings can be performed per charge.
What is vibration energy harvesting?
Vibration energy harvesting is the act of converting otherwise wasted mechanical energy in the form of vibrations into useable electrical energy. This electrical energy can then be used to recharge batteries or power low-power sensing systems. Midé performs this mechanical to electrical energy conversion by utilizing the unique properties of piezoelectric materials, which, when stressed and strained, produce electrical potential.
The drawback to these materials is that they are brittle. However, Mide uses a patented packaging process (QuickPack™ Advantage) to dramatically increase the robustness of the materials for use in harsh vibration environments.
How do you know if your application is suitable for vibration energy harvesting?
Midé’s energy harvesting devices - Volture™ Energy Harvesters - use what is known as resonant harvesting. This means that to be effective, Midé’s harvesters need to be “tuned” to match the vibration from which they are harvesting. To know if your application is suitable for this type of harvesting, the first thing to know is what your vibration “looks” like. To determine this, the vibration needs to be measured using an accelerometer. Depending on your application, this could either be a somewhat difficult and expensive task or a VERY difficult and expensive task….until now!
Slam Stick™ Vibration Recorder
|Max Acceleration (each axis)||16||g|
|Sampling Rate (each axis)||3.2||kHz|
|X,Y axis noise||0.016*||g|
|Z axis noise||0.022*||g|
|Max Recording Time at 3.2 kHz||240||seconds|
|Min Operating Temperature||-40||C|
|Max Operating Temperature||80||C|
*See noise histogram data in the Testing section of the Data Sheet
Slam Stick™ Vibration Recorder
“Noise” in this document refers to sample-to-sample deviations of the measured signal from the actual value due to internally-generated electrical interference, thermal effects and other factors (not acoustic sounds). Random noise is statistical in nature and follows a Gaussian (bell-curve) distribution, thus a histogram is normally used to quantify its distribution in a given signal.
For the noise measurement, a sample Slam Stick recording was taken at room temperature with no vibrations present. The Plots above represent the distribution of noise in the recorded data for each axis. This distribution is approximately Guassian, with the majority of samples falling very close to the correct value and less frequent outliers falling further from the correct value. This distribution is typical of random noise present in any measurement.
Each bin represents one LSB (Least Significant Bit, or the minimum G-level difference that can be represented, by a 1-bit change, in the recorder’s digital output), while the X axis shows the corresponding error in Gs and the Y axis value indicates the percentage of samples which contain this amount of error. The RMS error (or standard error) is the error value at which 68.2% of samples (one standard deviation) have an error lower than the RMS value.
Frequency roll off data was taken by exciting the Slam Stick at a constant amplitude (1 gee in this case) over a range of frequencies from 10 Hz to 3200 Hz. The measured vibration amplitude from the Slam Stick was then plotted versus frequency. Note that the roll off is severe after ~240 Hz. Different devices performed differently after this frequency and amplitude data should not be trusted when above this frequency. Frequency data is not impacted by this roll-off. For example if a vibration source was providing a 3 gee amplitude at 500 Hz, the Slam Stick would properly provide the relevant frequency information, however the gee level would be inaccurate, most likely providing a lower than 3 gee amplitude reading.
Slam Stick™ Vibration Recorder
Any complex waveform, such as Slam Stick vibration traces, can be expressed as the sum of sinusoidal frequency components.
Per Shannon’s Sampling theorem, in any sampled data such as that produced by the Slam Stick, the maximum frequency that can be reliably identified is limited to half the sampling frequency. This limiting frequency is known as the Nyquist frequency, critical frequency or folding frequency. Any frequency content at or above this limiting frequency will produce an alias, or false signal appearing at a lower frequency. This alias will appear as the reflection of the high frequency signal over the folding frequency. For example, if the folding frequency is 1600Hz (half of the Slam Stick’s 3200 Hz sampling frequency), an 1800Hz signal would appear as an alias at 1400Hz, and a 1900Hz signal would appear at 1200Hz, etc. Figure 1 shows an example of a real signal exceeding the folding frequency, and the apparent (aliased) signal that results.
Figure 1: Aliasing occurs when the real signal frequency exceeds the Nyquist frequency, resulting in there being not enough data points to correctly reconstruct the signal.
Sampling devices such as the Slam Stick include a lowpass filter at the folding frequency to minimize the appearance of such aliases (see the Frequency Response Graph in performance plot section); however, no filter can completely eliminate unwanted signals, so a very strong signal above the folding frequency may still produce a visible alias. The best course of action is to recognize that this can occur and take steps to identify aliasing if it occurs. For example, if the frequency content of the system is known to be (or found to be) dependent on an external parameter such as engine RPM, observing the response as this parameter is varied can help in identifying spurious signals.
Commercial Aircraft Vibration Characterization
The following shows a brief example of using the Slam Stick data to identify specific features in a recording. This example uses the included MATLAB script for data visualization.
Figure 2: Time domain response of a commercial jumbo jet during taxiing.
A series of vibration recordings were taken during a flight on a commercial jumbo jet. The first was taken during taxiing; the second taken during active flight. In each case, the recorder was placed on the floor of the craft with the button facing upward (Z-axis normal to the ground). The time-domain result is shown in Figure 2.
Unsurprisingly, the greatest energy is seen in the Z direction. The 1-D FFT result for this axis, shown in Figure 3, shows several features that may be of interest. The first is a comparatively strong peak at around 920Hz, some extremely low-frequency components (1.5Hz and 5Hz), and a broad swath centered around 260Hz or so. What’s producing these features, and are these consistent? Would any of them be useful in an event triggering or energy harvesting application?
Figure 3: 1-D FFT of the time domain data shown in Figure 2.
For a more detailed look at the vibration profile, the 2-D FFT was plotted with a low ceiling value (~ 0.05) in order to clearly show each feature of interest. For this Figure, the surface plot was rotated so that it is viewed directly from the top. Some of the identifiable features have been labeled.With this image, the contributors become more clear. The 920Hz peak is extremely consistent both in time and frequency, most likely contributed by the aircraft Auxiliary Power Unit (APU). Although it appears to have the highest amplitude in the 1-D result by far, it is not the highest amplitude signal after all – those in the 150-375Hz range have higher peak amplitudes, but appear negligible in the overall 1-D result since this energy is spread out over a wide frequency range. A closer look at the 150-375Hz signals, along with the similarly-shaped features at higher frequencies, shows that they all start off at multiples of approximately 64Hz and 156Hz. That is, 64, 128, 156, 312, … . Note that not all of these overtones are visible at t=0; the higher modes are visible mainly as they cross fixed resonant modes in the aircraft structure itself. Some examples of these mode crossings are near 750Hz, 900Hz, 1050Hz and 1200-1300Hz. The 150-375Hz signals are most likely the n/rev contributions of one engine being spun up, meaning they are engine-speed dependent and may not be reliable for e.g. energy harvesting, but may be useful to estimate engine RPM remotely.
Figure 4: 2-D FFT of the time domain data shown in Figure 2. Salient features have been labeled.
A couple other features are of note in this plot: At approximately 28 seconds, a brief horizontal ‘broadband’ feature is evident, most prominently from 920Hz all the way up to 1600Hz. This is the impulse response as the craft crosses a large crack or other discontinuity in the taxiway, creating a mild mechanical impact. Other sudden mechanical shock or impacts will produce a similar response. Finally, note the signal between 1200-1450Hz at approximately 13-15 seconds. As the engine speed continues to increase, frequency features in the inverse (decreasing) direction are evident. This is a good example of an aliasing artifact caused by a very strong signal beyond the maximum measurable frequency of the device. As mentioned in the Aliasing appnote, any signal energy remaining beyond the Nyquist (folding) frequency will be mirrored over this frequency and appear as a spurious signal below the Nyquist frequency.
The next series of plots shows data taken from the same location during flight at cruising altitude. The APU contribution is still visible in the Z axis, but the response is now dominated by “wind noise” at 400-600Hz (Y axis) and a similarly wide band centered around 380Hz (Z axis). The X axis shows contributions from both of these sources. Unlike the previous case, all of the peak frequencies are stable over time, making these more suitable energy harvesting frequencies.
Figure 5: 1-D FFTs of vibration data during cruise.
Figure 7: 2-D FFT of vibration data during cruise (X axis)
Figure 8: 2-D FFT of vibration data during cruise (Y axis)
Figure 6: 2-D FFT of vibration data during cruise (Z axis)