Evaluate your application’s energy-harvesting vibrational profile with a Slam Stick
Using energy harvesting to power an application from ambient mechanical vibrations has a lot going for it. Maybe the application is off-grid or simply too far from a power outlet and has a service cost that precludes batteries. But how to know if your application is suitable for vibrational energy harvesting?
Midé Technologies’s energy harvesting devices rely on resonant harvesting. This means that to extract the maximum available vibrational energy from its environment, the harvesters need to be “tuned” to match the vibration. To know if your application is suitable for this type of harvesting, you need to determine the vibrational profile by using an accelerometer. To ease the task of profiling, Mide designed the Slam Stick, a data logger with the form-factor of the familiar USB stick, that measures acceleration in all three axes.
I first heard of the Slam Stick from the folks at Linear Technology, which provide the power management ICs for Mide’s piezoelectric vibrational energy harvesters. (I think Mide also sells their piezo elements without any power management devices if you want to do that part yourself.) The Slam Stick sounded like a great candidate for an EDN “Prying Eyes” teardown. Fortunately the device comes in a clear plastic case so I didn’t have to rip it apart.
The Slam Stick uses a Microchip PIC18F25J50 microcontroller, which is an 8-bit MCU with an integrated full-speed USB 2.0 transceiver and a 10-bit, 10-channel ADC.
An overriding goal for the Slam Stick was that it behave as a USB thumb drive. This presented a challenge. A PC expects to be able to read and write single disk sectors (512 bytes), but the minimum erasable block of most Flash memory is often much larger (for the part on Slam Stick, 4KB), and a microcontroller may not have enough RAM to buffer the difference. The workaround, say Tim Gipson, design engineer at Mide Technology is to store some of Slam Stick’s file system in the firmware and hand-craft it to ensure key FAT filesystem areas align on 8-sector (4KB) boundaries, with only the first sector containing live data. “We can also then generate file data on the fly when the OS requests it – for example, substitute the correct FW revision info on the fly when the config file data are requested. These tweaks allow us to use low-cost and low pin-count Serial Flash.” (AT25DF641 / SST25VF064 depending on availability).
The Slam Stick’s 3-axis accelerometer is the Analog Devices’ ADXL345. Gipson explains that using a high-Q resonating piezo beam to harvest energy requires accurate knowledge of the vibration frequency, ideally within 1-2Hz. While the ADXL345 generates its own sampling clock internally, this clock frequency can vary from part to part. So they run an accurate 32kHz oscillator along with it to determine its actual sampling rate and store a correction factor in the recording file, giving fractional-Hz accuracy.
The reverse side of the Stick shows its lithium polymer battery. The device charges in 1-2 hours after you plug it into a USB port when a green LED indicates it’s ready to go. Gipson says they evaluated some thin-form-factor supercaps early in the design, but the lower self-discharge and flatter discharge curve of the lithium polymer battery kept the design small and simple.
As it stands now, the software doesn’t tell you how much energy is being generated. I asked Gipson if that was in the works: Turns out it’s not such an easy feature to include. He said, “The answer to “how much power” depends on several factors beyond the frequency and amplitude. For example, the same piezo beam with a small proof mass at the very end vs. a large proof mass located centrally might be tuned to the same frequency, but have very different power outputs in the same environment.”
The software could scarcely be easier to use: I’ll include some screen captures in my next post of a simple data-logging run I made comparing running in conventional running shoes vs the newer minimalist shoes.