I just discovered this part recently, they are quite interesting and I can imagine all kinds of neat applications for this device.
The principle of operation is very similar to a piezoelectric beeper, only the operating frequency is an inaudible 25 kHz, the unit has a diaphragm that is excited into vibration by a piezo-ceramic disc, the vibrations induce an airflow through the device at an impressive rate of up to 1 liter per minute and can produce pressure differences up to 2 kPa. (not at the same time!) It's also TINY! you can fit this device into tight spaces and do things that would be very difficult any other way.
I ordered a few samples from Digikey. ($15.98 each) The first look at the data sheet reveals a sparse set of specifications, not enough information to really understand what the thing is, or how to use it - a pathetic data sheet to say the least.
The data sheet shows the resonant frequency of 24-27 kHz, with a recommended drive of 20 V peak-to-peak, what they don't tell you is the fact that the thing must be driven exactly on resonance to perform well, if you are off-resonance more than 2%, the performance is miserable. I don't understand why they don't make an effort to explain this critical point in the data sheet!
I quickly realized that the only way to really utilize this part is to drive it exactly on resonance, I set out to design a simple auto-tuning driver that could keep the drive frequency dead-nuts on resonance, without adjustment.
Murata does supply a driver reference design that does this, but it is a half-bridge that requires a 20 volt supply for full excitation, I want to use lower system voltages so I opted for a full bridge design. The next step will be to design a 5V version that uses a transformer to boost the output voltage to 20V.
The circuit is based on a PIC16F1455 MCU that drives a simple transistor H-bridge from 12 volts to provide the drive, a sampling circuit analyzes the voltage waveform on the ceramic resonator to detect the resonant condition, adjusting the MCU's OSCTUNE register to tweak the frequency into perfect resonance.
The 25 kHz drive is bit banged, along with pulses phased to sample the voltage on the resonator at two points in each half cycle, the voltages are sampled and stored on two capacitors via a 4066 analog switch, an interrupt routine samples these two capacitors every few seconds to determine the phase relationship between the resonator and the imposed drive signal. When the part is in perfect resonance, the back emf is 180 degrees out of phase with the drive, in this condition, the voltages on the capacitors are about equal, the drive simply tweaks the MCU's clock until the voltages match.
A few subtleties:
The sampling cycle interrupts the drive, I balance the timing so when the drive turns on again, it's still in phase, otherwise it would make an audible 'click' and it would also snub the resonator until the phase sync'd up again.
The program samples less frequently when the drive detects resonance, more frequently when it's way off- so it converges quickly.
I added a GATE input to turn it off without losing the tuning parameters, this makes it start instantly, on resonance.
Now that I have a neat way to get the most out of this cool new technology, I will be exploring new ways to actuate mechanisms, move fluids and much more.