A PZT transducer is used to convert a high frequency ( 100kHz) and high voltage ( > 1000V) electrical signal into a mechanical displacement. This high frequecy movements are coupled to the object to be cleaned by sub-merging both the object and the transducer in a bath of a suitable solvent (can be water)(The requirement is that the speed of sound in this liquid to be of similar magnitude as that in the transducer and object to be cleaned
The high speed and high energy movements remove unwanted deposits from the object to be cleaned.
What do you mean by detune? These can be thought of as a two port network. There is the mechanical resonance and the electrical resonance. You can change the mechanical side by loading the back of the transducer with material. The electrical side is modified with L or C in series or parallel with the two wires. Generally, mechanical and electrical resonance are at slightly different frequencies. You have to do some modification to the electrical side to bring them together and increase the efficiency of electrical to acoustic power conversion.
The maximum mechanical output is at electrical resonance frequency.
But, the resonance freq is sharp and because of sharp sides of the response curve mechanical output energy varies with even the slightest variation in driving frequency. I want to know how to flatten the response so that mechanical output is available for some + or - variation around the centre freq.
Mechanical loading is the only way, but this makes the efficiency equally poor over a broad frequency range.
Your best bet is to use some method of controlling the drive frequency. Measure the angle between the drive voltge and load current and adjust the frequency for zero degrees.
I am not shure that "The maximum mechanical output is at electrical resonance frequency." From my expirience i know that the maximum is somewhere between electrical and mechanical resonance.
Additionally, the resonance frequency is very different between transduser to transducer. So, i didn't success to get a good efficiency without some callibration.
You have to tune the electrical side (as described several posts above) to have the resonance at the same frequency as the mechanical one. Then you get even better efficiency.
In the megasonic area one company touts using the zero phase condition. Another company measures reflected power. So there are all sorts of ways of determining how to automatically tune the drive oscillator.
All of this is not just academics. For high power (hundreds of watts), the load on the amplifier off resonance will damage the transistors. This is why the electrical resonance is the one used to determine oscillator frequency.
Back in the days of valves/tubes the transducer was used in a one valve oscillator as the frequency determining element. You could get several watts into the transducer this way and the frequency automatically tracked the transducer changes.
A usual strategy is to find the frequency of series resonance (please see the papers). Samples of output voltage and current are taken and compared with a digital phase comparator like the one which is inside of a pll chip like 4046. The output signal is used to modify the frequency of the oscillator. As it was stated in posts before, it is a critical. An ultrasonic stack could have a Q as high as 1000 or more and their frequency change by several reasons but the most important is temperature. With one or more KW applied over the stack, any shift WILL destroy the output stage.