Ultrasonic Power Transducer Driver for Ultrasonic Cleaning

I assume you're using an H-bridge or other switching design to drive the transformer? (I saw another wag was planning on using an automotive hi-fi amp, sheesh!) If it's properly designed (snubbers, Schottky clamps across the FETs, etc.), the driver shouldn't have any problems with strange loads, out-of-phase currents, etc., and the transformer shouldn't mind either. I speak from experience with a 500W 600kHz design. Blast away.

No. H-Bridge is not used. I am using a push-pull arrangement for transformer coupling.

Very first experiment I did with 1kV 12A fast MOSFETs and they did work for few seconds or minutes and then failed all of a sudden. Perhaps they are not proper for the power in use.

In next step I used IGBTs of 1200V and 40A and then they did work for 30 minutes and then some of them started getting damaged.

All these were mounted on heat sink and I used cooling fan to cool the heat sink.

At high frequencies, most of the MOSFETs work fine but typically at 25kHz range IGBTs are most suited.

I have 300V DC from which I have push-pull transformer driver IGBTs of 1200V each and in next step I will like to see if 600V IGBTs can work. There is a chance of these getting damaged from induced voltage reaching above 600V for which protection mechanism is planned.

Problem is not with this part of the circuit. Driver transformer output coil has series L and Load capacitance of Ultrasonic transducer almost 33nF. When transducer produces its own voltage, it deforms the filtered sine wave output. This what I am concerned about. There is no forced drive voltage at the filter stage and input to the filter is almost a square wave and output sine wave due to LC resonance.

Transducer generates its own energy from changing water pressure and this energy gets mixed up with transducer driver output and then waveform becomes nasty standing wave mixed waveform. This also happens in wireless transmitters due to energy reflections. here reflections are due to transducer itself. Unless driver can work against this energy, waveform will remain of mixed energy type.

In a properly-designed driver the switches won't get very hot. An H-bridge has the advantage that you firmly control both polarities of the waveform, whereas with a push-pull center-tap transformer leakage-inductance flyback is uncontrolled. Wiring inductance is a serious issue, along with directing the currents to R+C snubbers, otherwise the high dI/dt currents produce excess voltages that can damage the gates.

I don't understand why you want waveform purity, won't the fundamental and maybe other parts do their job in creating cleaning turbulence?

Transformer coupling gives isolation and series inductance of secondary reduces high frequency component.

Transducer is having high capacitance and can get damaged if square wave is injected. Entire energy will go waste in the transducer.

Transformer coupling is used in many H-Bridge drivers also. It is not very different from push-pull. H-Bridge Motor driver and this Ultrasonic driver are almost similar type. Motor also generate back emf.

Yes, yes, transformer coupling to the transducer, with a series inductance to eliminate the square-wave aspect, and do resonance to further magnify the voltage.

You may have not been able to read my drawing (CR4 dramatically reduces the resolution), but e.g., I used 2uH of series inductance, or 4 ohms at 300kHz. My transformer had further leakage inductance playing the same role. It's this increasing Zout at high freq that allows the transducer to reflect back unusual voltage waveforms. I think by definition if it's responding to the external pressure transients by showing corresponding piezo voltages, it's happy and not threatened.

My friend suggested that to break the effect of standing wave, 100Hz AM modulation is to be used. This perhaps may relate to the reflection arrival time for wavefront from vessel boundary to transducer which is responsible for causing sensed energy in the transducer. During the forced power for short period, waveform deformation does not take place.

This idea is not bad and I can easily sense the increased energy due standing wave and break the oscillations for 5ms or 10ms and then again pump energy for same interval.

I think this low frequency modulation to be selected for the size of he vessel. I have to first experiment with this idea but sure it makes a good sense.

With respect to your answer for transformer coupling and leakage inductance, I do take care by snubber circuits and current limiting circuits to keep peak waveform under control.

I have recently procured 199kHz ETD59/97 cores and these perhaps may inject greater power into load.

I have two options to use in IGBT voltages, 600V and 1200V versions but 600V have greater current and power ratings. I am bit worried about 600V versions from back emf from transformer coil switching.

I also face problems related to mains voltage fluctuations. Mains voltage fluctuates a lot here from 160V to 275V and sometimes may exceed 300V also for short time.

I'm that wag!

Actually I have started looking into radio amplifiers now .

I'm trying to build a monster.. muahaha!

(1kw-1.5kw)

Nothing like taking a bunch of used crap and re-purposing it for a new life. I enjoy doing things that some say can't be done. However, I'm just in the "Thinking" phase of the project.

At high amplitudes transducers are inherently nonlinear. Coupling to a mixed phase medium causes mismatch, with the attendant reflections. No clean waveform possible.

You are right.

I do not have much details of transducer other than rated operating power which is 50W each and there are 10 numbers in the Assembly. These are all fired in same phase. I am not sure about the linear power zone and peak power transducer can tolerate. I am limiting maximum power to 500W for this assembly.