Variable frequency ultrasonic driver for resonant sonotrodes

[amaurot]

Hello,
I have built a Langevin type sonotrode using Ti 6-4 and piezo rings. The system should resonate at around 38 kHz and be below 100W of power.
I would like to build an arduino controlled universal driver that I could use to drive future sonotrodes:

• It should be able to reach 100-150 kHz
• Should be able to power about 100-150W sonotrodes working at resonance

I normally see a Q around 10-14 in these kind of sonotrodes so it would be ideal to have a circuit able to withstand 1000V or so (normally I drive the sonotrode below 100V).
I've been thinking about Mosfets, but I am not sure if they are right for this application. I understand I might have to use a few in parallel, what could be the implication of this?
Any suggestion?
Many thanks

[Marco]

Similar ready made versions have a resistance at resonance of less than 20 Ohm, with 1000v across it at resonance it would blow up.

How about a very fast class D amplifier? I think a class D amplifier can recover the energy from the parallel capacitance without a matched inductor, whereas other amplifiers can't. There are some other circuits too for charge recovery.

You could of course just do a square wave universal driver and forget about the charge in the parallel capacitance of the Piezo entirely, but it's not very efficient. Will spike current and burn power in the MOSFETs.

[ale500]

I had such a project but the sonotrode would resonate around 20 kHz. The "off the shelf" driver had a pair of IGBTs and a transformer. I used a class-D amplifier and a transformer, a ready made (almost) class D from TI, some TAS something, that should have provided around 200 W. The first transformer I did couldn't provide the power, saturated core.