50KHz ultrasonic driver techniques 1

[Electricliff]

I have on my bench a stainless steel plate with a piezo ceramic transducer (50mm diam 5mm thick) bonded to it.
I need to build an efficient driver to pump power at 50KHz into the device. The end product will deliver ultrasonic energy into a glass container to excite particles suspended in a liquid.
The transducer is a two wire device.
Initial tests show the structure is sharply resonant at about 50 and 100KHz at which frequencies I detect activity in the liquid.
This activity dies if the frequency shifts by more than 0.2% either side of the resonant frequency.
Temperature changes in the stainless steel and transducer assembly can shift the resonant frequency by more than 0.2%.
Are there any text book techniques which I ought to know before designing the driver?
Suggestions welcome!
Electricliff

 

[nbucska]

Build an oscillator with the transducer as frequency determining element.


<nbucska@pcperipherals.com>

 

[Electricliff]

Thanks nbucska.
That was my first thought and the first attempt.
Unfortunately it would need to be a power oscillator because the drive requirement is about 30 Watts.
Have used a power amplifier to drive the piezo element and find that at 56KHz we get the desired activity in the specimen.
Unfortunately the resonances in the system (and there are a few!!)are not at this frequency. Do you call it Sod's law or Murphy's Law?
It looks like we are stuck with either driving our &quot;off the shelf&quot; unit with a home brewed power amp (the one we used for the experiment is very nicely boxed and full of expensive looking bits!!!)or hunting for a piezo ceramic assembly with a natural resonance at 56KHz.
I favour sticking with the &quot;off the shelf&quot; unit and designing a minimalist power amplifier driven by a tweakable front end oscillator.
I guess my real problem is knowing how minimalist you can be with these power piezo devices.
Do I need to concentrate on a low distortion push-pull design capable of emulating the expensive box we used for the experiments? Or can I just hit it with high energy pulses and let the properties of the piezo element convert these into the required input to the liquid?
I like you are semi-retired and have a few patents to my name, but ultrasonics has never been part of my love affair with electronic engineering.
Just thought someone on the Eng-Tips site might have experience of these things and point me in the right direction.
The problem was thrown at me by my old firm, who think I have time on my hands to pursue these things without it costing them an arm and a leg! And I like the challenge anyway.
But the solution must be inexpensive so I guess I need to start experimenting with single ended transformer coupled pulse generators and find out what happens.

Thanks for your input.

Electricliff

 

[nbucska]

There must be a strong resonance at the preferred frequency of the transducer but it may change due to temperature, accoustic loading/damping etc. -- so it is desireable to
use this as frequency determining element. The other
resonances you can suppress. I suggest you look up
older patents at <

http://www.uspto.gov>

or other literature.

I don't think distortion is critical, Class-&quot;C&quot; should be
OK.
If you want a frequency different from the self-resonance
of the xducer, it means MUCH more power loss...


<nbucska@pcperipherals.com>

 

[Electricliff]

Thanks for that info.
The frequency does shift with temperature and not surprisingly the temperature rises due to the 30 watt input.
I will start searching the data base!
Thanks again
Electricliff

 

[mws]

I have a project where I'm needing to build an ultrasonic horn to try to create microdroplets in a fluid stream. From what I've seen I think you need to get a crude model of your piezoelectric system's parasitic components, C and L, and then add an appropriate L to the driver output to set it's resonant frequency at the frequency of intrest.

 

[zsgunay]

This is an old posting, so before i suggest the solution, I would like to know if you are still looking for a solution. Please reply here.