Author Options:

Universal ultrasonic driver circuit - help required Answered

I would like to build a few, properly working, ultrasonic devices.
For example an ultrasonic soldering iron and an ultrasonic soldering bath.
But some small ultrasonic plastic welder or cutter is nice too :)
If you ever had one of the above to play with you know why they are great to have.

The development story so far:
I managed to destroy several driver boards.
The ones you find for cheap with 28 or40kHz transducers in your favourite online store.
In the beginning I knew I will have a need to repair or replace these boards but no clue why.
Take an ultrasonic cleaner and read the manual.
There it is always pointed out that a low water level can destroy your toy.
What does that exactly mean?
The transducer needs to be kept in resonance, if the water level is too low or something havy sits right at the bottom of the tank the frequency drifts off too much.
Very expensive untis can cope a bit better here, which gave me the idea for the universal driver.
During my experiments with hoorns I noticed that it is very hard to get usable results without extensive computer simulations first.
Just one mm too long or too short and literally nothing happens, go a bit further and a thin aluminium horn might start to crack under the stress.
And in all these cases the driver overloads, in one cheap case to the point that the transducer fused together.
Trying to examine these driver circuits while they operate turned out to be a total nightmare!
Place the probe from the ocsilloscope literally anywhere and the thing goes out of tune already.
By the way: Never coil up the wires going to your transducer.....
Only way I found that somehow works is by adding a tiny transformer around the wire going to the transducer and to measure the voltage generated there.
To make it short: Destructive testing provided the requirements a driver needs to match to keep the cost low.

Reasons for the premature death of cheap driver boards:
Almost all of these cheap drivers I could find generate the 28 or 40kHz signal from the mains voltage.
Means it goes through a transformer to get the desired 50-80V and some witchcraft turns that into a more or less smooth DC voltage.
This is then switched by some beefy transistors, mosfets or similar, depending on the circuit.
The actual feedback happens with a tiny ring toroid, similar to what you use to drive a ZVS system.
With this dirt simple design a fully tuned transducer - like when nothing is attached to it yet - would cause the driver to provide a voltage of about 6x of what the transducer is rated for.
Thankfully in most cases the transducer survives this a couple of times while the transistors fry within about 3 seconds no matter how good the cooling.
Slightly out of tune - like when mounted onto a cleaning tank - the resonant frequency is slightly off the tuned 28 or 40kHz.
The driver compensates this through the tiny feedback transformer.
But this only goes for a about 1-4kHz, drift away further and first the power drops, then the voltage spikes and it dies.
The feedback is not able to shift the generated frequency enough as it is ultimately derived from the mains frequency of your grid.

Reasons why a dedicated, low cost driver would open new possibilities:
Imagine you need to make a horn or sonotrode for your transducer.
Knowing that each half of it should be equal to a quarter wavelength of the operating frequency is nice and easy.
But if you add something like a blade for cutting or you need some pressure for welding then calculated dimensions become useless.
Programs to fully simulate complex sonotrode designs, especially if you need to add screws or blades are costly and out of reach for most of us.
Even if you would have access you still need to know the material properties to know the speed of sound in the material and how much it can flex in various directions without being subjected to metal fatigue.
For basically all hobby needs in terms of ultrasonic gadgets we are happy with a simple push pull motion.
the same motion our transducer offers by default.
And when it comes to attachments it turned out that quite stubby horns of light weight are a good compromise already.
A 50-50 ratio of diameter and length works reasonably well in most cases.
For example the standard 40kHz transducer of 45mm diameter is quite happy to work with a horn like this:
45mm diameter on the thick end, 20mm diameter on the tin end.
Thick part 40mm long, thin part 42mm long.
The extra 2mm are for the manual tuning by filing or sanding it off until there is good cavitation happening when you put the end into water.
This however is only good for simple testing purposes and some fun but as soon as you attach blades or a small pot with about 200grams of molten solder the tuning is way off and destroys the driver quickly.
To be able to deal with different pressure levels on the working end or just a different mass that is attached the driver needs to "know" the new self resonant frequency.

Basic idea for a dedicated driver:
Please bare with me on this one as my developing days got severly neglected once I moved to the other side of the globe....
Input should from a 12V power supply, preferably a PSU to keep costs and sourcing time low.
The operating voltage for the transducer shall come from a simple switch mode supply.
I was thinking of scrapping a PSU for the transformer and switching transistor.
This however would provide about 120-160V on 240V mains with the transformer of a PSU.
To match the required load changes it would be great to drive this first transformer by PWM means to regulate the output voltage with a potentiometer while keeping it steady within the set values.
Basically like every cheap phone charger but with an output voltage that can be adjusted and kept regulated.
The switching transistors for the transducer should be well over the required specs of an out of tune transducer.
I guess capable of switching 600V should be sufficient.
Main design change to the cheap driver boards would be the feedback.
A hall effect sensor could provide the proportional voltage to the current going into the transducer.
It would also provide the real operating frequency of the transducer for the feedback loop.
The resulting real resonant frequency of the running system is then used to drive the switching transistors.
As a result the transducer would always be driven at the exact right frequency no matter the load on the working end.
These transducers still have a quite limited frequncy range due to the fixed counterweight on the back - it is optimised to be self resonant without the transducer being mounted.
To explain this feature let me use a spring with a weight on it....
You can move your hand up and down to make the weight swing up an down with the spring force.
You can also push the weight to get the same effect.
But if the weight would just expand and contract there would be no change in the spring force or position or the weight.
Our transducer however is mounted to something and the weight on the back is heavier than what is on the front end of the transducer.
As a result the weight is pushed back and forth and because all is fixed together this movement is transfered to for example your cleaning bowl.
Without anything attached to the transducer it would literally start to rip itself apart until either the bolt or the ceramics fail.
The feedback loop needs to prevent this by adjusting the switching voltage going to the transducer.
Once too far out the system needs to shut off until it can reset.
The frequency control is not that fragile.
With the power controlled through the feedback even a wide drift in the operating frequency of about 5kHz would only reduce the effectiveness and amplitude of the moving horn/sonotrode.
Sadly my skill set in circuits is not that good anymore to have the required parts in my head and to know how to combine them properly :(

Why this concept is only really good for really basic applications:
Professional solutions utlise often less than 20W of ultrasonic power for a soldering iron or scaler.
For these devices the sonotrode/horn is spefically designed for the task at hand.
Same goes for any possible attachments - without them these things don't do much at all.
Finding these low power ceramic transducer rings for a good price is hard enough, making an amplifying horn even harder.
But when using these quite big 50 or 100W transducers we find for cheap online we can compensate the lower amplitude with the added power of the transducer.
Since we only need surface action but won't have to go through a few liters of liquid it might even be beneficial.
Fun fact: A 40kHz transducer has the second harminc frequency at about 170kHz.
Means we could design a driver for the second harmonic and enjoy total silence when working with it.
Would also mean that the ultrasonic power would be much higher.
Mass times acceleration and such things ;)
If you want some ultrasonic cutter then you don't want to waste weeks and lots of money trying to come up with a working attachment to your transducer.
Just keep it as short as possible and with about the same weight as the front part of the transducer.
At least the driver desing would make it quite easy to design an amplifying horn by trail and error through reducing the lenght of the thin end until it really fits.

Anyone with good circuit skills willing to volunteer? ;)



4 days ago

These cheap, chinese, Tesla coil kits you can find online provide about 100 to 200W in power.
The feedback is usually done through an antenna wire going into a small toroid transformer to provide a signal for the electronics.
Funny thing about these circuits is how powerful they are and how little energy is wasted on the switching transistors or mosfets.


4 weeks ago

I might have found an easy way out...
That is if the basic idea behind is feasable enough to be used.

These cheap, chinese, Tesla coil kits you can find online provide about 100 to 200W in power.
The feedback is usually done through an antenna wire going into a small toroid transformer to provide a signal for the electronics.
Funny thing about these circuits is how powerful they are and how little energy is wasted on the switching transistors or mosfets.
If you add the fact that their frequency range is surprisingly wide then it could be misused....

The circuit already drives a coil at resonance.
And almost all simple ultrasonic transducer circuits use a transformer on the output to drive the actual actuator plates.

Usually these Telsa coil drivers operate anywhere between 100 and 400kHz - depending mostly only the primary coil of the Tesla transformer.
A standard 40kHz transducer has its second harmonic resonance frequency around 170kHz - right in the ball park for these Tesla coil drivers!
The output mosfets or transistors switch a rectified mains voltage directly to the Tesla transformer.
Without a variac I see two possible options to join the two:

Option A:
The driver is connected to an air core transformer to provide the required 40-80V on the transducer.
Added capacitors would help to create better sine wave and a bit more power.
The feedback antenna could be replaced with a hall effect sensor to monitor the tranducer directly.

Option B:
A ferrite transformer for the 170kHz range is used to drive the transducer.
The feedback coul be taken by an added wire loop on the transformer or again with a hall effect sensor.

Problems I could imagine....
As we no longer drive a resonant Telsa transformer the circuit might struggle to start swinging at the desired frequency.
It might be required to make some modifications so the output is driven to start directly at 170kHz and then switches to the self controlled mode with feedback.

Benefits of an abuse like this:
The Tesla coil driver circuit is able to keep and track the resonance.
With the feedback directly from the transducer or driving transformer and minor outside effects will be compensated.
For example:
Adding a soldering tip to a scaler handpiece would change the resonant frequency of the system slightly.
But since we track this in real time our tip would still swing as planned, just a bit off the originally designed frequency.
Or the unavoidable contact.
Adding pressure in the direction of the tip movement is a slight problem for a scaler driver.
Adding sideways pressure is bad.
The Tesla driver however would just "see" the frequency is now way off and change it to optimal parameters that match what is happening right now.
Means, despite the contact the tip will still produce the desired cavitation effect.

Any thoughts or ideas about this?

Jack A LopezDownunder35m

Reply 24 days ago

I have noticed this habit, in your posts, of alluding to things, without ever, or rarely, providing a link to an example things. For example:

"These cheap, chinese, Tesla coil kits you can find online provide about 100 to 200W in power. The feedback is usually done through an antenna wire going into a small toroid transformer to provide a signal for the electronics."

It is as if you think everyone else has seen, knows about, all the things that you have seen and know about, like, as though everyone knows everything.

Anyway, I am not asking for an example link to a product page on eBay, or wherever.

Instead I wanted to give you a link to this page at uzzors2k.4hv.org


in particular, at this circuit diagram,


The tuning for this SSTC (solid state Tesla coil) is done by way of a PLL (phase locked loop), in particular the CD4046. It also gets its feedback signal by way of a stubby piece of wire, acting as an antenna.

Anyway, that IC has a VCO (voltage controlled oscillator) inside it, and there is some method to set up the range of frequency this oscillator runs at, using a handful of external resistors and capacitors.

The reason I mention this, is because you basically want to something that can tune your driver for your ultrasound transducer, and you are speculating about what frequency range your Chinese Tesla-coil kits can achieve, but I think if you can get your head around the CD4046 PLL, and how it works, to produce an oscillating signal locked in frequency and phase, to some signal it is listening to...

Well, I think if you can understand how Uzzors2k used a CD4046 to drive a Tesla-coil, that this will be the same, or better than, understanding how the Chinese Tesla-coil kit manufacturers are doing it.

By the way, Uzzors2k, said his circuit is the same as one discovered by Steve Connor, of scopeboy.com. That link is dead, but archive.org has some copies of it, e.g.


Did I mention you can set the frequency range of the CD4046's VCO, to like whatever you want? Erm... almost. Uzzors2k mentioned a maximum frequency of 2.7 MHz, and I am guessing that is a quote from the datasheet.


The frequencies you mention for your ultrasound transducer, e.g. 40 KHz,or close to that, are easily in this wide range, the whole range the 4046's VCO is capable of; i.e. 0 to 2700 KHz. I am naively guessing you can narrow it down to whatever custom range you want (e.g. 10 KHz to 60 KHz, or 35 KHz to 45 KHz, or whatever), through choosing some appropriate external resistors and capacitor.