Driver circuit for piezo transducer

[AccountRemovedPerUsersRequest]

Long story short: any design ideas how to drive piezo transducer using 24V?

A

[Berni]

Anything that can drive a small 24V motor in both directions will also drive a piezzo just fine.

The thing to keep in mind with piezzo speakers is that they are mainly capacitive loads. So you can't just pulse 24V on it and have it beep like a regular speaker does. You need to actually connect it to 24V, then short it to ground to discharge it, then apply 24V again. Or to get even more output out of it you can use a H bridge to apply 24V then swap the polarity around to give it -24V. This gives you effectively 48V peak to peak across the piezzo

[AccountRemovedPerUsersRequest]

Hi,

Yes, in deed. I have tried a few approaches and this (=attached image) is what does the job. However, I would like to have something a bit more sophisticated. I am actually using a fet instead of transistor (despite the circuit). C1 represents the piezo element - I have modeled it as a capacitor in this model.

A

[mikeselectricstuff]

Either an H bridge ( and voltage booster if required), or open-collector driver and inductor ( 1-10mH) to +ve with the transducer across the inductor.
For a low-cost software-controlled solution, it can be done with 5 transistors and a few discretes - flyback boost stage and a discrete  H-bridge

There are some chips made for this, e.g.
https://www.mouser.co.uk/new/diodes-inc/diodesinc-pam8904-driver/
https://www.onsemi.com/pdf/datasheet/fan8841-d.pdf
https://www.njr.com/electronic_device/PDF/NJW4814_E.pdf
https://www.ti.com/lit/ds/symlink/drv2700.pdf?ts=1626923693593&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FDRV2700

Some small H-bridge motor drivers can also be used.

[Berni]

One trick i like to use for compact power drivers are mosfet gate drive chips like this one https://www.digikey.com/en/products/detail/microchip-technology/MIC4223YMME/5701922

They are essentially a miniature half H bridge inside since they are also designed to drive capacitive loads such as the gate of a mosfet. Only issue is that its difficult finding ones that go to 24V since mosfet gates usually blow up at that point. But for driving small 12V loads like solenoids and the like i really like using them.

If you want a cheap and expandable solution you might look into adding a PNP/NPN pair power stage between the piezzo and your existing transistor. That way you can increase the value of the pull up resistor to create a weak signal going between 0V and 24V then run then use that PNP/NPN pair follower to boost the current of that signal to something powerful enough to drive the piezzo at the current you want towards both 24V and 0V.

Here is an example of a PNP/NPN follower output stage:

[Kryptychon]

Hi,

Berni made two nice proposals.
If you're looking for a MOSFET gate driver chip which can handle 24V, have a look at TLP250. It's not cheap, but easy to use and can handle up to 35V.

The cheap and simple solution is the shown totem pole configuration.

Regards
Kryp

[Zero999]

Swap the transistors round, so they're in common emitter configuration and AC couple the signal to the bases, via capacitors. Here's a circuit I designed for 12V operation, but will work fine at 24V. The input musn't be allowed to float, otherwise both transistors will turn on simultaniously and go pop.


Push-pull driver AC.asc

[AccountRemovedPerUsersRequest]

Hi,

Very nice circuits and ideas. Thanks!

General:
A single chip solution is preferred. Price is not a limiting factor.

Zero999:
Is the C1 and RL together representing a piezo element or is the element driven thru C1?

Kryptychon:
TLP250 looks good although it is an obsolete product. Maybe there is are newer versions available.

Berni:
NPN/PNP follower is what I am going to try next since I do not have any integrated chips in hands at the moment.

mikeselectricstuff:
Thanks Mike (and thanks for the interesting content you have shared on your channel for a long time!). DRV2700 looks like a nice option, but isn't it boosting >100V. Maybe there are similar below 36V] which is the absolute maximum.

Hopefully chip shortage is not affecting too much..

A

[AccountRemovedPerUsersRequest]

In practice, the version on left results almost twice as strong signal as the version on right (attached image). The fet is 7002 and transistors are BC327/337.

[TimFox]

Depending on how much voltage and current you need, where the current depends on frequency for a capacitive load and constant voltage drive, have you tried a simple 555 (BJT version) astable multivibrator?  The output totem pole has a substantial current capability, > 100 mA with Vcc = 15 V.  Max rating 18 V, recommended max 16 V, in TI data sheet.

[floobydust]

Long story short: any design ideas how to drive piezo transducer using 24V?

What frequency?

[AccountRemovedPerUsersRequest]

200kHz

[TimFox]

What capacitance?

[Zero999]

Hi,

Very nice circuits and ideas. Thanks!

General:
A single chip solution is preferred. Price is not a limiting factor.

Zero999:
Is the C1 and RL together representing a piezo element or is the element driven thru C1?

It was designed for a different application: a class D amplifier driving a resistive load.

In the case of a piezo, you could just connect it in place of the resistor and capacitor, as piezos are capacitive, thus block DC. One thing to bear in mind is that it's often not recommended to have DC bias on a piezo, so you might want to add a capacitor anyway.

What voltage swing do you require? The circuit I've posted is a half bridge, which will drive the piezo with a peak voltage of 12V, or 24V peak-to-peak. If you want a peak voltage of 24V, use a full H-bridge.

[AccountRemovedPerUsersRequest]

The capacitance is 2nF @ 200kHz.

I changed the driver transistor Q1 to 2N3904 and pulse started to get its shape. Also the amplitude is the same between the fet circuit and NPN/PNP version. This is good.
To get more juice, I need to study more about suggested single ship options (with boosting).


A

[Niklas]

I vote for the push-pull circuit that Zero999 showed first, but with some modifications.
- Add diodes across the base-emitter junctions, but pointing in the reverse direction. This will protect the transistor and speed up the switching.
- Add diodes across the collector-emitter on both transistors to provide free wheeling. Similar to the body diode in a FET.
- Adjust the coupling capacitance to match the switching frequency and the transistors. 1uF will probably not work for 1 MHz, but a ceramic 1nF will be closer to what you need. Double check that the capacitance is enough to keep the transistors on for the needed period time.
Also the 1k in series with the input signal might be too much, try 47-100 ohms.
- Consider to add a small resistor, 5-10 ohms, and a matching inductor in series between the driver midpoint and the piezo. With 2 nF you can start with something in the range 10-33uH and observe the waveform at the excitation frequency. The driver will give a square wave output and you want a sine wave over the piezo. I have seen more than 45 V over a 1 nF piezo at 1 MHz with only 15V supply for the driver.

[Zero999]

I vote for the push-pull circuit that Zero999 showed first, but with some modifications.
- Add diodes across the base-emitter junctions, but pointing in the reverse direction. This will protect the transistor and speed up the switching.
- Add diodes across the collector-emitter on both transistors to provide free wheeling. Similar to the body diode in a FET.
- Adjust the coupling capacitance to match the switching frequency and the transistors. 1uF will probably not work for 1 MHz, but a ceramic 1nF will be closer to what you need. Double check that the capacitance is enough to keep the transistors on for the needed period time.
Also the 1k in series with the input signal might be too much, try 47-100 ohms.
- Consider to add a small resistor, 5-10 ohms, and a matching inductor in series between the driver midpoint and the piezo. With 2 nF you can start with something in the range 10-33uH and observe the waveform at the excitation frequency. The driver will give a square wave output and you want a sine wave over the piezo. I have seen more than 45 V over a 1 nF piezo at 1 MHz with only 15V supply for the driver.

Those are reasonable suggestions, but I'm not convinced about #1, about the base-emitter diodes, which I think would slow down the turn-off time and increase cross-conduction. The reason why I selected resistors was so the base-emitter junctions become reverse biased, which should speed up the turn-off time. Replacing them with diodes would clamp the reverse voltage at 0.6V.

[Niklas]

We added the diodes mainly as a protection for reverse biasing of the base-emitter diode, but it also helped us to get better symmetry and speed on the output. Changing the supply voltage and the transistors shows that some combinations has no need for it, but some are just stopping up for the first switching periods. Try to disconnect the diodes in the attached LTSpice file and see the difference.

[Berni]

Ah 200KHz is pretty high. I assume this is then some sort of ultrasonic sensor.

At those speeds large BJT power transistor can indeed become pretty slow for the turn off part, but the power levels here don't sound particularly high so it should be fine with pretty small transistors.

For driving piezo transducers at really high powers (>300W peak) the way to do it is to find a center tapped switchmode transformer, connect the center tap to the supply voltage with a big capacitor then use two beefy low Rdson mosfets to alternatingly pull the ends of the primary to ground. The piezzo then connects to the secondary, this can adjust the number of turns to get any voltage you want(some piezzos need like 300V or even more). Due to the low losses and the transformer used as a forward converter topology means you can push huge amounts of power trough it in peaks. The solution has a low component count (1 transformer, 2 power transistors, possibly a gate drive IC) and provides a nice symatrical drive waveform with no DC bias.

[Zero999]

We added the diodes mainly as a protection for reverse biasing of the base-emitter diode, but it also helped us to get better symmetry and speed on the output. Changing the supply voltage and the transistors shows that some combinations has no need for it, but some are just stopping up for the first switching periods. Try to disconnect the diodes in the attached LTSpice file and see the difference.

Don't worry about damaging the base-emitter junctions, by reverse voltage, because the drive voltage is only 5V, which they can handle, but I can now see why the diodes are a good idea.

Adding D1 and D2 provide a return path for the base current. Without them, the return current has to go through the base-emitter resistors, which could be reduced, but they would then rob more base drive current. It means you can increase the base-emitter resistors to a much higher value. The base-emitter diodes should be changed to small silicon diodes, such as the 1N4148. The PMEG6010AED has a higher capacitance and the lower forward voltage drop is counter-productive, as it slows down the off time.

[Niklas]

I changed the transistors in the simulation to something that matched with the higher supply voltage. For the original 1 MHz ultrasonic application, with max 15V supply,  I have BC856B and BC846B transistors and the diodes could be BAT54 or similar. They will switch up to a few MHz, but the matching network is less optimal then.

[AccountRemovedPerUsersRequest]

For a single chip solution this looks pretty good (and is available across distributors)
https://www.analog.com/en/products/lt3469.html#product-overview

Using 10X gain the min load capacitance is not in range so I need to do something for it.

A

[Zero999]

For a single chip solution this looks pretty good (and is available across distributors)
https://www.analog.com/en/products/lt3469.html#product-overview

Using 10X gain the min load capacitance is not in range so I need to do something for it.

A

It's completely unsuitable. The maximum supply voltage rating is only 16V, for a start.

If you'd rather use ICs. Try a MOSFET driver. The TC4431 & TC4432 can be used to make a full H-bridge driver, for the piezo.
https://ww1.microchip.com/downloads/en/devicedoc/21424d.pdf

[AccountRemovedPerUsersRequest]

>It's completely unsuitable. The maximum supply voltage rating is only 16V, for a start.

Completely unsuitable? Well...
Is able to generate max. 33V from 2.5-16V which is more than fine to me if I want to have more juice than I can get from 24V. Any other concerns?

TC4432 is a nice option, in deed, if 24V is enough. It might be.

A

[Zero999]

>It's completely unsuitable. The maximum supply voltage rating is only 16V, for a start.

Completely unsuitable? Well...
Is able to generate max. 33V from 2.5-16V which is more than fine to me if I want to have more juice than I can get from 24V. Any other concerns?

TC4432 is a nice option, in deed, if 24V is enough. It might be.

A

Sorry, I should have read the data sheet. I just read the description,

EDIT:
It looks good. Probably overkill as it gives a nice linear output and I thought you just wanted a squarewave,