How to amplify a square wave cheaply?

[plukens]

I am driving a piezo disc at 100 Hz (or less) but am limited in voltage by the output of my signal generator to +/- 10 volts (20 volts total). I need to amplify the voltage to 50 or even 100 volts. What is a cheap way of amplifying the square wave signal (at 100 Hz) produced by my signal generator?

Will an audio amplifier work? They are designed for low impedence loads (8 ohm for instance). How will an audio amplifier work with a high impedence load? (my piezo disk)?

I'm looking for a solution that costs $50 or less. OK, $100 max.

Thanks!

Peter

 

[alec_t]

If you have an audio amp why not suck it and see? How many watts are you planning to pump into the piezo?

 

[ronsimpson]

Do you want to make or buy it?

 

[plukens]

I do not have one, but know they are inexpensive and, in principle, they amplify voltage. But don't want to waste my time if it not a worthwhile approach. I know from the literature with my piezo disk that I need to drive it at 50 or more volts.

Thanks for your assistance.

 

[plukens]

Do you want to make or buy it?

I prefer to purchase something that can be driven by my signal generator. However, waveform amplifiers seem to cost quite a lot.
Your suggestions are welcome! So far what I've heard is:

1. Use a NPN transister to amplify the voltage (not my first choice to build a device from scratch due to limited knowledge and materials/tools etc.)
2. Use a MOSFET with my signal generator's square wave switching it on/off in conjunction with a higher voltage supply.

 

[WTP Pepper]

Since you are running at such a low frequency with such a high impedance, the current drawn is likely to be very small. Build a boost switched mode PSU to run your switching transistor/FETs off it.

Or you could drive a mains transformer in reverse and use the output to switch a transistor to revert the signal to a square wave.

 

[plukens]

Since you are running at such a low frequency with such a high impedance, the current drawn is likely to be very small. Build a boost switched mode PSU to run your switching transistor/FETs off it.

Or you could drive a mains transformer in reverse and use the output to switch a transistor to revert the signal to a square wave.

Thanks, but am not familiar with the things you mention. Yes, the impedence is very high so low current. And the freq. I'm after is only 100Hz. I'm driving a piezo droplet generator. I'm seeking an off the shelf solution that can be driven by my signal generator. Thanks for the help.

 

[audioguru]

Most piezo transducers do not work well below their resonant frequency of a few kHz, not 100Hz.

 

[plukens]

Most piezo transducers do not work well below their resonant frequency of a few kHz, not 100Hz.

Thanks for the comment, but no sure what that means. Piezo transducers work at all frequencies. This disk is a part of a droplet generator that produces individual droplets (single pulse) or trains of droplets (in the Hz regime). I am not trying to achieve resonance. Driving it with a square pulse helps droplet formation. The displacement of the disk is a function of the amplitude of the pulse. 30 V seems to be the target identified in the literature, thus my need to amplify from the 10V my function generator can produce.

 

[audioguru]

Here is the frequency response graph of a name-brand piezo transducer (speaker, not beeper).
Its level is all over the place due to its strong resonance at 5kHz and smaller resonances at other frequencies. Look at the notch at about 8.5kHz.
Its outout drops 20dB or more just a few hundred Hz away from resonance and the output below 1kHz is almost nothing.

 

[RCinFLA]

Piezo ceramic is made by static charge polarizing polycrystaline ceramic. It is the electrostatic equivalent of charging a magnet. Usual chargng field is about 4 kV per mm thickness of ceramic.

When electrodes are place across the polarized ceramic is will compress or stretch depending on applied voltage relative to polarized direction. As with most mechnical devices there will be a resonance point that will create a conservation of energy resulting in a build up in the physical excursion.

If the resonator is placed in a highly dampened environment (acoustical coupling) the the resonator will de-Q. You get less excursion but wider bandwidth. This is what most medical ultrasound imaging machines do. They operated at a loaded Q of 1 to 2 versus unloaded resonator Q of about 30.

You can make a low frequency ceramic piezo transducer. It is just a matter of size and mix of ceramic. Anyway, in your application creating water droplets it is likely pretty well dampened. Just about all ink jet printers use piezo transducer to 'spit' out droplets of ink.

As to audio amps, most will be okay with a high impedance load. Some without a minimum load may have high frequency oscillation. Put a 20 to 50 ohm resistor in parallel with output and you should be fine driving the ceramic transducer.

As a side note, if you apply excessive drive you can depolarize the ceramic causing the piezo properties to degrade or disappear. This will start to happen a about 200 volts per mm thickness of ceramic. (for most ceramic mixes).

 

[plukens]

Thanks for the response.
I am using the piezo disk like a diaphram to pump a liquid out of an orifice. I am not trying the achieve resonance or maximize the Q (amount of energy stored divided by the energy lost per cycle) which you would want to do if making a transducer to produce sound or couple energy into a liquid (ultrasonic cleaner for instance). I'm not concerned with efficiency. I'm just using the deflection of this device to squeeze liquid out of a nozzle. The literature I'm following finds success with a 30 V p-p square wave, thus my need to amplify my signal generator's 10V p-p. A square wave is used because the rapid fall of the tailing edge of the pulse helps pinch off the liquid and eject a droplet. I hope that helps define the scope of this project. It's more mechanical than electrical. And it's a prototype device, not something intended for production.
Your input (bad pun...) is greatly appreciated.

 

[plukens]

Piezo ceramic is made by static charge polarizing polycrystaline ceramic. It is the electrostatic equivalent of charging a magnet. Usual chargng field is about 4 kV per mm thickness of ceramic.

When electrodes are place across the polarized ceramic is will compress or stretch depending on applied voltage relative to polarized direction. As with most mechnical devices there will be a resonance point that will create a conservation of energy resulting in a build up in the physical excursion.

If the resonator is placed in a highly dampened environment (acoustical coupling) the the resonator will de-Q. You get less excursion but wider bandwidth. This is what most medical ultrasound imaging machines do. They operated at a loaded Q of 1 to 2 versus unloaded resonator Q of about 30.

You can make a low frequency ceramic piezo transducer. It is just a matter of size and mix of ceramic. Anyway, in your application creating water droplets it is likely pretty well dampened. Just about all ink jet printers use piezo transducer to 'spit' out droplets of ink.

As to audio amps, most will be okay with a high impedance load. Some without a minimum load may have high frequency oscillation. Put a 20 to 50 ohm resistor in parallel with output and you should be fine driving the ceramic transducer.

As a side note, if you apply excessive drive you can depolarize the ceramic causing the piezo properties to degrade or disappear. This will start to happen a about 200 volts per mm thickness of ceramic. (for most ceramic mixes).

RC (aka Tin Man), Understood all you write and appreciate the input. Now I have an approach to try and can calculate a voltage limit (based on thickness of my cheap piezo disks). If I had more budget for this project, I'd just have purchased a high voltage pulse generator (HP Model 214B for example), or a beautiful piezo high voltage driver **broken link removed**. But I'm stuck with a 10V p-p signal generator.

If I do use an audio amp (with a resistive load in parallel with my disk) what kind of voltages can I expect? Is it determined by the power rating of the amp? (P = V^2/R, so V = sqrt(P*R)? Where R is the load resister I put in parallel with my disk? What amplitude of signal do you put into an audio amp? I am out of my depth here...

 

[plukens]

Right now I can see the miniscus of the liquid (recall that I’m building a droplet generator to produce droplets on demand of a liquid) moving at the tip of the nozzle, but droplets are not being ejected at 20 V p-p.

Someone suggested using a low voltage square wave from my signal generator to turn a MOSFET switch on and off, allowing a higher voltage supply to drive my disk. Since I’m only seeking 100 Hz or less (to produce 100 Hz train of droplets), perhaps this is a good way to go. Will this unit do the trick?

**broken link removed**

So then I need a 50V DC supply to be switched?
Any suggestions? Ideally, I’d like a variable Voltage supply. Do inexpensive power supplies go as high as 50 volts?

Based on the input of RCinFLA, I should not exceed 200 V/mm of piezo disk thickness. That means 55 volts for my disks. If I kill a few in the process of figuring this out, no harm.

 

[alec_t]

A cheap, simple, adjustable 50V+ power supply could be made using a dual secondary 20V or 24V transformer, and a switched MOSFET/BJT could then drive the piezo as mentioned above. I'm guessing a ~3VA transformer would suffice. The whole project could probably be built with <$20 of components (excluding carriage and fancy enclosure).
The item in the link is only a switching module; you would also need an adjustable power supply.

 

[MrAl]

Right now I can see the miniscus of the liquid (recall that I’m building a droplet generator to produce droplets on demand of a liquid) moving at the tip of the nozzle, but droplets are not being ejected at 20 V p-p.

Someone suggested using a low voltage square wave from my signal generator to turn a MOSFET switch on and off, allowing a higher voltage supply to drive my disk. Since I’m only seeking 100 Hz or less (to produce 100 Hz train of droplets), perhaps this is a good way to go. Will this unit do the trick?

**broken link removed**

So then I need a 50V DC supply to be switched?
Any suggestions? Ideally, I’d like a variable Voltage supply. Do inexpensive power supplies go as high as 50 volts?

Based on the input of RCinFLA, I should not exceed 200 V/mm of piezo disk thickness. That means 55 volts for my disks. If I kill a few in the process of figuring this out, no harm.

Hi there Pluk,

The piezo element will contain some capacitance that has to be driven. This usually requires a lower than usual impedance. Depending on your capacitance a higher than optimal impedance would cause distortion of the square wave. You could look up the effect of a low pass filter on a square wave to find out what kind of distortion to expect. Since im not familiar with your application i cant say how much degradation from a perfect square wave your app can stand, i guess you'll have to determine that. If it seems like it might be a problem, then you would have to go with a symmetrical driver rather than a single transistor driver. If you arent sure, i guess you can try it and see if you get the results you are expecting.

A regular transistor with a collector resistor constitutes a square wave 'amplifier', where the smaller input can drive a much larger output. The collector resistor has to be small enough to fully drive the capacitance. The trade off is the smaller the resistor the more waste power, but the larger the resistor the less current drive is available for that capacitance.
A symmetrical driver would have a transistor on top and on bottom (known as a half bridge). There is a delay introduced between turning one transistor off and the other on so that they both cant turn on at the same time even for a tiny fraction of the cycle. This kind of drive provides very low output impedance for the load with little waste.

Do you really need 100 volts?

You stated that you are using this to drive a liquid out of a nozzle. I was thinking of doing something like this too for an entirely different reason. What i wanted to ask you was are you using any kind of check valve or do you find you dont need one?
If you are using a check valve, how are you constructing it?

 

[alec_t]

I agree with MrAl that a half bridge drive would be preferable to give sharp square-wave edges which, I guess, would be more effective for droplet ejection.

 

[plukens]

Hi there Pluk,

The piezo element will contain some capacitance that has to be driven. This usually requires a lower than usual impedance. Depending on your capacitance a higher than optimal impedance would cause distortion of the square wave. You could look up the effect of a low pass filter on a square wave to find out what kind of distortion to expect. Since im not familiar with your application i cant say how much degradation from a perfect square wave your app can stand, i guess you'll have to determine that. If it seems like it might be a problem, then you would have to go with a symmetrical driver rather than a single transistor driver. If you arent sure, i guess you can try it and see if you get the results you are expecting.

A regular transistor with a collector resistor constitutes a square wave 'amplifier', where the smaller input can drive a much larger output. The collector resistor has to be small enough to fully drive the capacitance. The trade off is the smaller the resistor the more waste power, but the larger the resistor the less current drive is available for that capacitance.
A symmetrical driver would have a transistor on top and on bottom (known as a half bridge). There is a delay introduced between turning one transistor off and the other on so that they both cant turn on at the same time even for a tiny fraction of the cycle. This kind of drive provides very low output impedance for the load with little waste.

Do you really need 100 volts?

You stated that you are using this to drive a liquid out of a nozzle. I was thinking of doing something like this too for an entirely different reason. What i wanted to ask you was are you using any kind of check valve or do you find you dont need one?
If you are using a check valve, how are you constructing it?

Thanks for the technical input. I do not really need 100V. With guidance, I found that I should not drive my piezo disk at more than 200 V per mm of thickness. This comes out to 54V for my disk. So that is my amplification goal.

Regarding check valve. This droplet generator makes small droplets. Hundreds of microns (200um) by using a very small nozzle (100um ID). The diameter of the droplet is roughly twice the diameter of the nozzle for water. There is no mechanical check valve. Just an open nozzle at the bottom of a cavity with the piezo disk sealed to the top of the cavity. The piezo disk acts like a diaphram. The paper I'm following is linked below. See figure two, which pretty much tells the whole story.

https://www.nist.gov/el/fire_research/upload/Yang-A-Simple-Piezoelectric-Droplet-Generator.pdf

 

[alec_t]

I think this should do what you want.
The components to the left of C1 make up a DC supply whose voltage can be set from ~7V to 50V. Components to the right of C1 form a full H-bridge for square-wave driving the piezo in response to a 0-5V square-wave input from a function generator. The advantage of using an H-bridge is that piezo displacement amplitude is twice what it would be for a single-ended drive arrangement; thus the amplitude swing can be up to 100V while the actual voltage across the piezo is up to 50V (i.e. within the 54V limit mentioned above).

 

[harold777]

What I'm not understanding is why Op has attempted to drive 100v piezo from a low-voltage source, maybe this is just experimental?

Surely the easiest technique is to use a transformer to push up the V? The Piezo won't take much current and any rounding of the waveform may benefit the fluid-handling, as moving mechanical things are semi-sine-shaped.
Ideally the transformer will an audio-quality - remember OP transformers? Turn one the other way round and you have a HV op.