Mosfet driver advice for ultrasonic transducer

[waxtend]

Hey! I am designing a variable frequency ultrasound transducer driver with a PWM chip and mosfet gate driver. The pwm part works charming, but the mosfet gate driver has problems. I need some verification, if it could work this way.  I have attached the schematics of the driver. (credits for schematics http://tahmidmc.blogspot.hu )  Since the transducer is a capacitive load, am i thinking right that i should use the half bridge, and not only the high side driver? Is there any additional part i should consider when driving capacitive load? The driver takes care of the Miller effect as i know.
I would appreciate some help!

[uncle_bob]

Hi

What is the capacitance of your transducer? Do you have equivalent circuit parameters for it in the loaded state? Is the transducer a two terminal device or is it more complex? If it's two terminal are you running two of the high side drivers?

Bob

[waxtend]

Hi, it is a simple two terminal device which can be drived by sine or square wave. It is around 8nF. If we want to be extra precise, its equivalent is an RLC series, but mostly RC. Now i built only the high side driver, but i am concerned. The capacitor should be drained by connecting it to ground with the half gate?

Daniel

[uncle_bob]

Hi, it is a simple two terminal device which can be drived by sine or square wave. It is around 8nF. If we want to be extra precise, its equivalent is an RLC series, but mostly RC. Now i built only the high side driver, but i am concerned. The capacitor should be drained by connecting it to ground with the half gate?

Daniel

Hi

If you connect one side of the transducer to B+ and only have a low side driver on the other side ... nothing useful happens. If you connect both sides of the transducer to low side drivers ... nothing happens. If you hook one side to ground and the other side to a high side driver nothing useful happens.

To get useful work out of the gizmo, you need to change the voltage on it each cycle. It does not conduct DC so you need to have both a high and a low side driver on it. You can put drivers on both sides of the transducer, that will double the voltage on it and eliminate the DC bias. Your transducer may or may not care about DC bias.

If the Q of loaded transducer is < 1 in operation, you can treat it as a R-C circuit. If the system goes through resonance, you may have to be a bit careful about what the electrical load looks like.

Bob

[waxtend]

Hi Bob, thanks for the answer. So you are suggesting to hook up a half bridge ( high and low side driver). By this, the supply voltage varies between 0 and vcc with a specified xkhz speed, and the capacitor charges, and when the bridge ties to ground, it drains the cap. Is it what you are saying?
Thanks
D

[Kleinstein]

Yes the half brigde is what uncle_Bob suggested.

A mechanical resonance can make the piezo look like a resonant circuit, even with a rather high Q.

[uncle_bob]

Hi Bob, thanks for the answer. So you are suggesting to hook up a half bridge ( high and low side driver). By this, the supply voltage varies between 0 and vcc with a specified xkhz speed, and the capacitor charges, and when the bridge ties to ground, it drains the cap. Is it what you are saying?
Thanks
D

Hi

Yes, also consider that your nanofarad capacitance is charging and discharging at a rapid rate. That (at the 300V you show) can be quite
a bit of current.

You get 8 nano coulombs of charge for each volt you go
300 V gets you to about 2.4 uC
At 200 KHz, you are running a half amp.

If you need to double up (drive both sides):

It's now 600V so twice the charge
You run an amp at 600V

You either have 150 W or 600W flying around your system.

If the transducer plus mechanicals goes into resonance with a Q=10 (which is very easy) and you hit that as a series resonance, your circulating current (through the MOSFETS) is either 5A or 10A. If you hit it as a parallel resonance, you have 3KV on the transducer terminals ...Thus my earlier question about models and equivalent circuits. If you only running a highly viscous load and cavitation can not occur, maybe resonance is a non-issue. A lot depends on what is on the other side of the 150W / 600W transducer. It also depends
quite a bit on the drive frequency for average current.

Consider that the peak current in the FET's will be limited by their resistance. That will apply regardless of frequency and the parts need to be sized to handle that regardless. Say you have a one ohm FET (random number):

Your 300V transducer will put 300A into the FET when you do the switch. The part needs to be OK with that current. Very briefly it will see P = I^2 * R = 300 * 300 * 1 = 90KW of power. That's not as alarming as it might sound, but the part does need to handle it. Is one ohm good enough? Your 8 nF transducer with 1 ohm will have an 8 ns time constant. You want something out >> 20 time constants to have any sort of steady state rules apply. That gets you to 0.16 us.  One ohm is probably ok, ten ohms might be ok....

Bob

Bob