Ultrasonic >100w amp for vibration exprmt...

[krazatchu]

Woot! my first post...

So here is the situation...

I'm doing some experimenting with ultrasonic vibration, using a 100w welding transducer and later, when my cnc mill arrives, some custom made transducers (designed with FEA)...

I'd like to keep the costs reletively low ... so I've opted for doing the electronics myself... I have a bit of experience with electronics but not in the area of power amplifiers..

I came across a schematic in a pdf file about ultrasonics... It seems a bit old since some of the components are obsolete... I modeled the schematic, had to swap a few parts and simulated it... It runs alright when fed a 20kHz signal, but responds poorly outside of this range...

I need to run it with a sine wave as I hope to achieve standing waves of a particular frequency, square wave would give me too many harmonics ...

I will build a sine wave generator later for testing and finally a PIC to generate the signal.... input should be from 0 to 1 volt.



So... to summarize things... these are my basic requirements

Power: >150watts (more is better)
Frequency responce: 18kHz to 30kHz
Amp type: push/pull Type B or AB

And about the circuit...

D2 and D4 are supposed to be BZV46/1.5V Reglator reference diodes
I changed them to 2.2V zeners (and turned them around) since I could find that part... I couldn't get it to simulate with a similar diode

R1 and R9 are adjusted to better match the 2 zeners, they were 680 ohm and 91 ohm originally...

Q12 was a BUZ34, now its a IRF240
Q10 was a RFM10P15, now its a IRF9240

Output step up transformer is not shown, it would be about a 1:3


Please comment on what I've done wrong... I need some serious advice here...

Thanks a bunch!

Michael Grant

 

[krazatchu]

Ok, so in my quest for simplicity, I changed my search terms and finally found another circuit...

The link is https://www.edn.com/contents/images/120704di.pdf ...

Second article of the pdf...

This one is much less complicated ...

I modeled the circuit and simulated it with a 20kHz sine input, 2V ...

The ouput looks clean, however, I'm wondering why the since the op amp out is a bit dirty (see scope jpg) ... any ideas? I swapped the LF411 with a TL082 and it was similiar....

Also, the datasheets on the MJE340 and MJE350 don't state an operating frequency range...

So... does this suit my purpose?

Thanks,
Michael Grant

PS, I didn't include any transducer info as I will visit the company/factory sometime next week and talk to one of there engineers...

 

[Nigel Goodwin]

The second one isn't linear, it's rather crude, and relies on the fact that the transducer probably isn't very fussy about the quality of it's signal.

 

[krazatchu]

Yes, I'm not too concerned with the distortion, jsut trying to build a general purpose untrasonic amp...

Althou..
Upon simulation, circuit 2 shows abour 0.2% harmonic distortion with a 20kHz input @ 2Vp sine wave... It gets worse as the frequency goes up... also it wont take a square wave without getting crazy...

And circuit 1, under the same input conditions... gives a 4.9% distortion...

I believe this is due to my poor choice of component swapping...
just a hunch, but i think its those diodes D2 and D4 ... there supposed to be reference regulator diodes... I replaced them with zeners and turned them around... any suggestions about fixing that properly??

thanks
Michael

 

[audioguru]

D2 and D4 are supposed to be about 2.2V (low voltage zeners have poor accuracy) but your reversed diodes are only about 0.65V.

How can the circuit work properly when they are wrong?

 

[krazatchu]

Oh,

I thought the zener voltage was the reverse breakdown voltage...

Originally, d2 and d4 were supposed to be BZV46 1.5V...
Which is an obsolete "general purpose reference regulator diode"....

The 2.2v zeners I used as replacement were the closest thing I could figure...

Could someone suggest a cross reference for the BZV46 1.5V ???

Thanks,
Michael Grant

 

[audioguru]

I thought the zener voltage was the reverse breakdown voltage...

It is. You said you reversed the diodes (making them forward biased at about 0.7V?)

Originally, d2 and d4 were supposed to be BZV46 1.5V...
Which is an obsolete "general purpose reference regulator diode"....

The 2.2v zeners I used as replacement were the closest thing I could figure...

Could someone suggest a cross reference for the BZV46 1.5V ???

Zener diodes with a voltage less than about 3.3V are lousy regulators, especially very low voltage ones.

Their voltage depends a lot on their current like an ordinary resistor and their voltage changes with the temperature.
What current are the 2.2V zener diodes rated at? What is their part number?

You increased the voltage from 1.45V (rating for the BZV45 1.5V) to 2.2V which is an increase of 1.52 times. But you decreased the value of R1 4.5 times which was wrong. With a 2.2V zener diode, R1 should be 1032 ohms (1k).

You decreased R to 50 ohms when it should be 138 ohms (150 ohms).

I would use the original resistor values with 2 ordinary 1N4148 diodes in series. Forward-biased they make a total of 1.4V.

 

[krazatchu]

Originally, the two diodes were "pointing" down... that makes them foward biased, correct?

The 2.2v zeners are toshiba, the spice info is

model 02BZ2.2 ZENER (v_breakdown=2.2 i_breakdown=10e-3 r_breakdown=15 i_rev=5e-9
+ i_sat=1e-12 n_forward=1.7874)


Ok, so I just tried 2 1n4148 in series, pointing down... and changed the 2 resistors back to 680 and 91 ohm ...
on the output scope... I got couple of saw shaped oscillations and then flat at about 52 v... I tried 3 in series... that works... 4 also works but the result is the same

I tried a few different configs...

Allrighty... instead of trying to type this I'll just include a the scope and circuit pics as attachments...

What do you think is going on with this circuit?

Thanks!!

 

[audioguru]

You have the ordinary diodes and the zener diodes with the correct polarity.
The 4.4V sets of zener diodes work the best.

You could use 4.3V zener diodes like the 1N5991B, BZX55C4V3 or BZX79C4V3 which are all rated at 5mA.

 

[krazatchu]

Hey,

Thanks for that tip about the 4.3v zeners...

The BZX55C4V3 worked the best, my library doesn't have the BZX79C4V3...

I used the distortion analyzer and adjusted the two resistors until I achievd the lowest possible distortion... at 20kHz and at 30kHz.... 0.417% and 0.701% THD... see jpg for R values

My question is... Is this a good idea? Can I trust the simulation?
Of course tolerances come into play in the real world... but otherwise... how much faith should I put in the accuracy of the simulation???

Thanks so much,
Michael Grant

 

[Goldsphere]

Hi there, my first post here. I'm looking to build a similar device but am going to use a mosfet half bridge with a mosfet driver.
Have you thought of doing the same? It would simplify things a bit.

 

[krazatchu]

Woot!

I would love to simply things...

However... I have no idea about a half bridge driver ...

Ah.. half = h ... so thats an H bridge, that I know is commonly used for motor control...

Otherwise I don't know much else...

I'd be happy to give it a try, do you have a schematic?
And what are you driving? A welding tip, cavitator, spray nozzle, etc, etc??

Michael

 

[Goldsphere]

I see now that you are trying to drive it with a sine wave, but the half bridge will be a square wave. I don't know if that is what you want.
I haven't got a schematic as yet, still in my head. But was thinking to use two IRFP450 and a TC4422 driver.
Signal can be created from a 555 and run through a hex schmitt trigger. Although I am looking to use a PLL and divider to create harmonics.

 

[krazatchu]

Hey~

I'm trying to completely avoid harmonics, since my purpose is to create ultrasonic standing waves in an aluminum plate... hence the sine wave...

Would you be using pll to create a stable/frequency drive?

I would certainly be curious about that ... As I'm going to need a stable/accurate square wave frequency generator ... before converting it to sine...

I might have to generate using pic... not sure tho, I need an incrementally adjustable generator.... haven't research that part yet...

If I new enough.. I would try to implement the tc4422 you mentioned.. it looks like a more elegant solution for my circuit... but thats just a guess...

Thanks,
Michael

 

[audioguru]

The distortion in your amplifier is much better than before.
Most modern amplifier circuits have distortion much, much less than 0.417% at 20kHz.
Your simulation applies only to transistors that have exactly the same spec's as the ones in your simulation. Transistors have very wide spec's.

Look here for some amplifier designs with fairly low distortion:
https://www.ampslab.com/projects.htm Look at the C300 amplifier.

 

[krazatchu]

Wow~~ Loot on that link!

Thanks for the link, any idea how the amps on that page will respond past 20kHz??

I'd like to work in the 20kHz to 30kHz range....

THanks!!

 

[Goldsphere]

Well simplest would be to use a function generator and drive the TC4422 input with that, purely one frequency, no harmonics. The PLL is for my purpose of driving different peizos at harmonics. I think the square wave that the H bridge would produce wouldn't matter much for Chladni figures, it is just the frequency that the plate is vibrating at, not the clarity of signal that is important.

 

[audioguru]

The ampslab amplifiers have a lowpass filter at the input that attenuates high frequencies. Remove it of increase its cutoff frequency.

 

[krazatchu]

Great... I'll model that circuit and give it a try...

On another note.... The generator note....

I think I may try using a CD4046 to give me stable/incremental frequency control...

Here is an example schematic **broken link removed**

Then a max292 or similair for sine converison from square....

woot~~
Michael

 

[audioguru]

I made an extremely low distortion oscillator with an older National Semi switched-capacitor Butterworth lowpass filter IC. I made a 10-steps sinewave with a CD4018 and 4 resistors (10-times oversampled) and filtered it so the distortion is almost unmeasurable.

I made a similar distortion analyser's notch filter with a general purpose switched-capacitor filter IC using the same clock as the oscillator so they tracked.

I tried to make their frequency adjustable with a 74HC4046 phase-locked-loop but the clock had a little jitter that increased the measured distortion.