Driving UltraSonic Transducers from 12V DC

Hi There,

I have 4x MA40S4R/S Ultrasonic Transducers, and a 12V Lead-Acid Battery. I need something to drive the ultrasonic transducers to about 40Khz at 20 Vpp TX. I have come up with using an astable 555-timer to create the 40Khz square-wave and I have tested it on the breadboard and on multisim and it works fine when I use one of the four transducers I have. The problem is I need to amplify the output so its output voltage is around 20Vpp so that way it works better. I have been messing with it in multisim but wondered if this was the best way to do it, or if you could help me by providing me recommendations to my circuit or if there is an even simpler approach to driving 4x transducers at 20Vpp 40Khz.


Datasheets:
https://www.mouser.com/catalog/specsheets/Murata MA40 Series.pdf
https://www.ti.com/lit/ds/symlink/lm555.pdf

see attached pic of my multisim circuit, any help is greatly appreciated.
https://obrazki.elektroda.pl/2469135200_1434976220.png (also why does my output looks so funny, it should be more of a square-wave).

Some options to increase the transducer voltage:
- use a H-bridge configuration
- use an output transistor with inductor load

The biggest problem driving these transducers is that the resonant frequency of the transducer is a very sharp peak.
The driving frequency must coincide exactly with the resonant peak, or very little energy will be transmitted.
Its not just a case of feeding more power into it, but feeding in the exact frequency which is fine when you are tweaking it on the bench, but building something that will stay exactly in tune for any reasonable amount of time can be very difficult.

Best approach is to use the transducer itself as part of a feedback oscillator so it automatically oscillates at the energy peak.

There are actually two peaks very close together, one is a voltage peak, the other a current peak.
The current peak might be better when trying to do this from a low supply voltage.

You might like to look at some of the modular bridged audio power amplifier chips for doing this.
If you use current feedback from the transducer, you should be able to get it oscillating AND absorbing significant power at a fairly low drive voltage.

Murata 40 kHz air ultra sonic transducers (also similar products from other manufacturers) have moderate bandwidth of about 1 kHz, so meeting the resonance frequency is not so critical. Most processor based distance measurement systems are operating it just at the nominal frequency.

For a simple drive circuit, you can refer to figure 14 in the application manual. I couldn't find it presently at murata.com, so I uploaded a backup copy.

A 555 has plenty of output current to drive a ultrasonic transducer [b[without[/b]15A power transistor. Use two 555 ICs, one as an oscillator and the other as an inverter. This circuit can produce about 20Vp-p:

The Murata device has 2550 pF capacitance which at 40KHz has a reactive impedance of 50 Ohms and 20Vpp max driver rating. THe resistance or real energy depends on the motional vibration levels but is likely to be in the same range at maximum voltage of 20Vpp for low Q pulse sonar.

This is easily achieved using 10V full bridge to get 20V differential. Matching the driver impedance may be important for transient response when turning off if this is intended for SONAR testing.

THe figure illustrated in MUrata's datasheet offered by FVM shows a pair of 4049B CMOS drivers AC coupled in differential mode using Vcc=12V The impedance of 4049B's as I recall is around 200 Ohms at 15V thus in a 2S2P matrix ( 2 Series+2Parallel) the net impedance is still 200 Ohms. Thus significant drop is expected from each driver. THis impedance lowers at rising Vcc and varies considerable among vendors or 4069B.

I would consider a dual half bridge at 10V with RdsON < 1 Ohm

I would use the fastest CMOS inverter to produce the complementary input and send the burst to the device and then momentarily for turn on both drivers low or High to shunt the device before floating them to reduce the resonant dwell time after end of burst to get a crisper burst (20us) after Tx before Rx is processed. This could be adjusted to produce the best transient response.

THese are recommended for Vcc=11 so if you used 12V add an RC current shunt to drop 1V at expected current, which may be in the range of 20V/50Ohms= 400mA Peak, so 2 or 3 series 1R can be tried so as not to exceed the 20Vpp max rating for current and power dissipation reasons in the device.

You can use induction at the end of your transistor.

2550pf is 2.55nF. I calculate the impedance at 40kHz as 1558 ohms, not 50 ohms. Two 555 ICs as a bridge can easily drive it.

Murata's applications sheet shows two CD4049 inverters in parallel as half a bridge. Then its minimum output high current is 15mA and has a 2V drop with a load that is 667 ohms. The 555 has a typical output current of 200mA.

My pocket calculator also tells Xc = 1560 ohm.

But more important, the input impedance at the resonance frequency isn't given by the sensor capacitance. It's a real value at the parallel and series resonance. A transmitter would be operated at the series resonance which has typically a few 100 ohms real impedance.

FvM said:

My pocket calculator also tells Xc = 1560 ohm.

But more important, the input impedance at the resonance frequency isn't given by the sensor capacitance. It's a real value at the parallel and series resonance. A transmitter would be operated at the series resonance which has typically a few 100 ohms real impedance.

Click to expand...

Then why does Murata recommend driving it with a weak little ordinary CD4049 Cmos inverter/buffer IC? They show two inverters paralleled for each half-bridge and a 12V supply. The supply current is only 15mA and the output is 17Vp-p on one schematic and is 20Vp-p on another schematic. Then the piezo transducer load is 1133 or 1333 ohms.

Driving ultrasonic resonators effectively is far from easy.

At parallel resonance the impedance will peak and the drive power will dip down to a minimum. It may look really good like that on an oscilloscope, but acoustic power output will not be anywhere near the maximum possible.

At series resonance the driver will be quite heavily loaded, and that is actually what you want to see happen. But driving it at that exact precise series resonant frequency is very difficult to do consistently, unless you have some mechanism to lock the frequency of the driver to the exact series resonant current peak through the transducer.

Many years ago I was attempting to continuously drive ultrasonic transducers in the 3Mhz to 5Mhz range with several tens of watts. The only reliable way to do it was to use the ultrasonic transducer itself as the frequency determining element, by taking feedback from a current transformer placed in series with the transducer.

You then need a high gain limiter amplifier (which saturates) to drive the power output stage. System noise starts the action, and amplitude quickly builds up at the transducer series resonant frequency. Its very long term reliable.

You get max feedback at the minimum impedance point, and it will oscillate merrily and draw a lot of current. The oscillation frequency also tracks any load changes which will occur if you try to tightly couple the transducer into fluid or solid objects.

This only works for continuous output. If you are planning to pulse it, the problems are going to be much greater.

There is far more to it than just generating 20 volts of ac across the transducer terminals.

We are talking about Murata's 40kHz ultrasonic transducer that has its datasheet posted here, not a completely different radio frequency thingy.
The graph of sound pressure vs frequency shows a peak at 40kHz but no "high impedance null".
The frequency fed to it does not need to be very accurate since its output level is almost the same (-3dB) from 33kHz to 50kHz.

That is not what that curve is telling me !!!!

I see a change of more like 40db between 30 Khz and 50 Khz.
The output varies from something like 80db to 120db.

Its not 3db down, its more like a 40db power variation over that frequency range.

OOPs, the graph shows dB, not actual levels. The peak is narrow. That is why Murata shows a frequency adjust trimpot on their schematics.
Maybe you can adjust the trimpot for maximum power supply current (only 15mA).

FvM said:

Murata 40 kHz air ultra sonic transducers (also similar products from other manufacturers) have moderate bandwidth of about 1 kHz, so meeting the resonance frequency is not so critical. Most processor based distance measurement systems are operating it just at the nominal frequency.

For a simple drive circuit, you can refer to figure 14 in the application manual. I couldn't find it presently at murata.com, so I uploaded a backup copy.

Click to expand...

I agree with FvM and apologize for my previous error of 50 ohm mis-calc in my head.

All crystals have the same equivalent circuit with series then parallel resonance. The difference is, resonators which operate in high impedance mode use the parallel capacitance which has very little impact in this case as it is used in series resonant mode. These are known to operate << 100uW and used with CMOS negative feedback oscillators as the phase shift is 180 degrees.

THe power generated by the lattice structure is the small motional capacitance in series and the voltage across the series resistance is the indicator of current and work done by the crystal. with the series resonance having a phase shift of 0 deg, which is irrelevant in this case as an open loop stable frequency can easily drive these near maximum power using an accurate reference such as a binary scaled down Crystal Osc or from MCU PLL.

I found the values for this part in a Microchip App Note AN1536 and simulated the transfer functions with changes in impedance in/out and then change output from voltage to current.

1. high source and load impedance across the crystal to measure voltage with the series dip 41.5kHz and parallel peak near 43.3kHz
2. lower R load value eliminates parallel high Q resonance peak
3. lower source R flattens voltage response
4. changed output to current response across Xtal series 340R shows true ultrasonic curve peak for series resonance.



However the Murata App Note that FVM linked shows two devices, one for Tx and one for Rx and driving with a low impedance source will not work in half duplex mode sharing one crystal.

So If you go this route 2 parallel 4069B buffers in single or bridge mode will do the job.

Otherwise if using only 1 crystal, you need to use a tristate MOSFET bridge as I indicated earlier with burst, short burst then open circuit to suppress the ringing of the series mode resonance.

You can compute the series Q from the table of values for Series RLC where Q=2pi*41kHz*48mH/340R and then from the Bandwidth determine the absolute minimum ringing time with a shorted burst. It may lead to think you will need two devices for short range doppler pulsing.

- - - Updated - - -

I hope my screen shots can be zoomed clearly and my quick explanation is adequate.

Any questions? Current drive should not exceed 20Vpp rating/240 Ohms or around 10 mAdc to the drivers accordingly or maybe 15mA as audioguru indicated by sensing DC current and tuning a VCO or relaxation oscillator to CMOS drivers ~ 100~200 Ohms. Careful review of specs will determine this,

The problem is that both the source frequency and the transducer resonance move around.
Adjust it perfect on Monday, and by Friday it will need readjustment.
I have no idea how frequency critical those 40Khz transducers really are, but I would guess even +/-100 Hz would give a significant fall off in acoustic output either side of the peak. Only a few db, but a few db is a lot.

The resonance peak is a very sharp spike, not a smoothly rounded off pyramid as that diagram tends to suggest.

Its all possible....
But you will be tweaking that thing for the rest of your life unless you can make it self tune somehow, which is definitely the way to go if you are planning for a continuous acoustic output.

Most uses of these things require pulsing, which creates a very big problem.
However, in this day and age, it may be possible to use software to tune it and make it slowly self correct over time. Never tried that, or seen it done before, but it may be something to think about for a serious application.

Please do not mention the CD4069 that has much less output current than the CD4049 that Murata recommended.

How about one of those dual mosfet gate driver chips ?
They would drive 2nF or more, rail to rail easily, without raising a sweat.

If the tolerance on the resonator is 100 Hz or 2500 ppm then your clock source should be 100pm like any cheap crystal or ceramic resonator or watch crystal.

Make a CMOS PLL with dividers for a fractional N to a common denominator for mixer like 1KHz and have VCO drive the bridge using any crystal on a CD 4040 binary counter
Can you synthesize the required stable frequency?

of course temperature drift is another factor for the resonant frequency with power 20V*10mW or 200mW self heating.

OK?

- - - Updated - - -

Audioguru said:

Please do not mention the CD4069 that has much less output current than the CD4049 that Murata recommended.

Click to expand...

Good point Is it rated for 20Vpp/340R ?? 59mApp .

On second thought, the Vcc current should be sinusoidal since harmonic voltage will not draw much current plus the spike for Ic=Cdv/dt for duration RC.


In any case , I think Murata suggested dual sensor and ganged CD4049 driver with separate receiver is going to be best start for now.

Except using a stable 100ppm clock source that be tuned over +-2500 ppm range around centre.

Microchip has good info for the receiver
ww1.microchip.com/downloads/en/AppNotes/00001536A.pdf

- - - Updated - - -

Warpspeed said:

How about one of those dual mosfet gate driver chips ?
They would drive 2nF or more, rail to rail easily, without raising a sweat.

Click to expand...

The parallel Cap only detunes the centre frequency in series resonance very slightly as the series cap does the work.

As FvM indicated the vibrational load resistance will be in the range of a couple hundred Ohms as I graphed in final simulation.

In both cases Driver impedance ~ 150 Ohms or two 15V HCMOS buffers is no problem. @15V Vol=1V @35mA or 286 OHms ( usually 200~300 Ohms each)

- - - Updated - - -

Warpspeed said:

There is far more to it than just generating 20 volts of ac across the transducer terminals.

Click to expand...

Right
you need a series resonant oscillator ( Non inverting) with DC self biasing (negative feedback) using two inverting stages.
Low power methods usually use two TTL gates self biased with 200 Ohms ac coupled to connect crystal after 2 stages of gain.

Perhaps complementary bridge from a higher voltage source self biased with low input impedance in series with low output impedance for low loss, non inverting AC feedback but self biased for DC square wave symmetry.

If this thread is still dedicated to 40 kHz air ultrasonic transducers, it would be good to refer to actual transducer data. As already mentioned in post #4, they have some bandwidth. If you have a stable oscillator driving the transmitter, there's no tuning required. In case of doubt, look at the various distance sensors offered everywhere.

For designs with separate TX and RX that implement low impedance TX and high impedance RX matching, Murata and other vendors used to offer transducers with slightly different resonance frequency. So the transmitter is operated in series and the receiver in parallel resonance (in terms of the equivalent transducer RLC circuit).