20kHz to 120kHz acoustic transducer (speaker)

[jstuart]

I'm trying to test some ultrasonic acoustic filters and attenuators. I have tried several piezo devices, but their high Q limits their frequency range. What I'd really like is a low power speaker (1 watt or less) with a fairly flat response to 120kHz.

I appreciate any suggestions. Thanks.

[BrianHG]

If you are lucky to find them, though the final speaker as a whole are usually rated to 20khz, the 2 dimensional flat planar ribbon tweeters on some really old Clements speakers can go up to 100khz @ 25 watts, but even if you can find replacements, both the motor and diaphragm, it will cost you more than 500$us each.  That's just the tweeter's diaphragm and it's huge magnet outside of the case.  You also still need the front acoustic Fresnel lens which is part of the speaker cabinet to direct the audio forward.  Say another 100$ or more.  Way back, I hear that these tweeters were used working with bats in experiments with their ultrasound.

[DaJMasta]

Do you need a sweep or can fairly coarse steps work?  If steps are an option, you could go with an array of ultrasonic transducers, they seem to be available up into that range, but generally don't respond well far from their resonant frequency.  Elements designed for ultrasonic cleaners may be an option too, since some of them offer a range of frequencies or a sweep cleaning pattern.

[BrianHG]

I found cheap planar tweeters which are available at the links I posted below, though they are rated flat to 40khz, their higher frequency response wont die off like normal round type tweeters with a standard voice coil.  Since you only want support for 1 watt, maybe with a good 10 watt custom amp with compensation for the higher frequencies will easily get this thing to be flat from 20k to 120k.  Looking at the frequency response, I wouldn't be supersized if it was flat to 80Khz and you might just need +3db or +6db at 120khz.  This means a 5 watt amp would do feeding the tweeter with a resistor divider with a parallel cap on the series resistor boosting the higher frequencies.  A little tuning of this filter circuit will need to be done.

Here are some US sources:
http://www.parts-express.com/fountek-neo-x-10-ribbon-tweeter-black--296-705
https://www.amazon.com/Fountek-Neo-Ribbon-Tweeter-Black/dp/B00CWEJXQE

Remember, planar tweeters have over 10 times the response speed of normal speaker tweeters.  If you are asking for a speaker full range type frequency response up to the insane 120Khz, I doubt there are any other affordable devices you could find which will be so flat and make it even a fraction the way to 120Khz.

[jstuart]

Thanks for the suggestions.

I would love to be able to sweep through the decade from say 15kHz to 150kHz, but I can probably get away with three discrete measurements at 20, 40 and 80kHz.

I did some testing with the Murata MA40S4S. Ding it with a 10uS 5V pulse and it rings with very consistent amplitude very close to 40kHz for about 250 uS, which is perfect for my use. If I can find similar devices that work at 20 and 80 that would be a good start.

I just found Durham Instruments (http://www.disensors.com/) make a low-Q air transducer with a center resonance of 80-90 kHz. I'll look into that.

Nice thing about these piezo air sonar devices, they are really simple to drive.

The Fountek tweeter is interesting too. I think I have parts to make a low power amp that could drive it to 40kHz at least. It doesn't have to be hifi for my purposes.

Thanks.

[BrianHG]

The Fountek tweeter is interesting too. I think I have parts to make a low power amp that could drive it to 40kHz at least. It doesn't have to be hifi for my purposes.

Thanks.

Any simple Class A low power amp, say 2-5watts, IE, a small power mosfet with a 2-3 ohm power resistor tied to to GND, +12vdc supply & a small heat sink with biasing resistor should give you a good 1MHz response.  A series gain resistor on the input gate, with additional series resistor and series cap will boost the high frequency flattening out response of the tweeter above 80 Khz.  Don't forget the series cap at the input and another at the output to protect the tweeter from a DC voltage.
Careful, though low power, you can pierce your ear drums since this circuit would function at a few Khz and up if you feed it so.

Remember, the Fountek delivers 92db at 1 watt...

My ears hate pure sine waves at 15KHz through 18KHz, I can no longer hear frequencies above that.

[cs.dk]

* If you try using the "hissing" noise sound from either a compressed air jet or maybe a turbulent water spray / jet that might be fairly broadband.  As I recall they're both commonly ultrasonic in reach, probably at least 10s of kHz, but I don't know about beyond that.

I've got one of those ultrasonic leak detectors. Most of them listens between 38 and 42 kHz. They can also find electrical arcing, so it must be in the same frequency range as air leaks.

[_Wim_]

Maybe a plasma tweeter could be a solution (due to the extremely low mass of the "membrane", I expect a very high frequency responce must be possible, mainly limited by the electronics only):
http://www.audiocircuit.com/DIY/Ionic-Speakers/Project:Plasma-loudspeaker-by-Ulrich-Haumann

[Marco]

Doesn't go up high enough but the Kemo L010 is interesting in that it's a wideband piezo speaker, it seems to combine a diaphragm with a piezo transducer.

Anyway, for enough money these people have something ready to ship for you.

[BrianHG]

Anyway, for enough money these people have something ready to ship for you.

LOL, their transducer, looking at the chart, almost flat to only 60KHz, then, a -12db nose dive at 100KHz response seems to have the same output bandwidth as the 79$ Fountek tweeter, maybe worse.  I bet if you removed the Fountek's internal driver transformer and drive the diaphragm directly, something like a 0.5ohm impedance, only about a foot of tracing for such a small tweeter, it would work to even better only because the original owner of the patents, Clements transformerless drive diaphragms on my speakers are rated to 100khz flat, but they are seriously expensive...

[Marco]

It's easier to rate something flat to 100 kHz than have it measured flat.

[jstuart]

Thanks for all the suggestions. I located a 92kHz piezo device (SMD22T25R211WL made by Steiner & Martins). I expect this will produce consistent SPL, like the Murata @40kHz. If I can scare up a piezo tuned around 20kHz that should allow me to detect a 1st order filter response.

So I think my first shot at this will be to test at 3 fixed frequencies, with & without attenuators & filters, look at peak voltages with a DSO, you get the idea I'm sure. If the transducers are consistent, it should work.

BTW the mic I'm using is the Knowles SPU0410LR5H-QB. Like a lot of MEMS mics, the ultrasonic response is down relative to the audio band, but the noise floor is down too, so with the right filtering it can be useful in the 20kHz-100kHz range. With this particular mic I'm thinking two 1st order highpass filters at 40kHz will eliminate most audible noise, and make the response pretty flat from 20kHz up to its limit. (I know it's a droopy design, but take a look at the datasheet http://www.knowles.com/download/file?p=SPU0410LR5H.pdf
. Droopy may work for once.)

What I'm hoping to produce is an acoustic highpass, followed by an electronic highpass, both at 40kHz. The former is to help mitigate overloading the mic with audio noise. As others have noted, these mics don't respond well when overdriven.

Can't wait to see where this goes....

[mark03]

Hey, I'm using four of those mics right now, summed together, high-pass filtered, digitized, then heterodyned inside an STM32, for an ultrasonic receiver (listening to bats).  The prototype works ok but I'd like to move to some better (higher dynamic range) mics from Invensense.  According to the MEMS people, most bottom-port analog-out microphones should have usable ultrasonic response, but as you've discovered, it's not easy to characterize.  Based on my literature survey, a short spark gap (~1 mm) should emit a sufficiently broadband impulse to assume flatness and permit the design of a compensating filter.  I haven't tried it yet, though.  Please let us know (or PM me) how your project goes.