Ultrasound transducer for bat detector project

Hey. I've started thinking about the possibility of designing a home-brew bat detector. I've researched the idea and I'm going to go for the most simple design, the heterodyne, to begin with. This design is essentially as follows: a transducer picks up the signal, before it is amplified and fed into a mixer to convert it down into the audible range. The user can then 'listen' to what's going on in the ultrasound range. The user tunes the local oscillator in order to scan through the ultrasound band.

So I thought I'd start at the start by looking at ultrasound transducers. This is where I could use some help and advice. Ideally I'd like to be able to 'tune' between 15kHz and 125kHz - but I can't figure out how to achieve this without spending a silly amount of money. I want to keep the cost down. From what I've read, I could go for:

1. Piezoelectric ultrasound transducer
Cheap, but narrow bandwidth. Possibility of increasing bandwidth (at the cost of sensitivity) by placing an inductor in parallel with it:

https://bertrik.sikken.nl/bat/detuning.htm


2. Electret condenser microphone - normally designed and specified for audio frequencies, with no specs given above 20kHz. However, apparently, these can sometimes offer some performance into the ultrasound. I have read that the following mic cartridge has some performance up to 60kHz:

https://www.electro-tech-online.com/custompdfs/2012/03/em06_wm61_a_b_dne.pdf


3. I've also looked at MEMS microphones, but again there's not much info given on their performance at 20kHz+ frequencies. For example:

https://www.electro-tech-online.com/custompdfs/2012/03/SPM0406HE3H-SBPDF.pdf

4. Electrostatic microphones apparently have good performance right up into the frequencies I'm interested in, but they require very high voltages to operate and I'd rather not go there.

https://bertrik.sikken.nl/bat/mics.htm


Since there's little information about higher frequency performance of electret condenser microphones I'm not sure if I could achieve my desired bandwidth with them. I could also consider using multiple piezo ultrasound transducers, but again not too sure how difficult this would be.

So if anyone could shed some light on this subject I'd appreciate it!

Cheers

My city used to have plenty of bats and plenty of mosquitoes. I enjoyed watching the bats flying around at dusk. Then the city put lavacide in all the street drains and now all the bats and mosquitoes are gone.

There is a recording of one of these sound frequency converter circuits playing bat noises. They are mostly clicks.

Yeh I've read about how different species sound different due to different echolocation pulses etc.

How would you go about picking up ultrasound in the 15k to 125kHz region?

This thing will get you to 65 khz -
http://search.digikey.com/us/en/products/SPM0404UD5/423-1086-1-ND/1587388

It's cheap, small, wideband, and fairly easy to use.

A Senscomp 600 series transducer will get you out over 100khz, but it has to be biased at 200Vdc to do the sensing -
https://www.electro-tech-online.com/custompdfs/2012/03/series-600-instrument-grade-transducer.pdf

You can get them for about $20 -
http://www.acroname.com/robotics/parts/R13-INST.html

Unfortunately, nearly every other ultrasonic sensor is very narrowband.

I would never buy anything from an American? Chinese? company that cannot spell simple English words correctly on its datasheet.
Are its engineers also little kids?

duffy said:

This thing will get you to 65 khz -
http://search.digikey.com/us/en/products/SPM0404UD5/423-1086-1-ND/1587388

It's cheap, small, wideband, and fairly easy to use.

A Senscomp 600 series transducer will get you out over 100khz, but it has to be biased at 200Vdc to do the sensing -
https://www.electro-tech-online.com/custompdfs/2012/03/series-600-instrument-grade-transducer.pdf

You can get them for about $20 -
http://www.acroname.com/robotics/parts/R13-INST.html

Unfortunately, nearly every other ultrasonic sensor is very narrowband.

Click to expand...

Thanks for this. I've considered electrostatic but I don't want to deal with the high voltages.

Before reading your post I was thinking about going for one of these transducers:

http://www.mouser.com/catalog/specsheets/KT-400244.pdf
http://www.meas-spec.com/downloads/40kHz_Transmit_Recvr.pdf

But after comparing the sensitivities of the devices, the Knowles MEMS sensor you linked is by far the best choice:

Kobitone: 10mV/Pa at 40kHz, dropping to around 2mV at +/- 5 kHz.

Measurement Specialities: 2.5mV/Pa at 40kHz, dropping to around 1mV/Pa at +/- 5 kHz.

Knowles: 8mV/Pa from 1kHz - 10kHz, rising to around 30mV/Pa at around 40kHz, and then dropping to around 1mV/Pa at 100kHz.

So the Knowles device has a much greater bandwidth. I'll go with that.

Does anyone know roughly how much SPL / Pa is in an average bat call? I need to look into this.

Dunno from bats, but I would suggest a heterodyne detector instead of one of those simple crappy counter-chip circuits. I've built both and there's a HUGE difference. Ianni has a working heterodyne detector in "Electronic Gadgets for the Evil Genius". It's a few more parts than the counter-divider, but it actually gives you AMPLITUDE, which that 4024 binary divider chip doesn't.

Heterodyne detector's like this: (jingle keys) jingle-jingle, JINGLE-JINGLE
Binary divider's like this: (jingle keys) ...silence... SCREETCH-SCREETCH!

Other people posted recordings from a heterodyne detector from bats sounds and it is just some clicks.

....they obviously hadn't trained the bats to jingle keys

audioguru said:

Other people posted recordings from a heterodyne detector from bats sounds and it is just some clicks.

Click to expand...

All you hear is clicks? That's too bad, you should get your hearing checked. I can clearly tell these sounds are more complex than just "clicks".

I hear many "clicks", one buzz and a few very short duration "chirps". Simply noises.
What do you hear? Sentences?

Maybe the clicks are caused by the opamp clipping.

That's the reason I went for a heterodyne - better sound fidelity.

I could use some advice on choosing an appropriate op-amp for the next stage. I assume an op-amp is the right way to go about amplifying the sensor signal?

I'm intending on running this circuit using batteries. Could be AA, AAA, whatever.

So I'm looking for high input impedance, probably a JFET-input, low noise and flat gain across ~10kHz to ~125kHz.

Trouble is, the gain of most low-noise op amps seems to drop after around 10kHz. Is this normal? I'm looking at DIP packages since I'd like to construct this on a breadboard:

https://uk.farnell.com/jsp/search/b...skipParametricAttributeId=&prevNValues=203333

Also, are these op-amps battery-friendly? By that I mean is it easy enough to generate the +/- supply rails from a battery power supply?

Don't use a lousy old 741, LM324 or LM358 opamp that have gain that drops at a low frequency and have a very low slew rate that cuts high frequencies.

A TL07x opamp has gain that drops above 100kHz when the gain is 33 times.
An OPA134 opamp has gain that drops above 100kHz when the gain is 100 times.

I used an LM6132. Higher gain-bandwidth product than the OPA134, has rail-to-rail inputs and output, works down to 2.7V, current draw is only 360μA.

Thanks for the suggestions. Think I'll go with the LM6132. The greater bandwidth means I can have a flatter gain. The low supply voltage and current draw is great too, since I'll be running this off batteries. On top of that, I believe the noise performance is comparable to the other two also.

A question about this device though - does it generate its own virtual ground? Since there's only 3 pins per amp, and pins for V+ and V-, I assume it does.

Originally I was thinking of designing for a BW of 150k. But if I go for an octave higher at 300k this will ensure that the gain hasn't dropped by more than 1dB at 150k - giving a much flatter gain. The gain itself is only ~30 but I can always cascade the second amp to get 30 x 30 = 900.

For the feedback resistors I'll use R1 = 3.3k and R2 = 96.7k. This will result in my gain of around 30 and also minimise feedback current to <1mA.

Please feel free to comment on this - not the most experienced with op amps.

You make a virtual ground with a voltage divider. looks like this -


Use two 10K resistors. Vout is your "virtual ground". PROTIP: put a .1μf cap across R2. This will lower the impedance to AC signals even more.

If you are using an existing schematic for a heterodyne ultrasonic detector, it will probably have a circuit like that in it already. If it doesn't, post the schematic, because you may need the active version of this thing, which is called a "railsplitter" and looks like this -

I'm not using an existing schematic for this - I wanted to do it myself.

I've attached my schematic - I wasn't sure what value to use for the input and output AC coupling caps? All I need them to do is block DC - and be of minimal impedance to AC. So is 100uF ok?

Actually - the cap you suggested putting across R2 of the divider is there for the same reason - to short any noise on the virtual ground line, yes?

breakshift said:

So is 100uF ok?

Click to expand...

Yes, though it's overkill. The formula for capacitive reactance is

[LATEX]X_c &= \frac{1}{2 \pi f c}[/LATEX]

So at 20khz the 100μf caps have .079Ω, and it just gets lower as the frequency goes up. You could put a 1μf in there and it would still be bigger than necessary.

breakshift said:

to short any noise on the virtual ground line, yes?

Click to expand...

Yes. At 20khz the .1μf has a 79Ω impedance, three decimal places lower than the two 10k resistors... though still a lot higher than the 100μf caps.

Your circuit biased the wrong input of the opamp. The (+) input must be biased at half the supply voltage but yours had no bias voltage so your amplifier will not work.

Your 100uF input capacitor would take maybe 1 or two days to charge because its value was a wrong guess at about 10,000 times too high.

Do it like this:
This website does not allow me to upload attachments anymore. It times out after I wait about 1 minute.

EDIT: I tried to upload the schematic as an attachment again. But after waiting and waiting it times out every time I try it.

Today I will try to attach a schematic.
This thread took 15 seconds to load.
"Go Advanced" took about 20 seconds to load.

But "Manage Attachments" timed out after about 1 minute.