I am not an electrical engineer, but I am up for a challenge and this is my first venture into electrical engineering. However I am beginning to think this may be a step to far!

As a hobby I survey insects, at the moment particularly those that 'sing' in the 6kHz to 20kHz range. Some of that range I could never hear and the rest I have mostly lost with age!

Last year I built a simple parabolic mic (an old car headlight reflector) with an Electret microphone mounted at the focal point, wired to a 3.5mm jack into my phone (Android). There I have an app, which visually displays the input from the microphone, allowing me to see the frequencies in which I am interested but can no longer hear. This didn't give me the detection range that I needed (it only gave up to about 5 metres and I'd like 10 times that or better).

So I started researching off-the-shelf kits to pre-amplify the signal and found some available online, but I need something more specific. Ideally I am only interested in amplifying a small band of frequencies from 6kHz to 20kHz. So then I started looking at active bandpass filters and have found some education material that shows calculations for the bandpass components, but I am getting a bit lost as to which amplifier to use.

Am I jumping in too deep for a complete novice? I hope that it is feasible to design/build a simple circuit powered by the Android 3.5mm jack, which will filter the frequency range in which I am interested, and amplify it to a level that doesn't damage the phone (but does increase my detection range)?

I'm assuming that the key input to the design would be the voltage from the phone jack, but what other attributes would I use to filter the choice of opamp? How do I ensure that the gain through the circuit isn't going to harm the phone?

Another input is that I'd like the resonant frequency of the bandpass amplifier to be close to 14kHz (where the hardest to detect insect 'sings').

Any pointers to appropriate references to help me progress or other considerations in the design that I have missed would be greatly appreciated.


Thanks for comments/suggestions.

The (hand)shotgun looks interesting but I think I will iterate on my parabolic mic first - a quick read suggests that the shotgun is more sensitive to be pointing accurately at the noise source whereas I will be 'scanning' with the parabolic mic as a search aid to find the source.

Should have also stated that the app I am using is actually a BAT recorder so has all the recording etc capabilities that I need (and I do use audacity to post process) but the key for this problem is detection/discovery.

Noted the advice regarding electret mics tailing off at the high end.

I think the suggestion of separation from the mobile power supply would set me off on the right tracks and give more flexibility. However, looking at an even simpler option (not being an engineer) , if I did the filtering visually in the app (as i do today) could I just use something like this: https://www.adafruit.com/product/1713 ? Is the electret frequency roll-off between 10khz to 20khz going to be significant? If so, maybe use the MAX9814 with a different mic?


You have several problems if you want to power your amplifier and filter through just the bias for the electret microphone.

  1. Voltage. You've already mentioned this. There's only a couple of volts available as bias for the microphone on your cellphone. There are amplifiers that can operate on just a couple of volts, but they are rarer and more expensive than your run of the mill operational amplifier.

  2. Current. Besides the voltage, the available current is limited. There's typically a resistor of around 2000 ohm resistance in line with the bias. You get like 1 milliampere of current that you can use.

  3. You have to share the available voltage and current between your amplifier/filter and the microphone.

The solution is simple:

Don't bother.

Use a small battery to power an amplifier, the filter, and the microphone.

You have to build something to hold the microphone and the dish together. Just go ahead and make the handle large enough to hold a 9V battery (or other small battery.)

Find online calculator for a sallen-key bandpass filter and give it your frequency limits. It will suggest a circuit and a set of part values that will approximate your requirements.

Use that as your filter.

Use a simple op-amp amplifier for gain (to make it louder.)

Power it from your battery, and pack the whole shebang in the gadget you use to hold your microphone and dish.

A resistor is series with the output will take care of things if you are worried about your phone's microphone input. Shouldn't be a problem, though, as op-amps don't provide a lot of current.

Instead of a dish, you might consider one of these.

That link goes to a DIY "shotgun" microphone.

It looks like this:

enter image description here

That one covers a large range of audio frequencies.

The pipes are of various lengths to resonate at different frequencies.

If you make it using just lengths corresponding to the frequencies you are interested in, then you get your bandpass filter implemented acoustically.

You can still use an electronic filter with it, of course. And an amplifier.

Using just the frequencies you mentioned will make the whole thing shorter - more of a "shot-pistol" than a "shotgun."

Something else you should consider:

If you want to record the critters, then you may be able to get by without any special hardware.

Record using an uncompressed file format (.wav or aiff instead of .mp3 or .ogg) Use a high sampling rate, and 16 bit samples

You can then use a program like Audacity to filter and amplify the chirps.

You can gain a lot of volume that way.

There may be programs that let you do all of that live, on your phone.

The module you linked to is probably a very good start. It has plenty of gain (amplification) and will lower the gain if the sound is too loud.

The microphone loses maybe 3 dB from 10kHz to 20kHz. You would be hard pressed to notice the difference. Consider the loss of 3dB (factor of 0.5) as compared to the gain (60dB, factor of 1000.) The little bit of loss doesn't really matter

| improve this answer | |
  • \$\begingroup\$ The shotgun mic won't work. Try listening thru a short straw and see if that blocks all the audio except HF. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 2 '19 at 21:30

\$\lambda=\dfrac{v}{f} ~~~~~ v=1,100 f/s = 335.28 m/s\$ for the speed of sound in air.

The Helmholtz resonance occurs at f for half wavelengths, λ/2 and all harmonics or octaves of this frequency when the air impedance is matched to the tube but is λ/4 and all harmonics when it is shunted by a mic at the end.

Thus for 14 kHz, where \$l=\dfrac{v}{4f}= \dfrac{335 ~m/s}{14 ~kHz}= 12~mm\$

You can recalc for 6 to 20kHz.


This device in this picture won't work for near ultrasonic acoustic pressure because the wavelengths are too short and difficult to resonate with low l/d ratios.

This is 1/2" and is 1/2 of the length of the shortest piece in the classic shotgun mic using 37 alum tubes from 1" to 36" enter image description here

However the length to diameter \$l/d\$ ratio also has an effect on the Q of the resonator. Since you want high Q narrow bandwidth to reduce stray noises you need at least l/d = 10:1 or a diameter of 1.2mm then a large qty of these tubes to capture more sound pressure.

But then you have to contend with wind noise and vibrations so the exposed ends need a sock over the end.

Next comes the electronic design with a low noise audio preamp with band pass filter.

For recording I suggest Audacity so you can record, watch, view the spectrogram and even listen to it with the frequencies shifted down.

So what will work?

For acoustic pressure near ultrasonic f where the ambient sounds are noise and the signal is much smaller meaning SNR <<1 means you need a tuned piezoelectric sensor that blocks out human hearing range < 8kHz

enter image description here

This also works for Bat sounds.

| improve this answer | |

Not the answer you're looking for? Browse other questions tagged or ask your own question.