Bats ultrasounds detector with LM324

Question

I have some questions about this small project, a device to detect bat ultrasounds and reproduce in a speaker. It's based in a post of the Element14 community, Episode 578. I have adapted it and I'm testing in a breadboard. It somehow works but not very well, it's unstable and lossy. Here is my schematic:

This is my understanding of the concept. Bats emit ultrasounds to navigate and locate insects for dinner, in different frequencies from 20 KHz to more than 100 KHz depending on the specie. Humans can hear in the range of 20 Hz to maximum 20 KHz, so we need to transform the signals. To do this I use a ultrasounds sensor and the 4 op-amps in the LM324N package. Two of them for two pre-amplification stages. I suppose that it is because the LM324 has not a high gain for the range of frequencies around 50 KHz.

Then there is an oscilator with a potenciometer that produces a wave adjustable between 12 KHz and 90 KHz aprox. I mix this wave with the 'bats signal' to get a heterodyne signal with all the combinations of the frequencies: separated, added and substracted. We are interested only in the substracted signal, the others will be higher and can be filtered. For example, if a bat is emiting at around 40 KHz and I adjust the oscillator to 35 KHz, the substration will result in a signal of around 5 KHz. Then this signal is sent to a power amplifier and speaker.

The microphone nominal frequency is 40 KHz, I don't know which usable range it has. Sound quality is not important as you hear mainly 'clicks' time to time when the bats are near.

After a lot of debugging it works, I can hear the 'clicks' of the bats, but it's quite unstable. As the bats are not always willing to collaborate, for testing I move a key-ring in front of the sensor, it produces many frequencies.

I'm not sure if it is working properly or could be improved. I have some doubts:

• The pre-amps should filter the frequencies below 20 KHz. Is this done by C3? Or by C2 and C5? they look to me like a low pass filters to remove noise
• I check the oscillator with the oscilloscope in the frequency domain. It produces a kind of triangular signal with a main freq but several higher peaks separated about 20 KHz. If I move the potenciometer it goes from 12.5 KHz to 90 KHz and the other peaks move also, sometimes there are less. In theory it should be a single square wave, right?
• At the output of the pre-amp stages there is always a constant signal of 55 KHz and about 280mV in the 2nd stage. It's only a bit smaller than the mic signal, I don't know if this comes from the mic itself. When I connect the probes directly to the microphone the oscilloscope introduces more noise.
• I don't know if the rest of higher frequencies go to the speaker and could disturb the bats, after the mixer I see freqs in all ranges. Maybe the amplifier and the speaker aren't able to manage those freqs anyway. In the output of the amplifier I can't see it clear.

I would like also to be able to identify at which frequencies actually emit the bats. I guess that it is around 40 KHz, but I'm not sure. Any hint on improving it or testing would be very helpfull.

Thank you!

Answer 1

Any hint on improving it or testing would be very helpful.

You need a proper mixer (multiplication) and not an audio-type filter (linear addition) if you are going to successfully convert ultrasonic sounds to within the human hearing range with decent sensitivity and accuracy.

Maybe if sufficiently overdriven the LM324 inputs can form a crude multiplier but, at best it will be insensitive and unreliable.

I check the oscillator with the oscilloscope in the frequency domain. It produces a kind of triangular signal with a main freq but several higher peaks separated about 20 kHz. If I move the potenciometer it goes from 12.5 kHz to 90 kHz and the other peaks move also, sometimes there are less. In theory it should be a single square wave, right?

A square wave or triangle wave naturally produces harmonics. That is what you are witnessing. Only a sinewave produces no harmonics. Square wave from wiki: -

Also, due to the harmonic content, a square wave may not be as useful as a sinewave when converting to baseband.

At the output of the pre-amp stages there is always a constant signal of 55 kHz and about 280mV in the 2nd stage. It's only a bit smaller than the mic signal, I don't know if this comes from the mic itself. When I connect the probes directly to the microphone the oscilloscope introduces more noise.

That sounds like a bad PCB layout problem or, maybe you are fouling things up using breadboard?

Answer 2

^{simulate this circuit – Schematic created using CircuitLab}

A piezo microphone can go directly to mixer input pins (1,2). The inductor L1 might require a different value depending on piezo characteristics. Its purpose is to broaden bandwidth of a piezo transducer. The SA612 shouldn't be supplied with DC voltage over +8V.

Local oscillator sine wave can be applied to pin 6. If amplitude is too high, the mixer begins to respond to odd-order harmonics. However, gain increases when you drive pin 6 with amplitudes up to 300mV.
Construction must be done with care - this mixer is sensitive to many hundreds of Megahertz.

Another stage of audio amplification might be needed. OA1 is a differential amplifier...R4 = (R5*R2)/(R5+R2).

Answer 3

Finally I got some parts and some time and I have also found more information in the direction of using the SA612 as @glen_geek suggested.

I'm using a ICL8038 to generate an adjustable sine wave from about 15KHz to 100Khz. For what I have been reading a square wave should also work. I have reworked part of a pcb from a kit: With the last potentiometer I reduce the amplitude to 200mV. It is constant all the range.

And the circuit that I'm working on:

Here: https://www.teensybat.com they are working in this kind of devices since years ago. This could help other people looking for information about bat detectors.

They have a similar version: https://www.teensybat.com/simple-heterodyne-fd and have also a nice one with a MCU, mini audio card overclocked to work up to 100KHz, display, etc. They have a lot of resources, information, detailed explanations, hints, other approaches like the freq divisor, etc. For example: http://bertrik.sikken.nl/bat/ne612het.htm

I think I will have to work in the noise and interferences, maybe move to a soldered board.