I'm building a heterodyne ultrasonic bat detector with a mems microphine (Knowles SPU0410LR5H-QB) and the SA612 mixer. I have built it on a breadboard and it works. It has some hum and noise but I have managed to reduce it. I can hear the sounds of a key ring about 10m away. I'm not sure if I would need more gain.

I'm designing a two layer PCB. I hope that it will perform a bit better than on the breadboard. All ICs are SMD except the ICL8038 and SA612.

enter image description here

What I am trying to do is to filter the microphine signal below 20KHz and the output of the mixer above 15KHz, more or less. Sound quality is not that important but I am trying to reduce the noise or posible oscillations.

The main difficulty is that the 7660S adds noise to the power rails, that's why I added some 10R resistors to Vin in the other IC's and decoupling capacitors everywhere. I have to replace it with the TC1044 that I hope is less noisy.

I'm using a voltage inverter because the minimum voltage of the oscillator (U4) is 10V, and I haven't found any other easy solution.

The ICL8038 is not easy to find and I don't know what could be a good replacement. It produces a quite nice sine wave of constant amplitude from 10KHz to 100KHz all the way with just one potentiometer. The SA612 is also not easy to find, but I've got some from aliexpress that work.

I'm not sure if I should add more gain to get better sensitivity. Now the bats are gone for winter holidays and I can only test with the key ring noise or similar. I have tried many configurations to improve the filter. This is still kind of arcane science for me, I'm just a hobbyist.


2 Answers 2


The design is neat. Very traditional, with all it entails - including ICs that aren’t made anymore.

One thing that stands out is using NE/SA612 at 9V. That is its maximum absolute rating. It should be running at a lower voltage - say 6V.

I’ll address the use of obsolete ICs.

The local oscillator for mixing could be a very mildly filtered square wave. 2nd order roll-off should be enough. The microphone won’t be picking up anything at the higher harmonics of the LO anyway.

Alternatively, with the LO set to a low end of the range, the harmonics help hear higher frequencies without a need to sweep the LO - the harmonica are like having multiple LOs at once. So, if anything, you could add a high-pass type filter on the square wave LO to get the harmonic amplitude up, and feed that to the mixer. That way you can keep LO fixed, and will easily hear every ultrasonic input up to say 200kHz or the limit of the microphone.

The mixing could be done with a pair of op-amps to amplify and invert the input, and a CMOS switch (4016 or 4066) to flip the polarity of the input signal. On the output of the switch, the downmixed signal would be at DC. Bandwidth would be determined by a low pass filter on the output of the switch.

So, the circuit could be simplified quite a bit that way, and it would use ICs that you can buy.

So, I’d see the chain as follows:

  • 555 square wave source
  • gain stages after input
  • -1 inverter stage
  • switch to demodulate the signal (clocked by the 555)
  • low pass
  • gain stage
  • variable attenuator (volume control)
  • output driver

ICs needed: quad op-amp, 4066, 4041 (preferred for complementary switch drive) or 4069, 555.

All of those ICs, opamps included, are still made in a DIP package by TI, if you wanted to go that route.

The whole thing can then run from a single 12V or 15V supply for both analog and digital ICs, no need for negative supplies.

You’ll likely need some shielding around the input amplifier and microphone. A fine copper mesh will work well for that purpose. That’ll help keep electrical noise from external sources at bay.

  • \$\begingroup\$ Ok, many good ideas! To start with the square wave, I had consired it, but what you mean is to keep all the harmonics in the range separated about 5 or 10 KHz? it would translate all sounds in the range to the audible zone at the same time, right? That would be perfect and I would get rid of both noisy IC's. That could be a first improvement. I think I have the 555 both in dip and smd. I prefer smd for the pcb, and for testing in the protoboard I have adapters, but if I have both versions it's easier \$\endgroup\$
    – Gos
    Oct 27, 2023 at 17:41
  • \$\begingroup\$ It will be powered by a 9V battery, and I think that it's ok with all these IC's. For quad opamps I have the TL974, smd and dip. \$\endgroup\$
    – Gos
    Oct 27, 2023 at 17:51
  • \$\begingroup\$ The SA612 is running at 7.5V, I have put the zener there for that. \$\endgroup\$
    – Gos
    Oct 27, 2023 at 17:57
  • \$\begingroup\$ The topic of the switch is not completely clear to me. I have been checking it and I understand that I should connect the switch with an opamp in a way that multiplies the signal by +1 or -1 based in the square wave, right? \$\endgroup\$
    – Gos
    Oct 30, 2023 at 20:25

Following the hints from @Kuba I have reworked the circuit and it works very well on the breadboard and solves most of my questions.

Now the mixing is done with analog switches as suggested. It simplifies everything, fewer components, almost no noise on the rails and all available and with SMD version.

enter image description here

The two analog switches in the middle are driving the opamp U4A inverting/non-inverting the signal alternativelly, clocked by the asymetrical square wave produced by the 7555 timer. It can be done also with just one switch, but in the package are two and are complementary, so I can build it very simply.

The 7555 timer generates a variable asymetrical square wave, this way I get all harmonics, not just the odd ones. With just the second or third first harmonics (e.g. 20, 40, 60 KHz) I think that I can cover a wide range up to 100KHz just moving the potetiometer a bit, but I have not yet tested with the higher frequencies.

Testing with the keyring it detects it from more than 25 meters away clearly, and rubbing fingers from about 2 meters.

With the scope and injecting sine waves it works fine also. I would like to test ultrasound at concrete frequencies, maybe with an ESP32 development board and an ultrasonic microphone when I have some more time.

I will order a PCB and later check the isolation as well.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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