LM567 or Custom bandpass filter

Question

I'm using a array of hydrophones to localize an underwater pinger (37.5 kHz and 45 kHz) and I'm thinking of using LM567 tone decoder as a frequency filter. The other option is to design a custom bandpass RC/RLC filter.

1. What are the pros and cons of both options?
2. What order and type of bandpass filter should be built and why?

Edit: this will be deployed on an AUV with limited space. Planning to use a Teensy 3.6 ARM board for pinger localization algorithm.

Answer 1

Neither options seem very reasonable in 2019.

• the RLC is unnecessarily bad a filter, unless you do many stages, which becomes very complicated,
• tone detection is very sensitive to more than one tone being present, so not usually an option for audio / hydrophones.

You'd want to do your localization in digital anyway, so do the filtering in digital, too.

You don't need very much bandwidth by modern standards. In fact, this would be pretty trivial to build with but a cheap soundcard (e.g. the one integrated in your PC/Laptop/shmartphone) with few external components.

You'd want to use your sound card to sample the hydrophones. Now, sadly, soundcards are meant for humanly audible sound, so you can't use them to pick up 37.5 kHz (or even higher) directly.

What's easily possible, however, is frequency-mixing your 37.5 kHz and 45 kHz down to lower frequencies that you can, in fact, easily pick up with a sound card. Then, in software, throw some bandpass filters at the problem – software bandpass filters can be arbitrarily steep (given that audio sampling rate signal processing doesn't pose a serious load for any modern smart phone or PC).

So, what you'd want to do is

1. Generate a tone at frequency f (we call that local oscillator, LO), so that both |37.5 kHz - f| and |45 kHz -f| are within what your soundcard can record. Notice the absolute value there!
2. Use that tone to mix down your very coarsely (e.g. single-stage RLC filtered) filtered hydrophone signal.
3. Digitize the mixing results with your sound card, and build digital filters to select |37.5 kHz - f| and |45 kHz - f| as narrowly as you want, to maximize SNR.

Tadah, you've just built a superheterodyne receiver for hydrophones, with IF filtering implemented in software, where filters are free and trivial to construct precisely, and for location purposes very importantly, with linear phase and thus defined group delay. That's not even possible with RC/LC/RLC, mathematically!

A down-mixer is just any nonlinear device (e.g. a diode) that you feed with the sum of the LO and the high-frequency signal. Then, the result is low-pass filtered (single-stage RC filter), to eradicate all the harmonics you don't want.

If you're able to do a bit of digital design yourself: get a microcontroller with as many ADC channels as you need. You can do bandpass subsampling to directly mix down the signals with (multiples of) the sampling frequency. That would still require you to do a very rough analog bandpass filter, but definitely minimize the complexity of your design.