Converting the ultrasonic signal down to the human audible range can be performed in some different ways: frequency division, heterodyning and time-expansion. Also envelope detection can be used.
These are explained below.

The frequency division method is in my opinion the simplest method to implement, but the spectral characteristics are not so good, because the frequency details are compressed.
The heterodyne method sounds much better and is theoretically better in noise performance because of the limited bandwidth. It requires however tuning to a particular frequency, although a lot of bats emit sound at a wide range of frequencies, so it is not very likely to completely miss a call.
I think time expansion is the best method for analysis because all frequency and time details are kept at full resolution, but is not good for real-time listening and very hard to build.
I don't know much about envelope detection yet.

Frequency division

See also Tony Messina's Enhanced simple bat detector.

A frequency divider converts the whole ultrasonic spectrum into the human audible range by dividing the incoming frequency by a fixed factor. If this division factor is 16 for example, a 40 kHz bat call would be converted to 2.5 kHz.
Drawbacks of this conversion method are that only the dominant frequency will be converted and that the amplitude information is lost.

Amplitude-retaining detector

An amplitude-retaining detector maintains the loudness of the bat signal in the divided-down output signal, to give a more faithful representation of the bat call. This allows you to estimate the distance of the bat you're hearing, for example.

See my frequency division ideas for some sample circuits.
Im also thinking about the design of an advanced frequency division detector.


In a mixing detector, the bat signal and a locally generated oscillator signal are multiplied. This results in two new frequency components, one with a frequency equal to the sum of the original signal frequencies and one with a frequency equal to the difference of the original frequencies. The difference frequency is the one we're interested in.
In contrast to a frequency division type detector, a mixing detector must be tuned to a certain center frequency. Making an unfortunate tuning choice can cause you to miss a bat call. On the other hand, it may be possible to discriminate between two different species by selection of the listening frequency.

In my personal opinion, a mixing detector sounds much nicer than a frequency division detector. The ultrasonic spectrum is shifted down instead of compressed, so you can hear differences in frequency very clearly (maybe even Doppler shifts). Also the theoretical signal-to-noise ratio is better (up to 13 dB), because of the limited bandwidth.

Direct conversion

With direct conversion, the oscillator signal and the bat signal are of approximately equal frequency. The signal that results from multiplying contains a difference frequency component in the audible range and a high-frequency sum frequency component. The high-frequency sum component is removed by a low-pass filter, while the audible difference frequency components is passed.
A disadvantage of this method is that bat frequencies slightly higher and slightly lower than the oscillator frequency sound the same. For example, bat signals of 39 kHz and 41 kHz both give a 1 kHz difference frequency when multiplied by a 40 kHz signal. Bat calls often consist of downward frequency sweeps, so a direct conversion detector would yield a combination of a downward and an upward sweep in the audible frequencies, when tuned to the middle of the bat spectrum.

Heterodyne converter

A method to eliminate the problem given above, is to use two mixing stages. In the first stage, the signal frequency is translated up to a high frequency, where the desired part of the signal is filtered out. Then the signal is translated back down to the audible range by another mixing stage.
For example, in the first mixer the ultrasonic sound of 40 kHz is mixed with a signal of 495 kHz. This gives you a frequency difference signal at 455 kHz, which can be filtered by a 455 kHz ceramic resonator. In the second mixer, the 455 kHz signal is mixed with a 458 kHz signal to give a 3 kHz difference signal. The slight mismatch between the filter central frequency (455 kHz) and the second heterodyning frequency (458 kHz) makes it possible to select only one set of difference frequencies.

Often the direct conversion method is also called heterodyne, but strictly speaking I think it is not. With direct conversion the oscillator signal and bat call signal are of approximate equal frequency, differing only a few kHz, while with true heterodyning they are much more different in frequency.

AM-radio conversion

The normal operation of a MW-radio is to mix (multiply) the antenna signal with the signal of a tuneable oscillator, then filter this through a fixed 455 kHz filter and recover the envelope of the resulting high-frequency wave. By replacing the antenna with a microphone, changing the range of the oscillator and adding a beat-frequency-oscillator (BFO), the MW-radio can be converted to a bat detector. The BFO signal is injected (added) just before the envelope recovery.
Adding the BFO signal results in a high-frequency signal modulated by a signal of a frequency that is the difference of the input frequencies. The built-in envelope detector then recovers the low-frequency envelope and turns it into an audio signal. The beat-frequency signal is essential in the conversion of a radio into a heterodyne detector. Without it, no tonal quality is heard and the detector works like some kind of tuned envelope detector.

This is a nice illustration of the non-linear mixing principle: by adding the frequency-shifted bat signal to the BFO-signal and putting the sum through the (non-linear) envelope detector, the frequency difference between the two signals is converted into audio.

Take a look here for some oscillator and mixers ideas.

Time expansion

With time expansion the bat signal is recorded (sampled at 300 kS/s for example) and played back at a lower rate (e.g. 10 times). Bat sounds are heard at 1/10th the frequency and 10 times the duration, so no frequency information is lost as is the case with heterodyning or frequency division. A 5 ms bat call at 40 kHz would sound like a 50 ms call at 4 kHz. A time expansion detector needs some kind of memory to store the bat call.

Envelope detection

With envelope detection, only the envelope of the bat call is preserved. A bat producing 20 calls per second will sound like a 20 Hz tone (with some extra harmonics). The amount of information that can be obtained with this method is very limited.

Envelope detector

On the right a possible circuit for deriving an envelope signal is shown. This circuit looks a bit like a voltage-doubler with an opamp inserted in the middle.
The combination C1 / D1 pushes the negative parts of the incoming waveform up. The signal is then rectified by diode D2 and stored on capacitor C2. By taking the feedback for the opamp from capacitor C2, the influence of the voltage drop of D2 is reduced (The opamp/diode combination is a so-called super-diode). The resistor R1 parallel to C2 determines the hold time of the envelope detector. I guess that R1*C2 should be around 1 ms or slightly less.

Synchronous rectifier

Diodes act like little switches that go on when the voltage across them has the right orientation. A disadvantage is that they only switch on above a certain forward voltage (0.6V for silicon diodes). Ofcourse you can do much better!
Instead of relying on the diodes, you can use a sensitive comparator to determine voltage polarity. The comparator output signal is then used to switch between the input signal or the inverted input signal. This type of circuit is called a synchronous rectifier and it is actually quite similar to the design of a heterodyne mixer!

This page was last updated Saturday, June 24, 2000