Picture 1

batty_het-1 (mp3)
Four batty pulse trains, as heard through an sk207 detector. The first two are recorded with the detector tuned below batty's ultrasound frequency, the last two are recorded with the detector tuned above that frequency.
The parameters of the batty call are of course the same as measured with the division detector.
It is already clear from the sonogram that if the detector is tuned below the detected frequency, then a descending frequency will be detected correctly (as a descending frequency). But if the detector is tuned above the detected frequency, a descending frequency will sound like a rising frequency. This is not unexpected, but it's still good to keep in mind.

Picture 2

batty_het-1b (mp3)
This is a closeup of 4 pulses, 2 at each tuning frequency. Although the plot looks like Picture 1, the time scale is very different. The blocks in Picture 1 are each pulse trains 12 pulses long; Picture 2 is just 2 pulses from 1 train and 2 from another.

The sonogram shows that a heterodyne detector is more sensitive to frequency shifts. The rich set of harmonics, at multiples of the shifted frequency, is due for the most part to harmonics of the original signal mixing with those of the detection signal. Those two features are the strengths of this type of detector.

The tuning frequency can be estimated from the sonogram and the division detector result (52.8 khz). In the first case tuning is around 52.8 - 2.5 = 50.3 khz and in the second, case around 52.8 + 2.5 = 55.3 khz. By accident the offset (2.5) is about the same in both cases. This also illustrates the weakness of this simple heterodyne detector: I should be deducing the bat frequency from the tuning, not the tuning from the bat frequency! Some way of recording a reference frequency is essential.

Picture 3

Closeup inside one of the pulses.