Beware of reflections from your detector

You may be using a detector that has a microphone that clearly sticks out from the detector itself, for example the D980 or the D1000. This in contrast to the detector type, such as the D240, that has a built-in microphone. Here I would like to talk about a problem you may (or may not) encounter with detectors that have protruding microphones, such as the D980. This article is based on some limited experience I had in using the D980, many years ago. I am not 100% sure whether what I will claim here is correct, but if it is, it would be quite a serious problem, that is probably quite easily solved. It is up to the users of D1000s, D980 or similar detector types to check what I claim. Please report it back to me so I retract or confirm. Just to be politically correct: I very much like the equipment Lars Pettersson makes and I guess the D1000 is probably the best detector that presently exists! The problem I encountered with the D980, or at least I thought I did, is depicted below in the drawing:


Pictures 1 to 3 show in temporal order what could happen during the reception of a pulse. As you can see, the detector on the drawing has a microphone, protruding from a flat surface. Both the microphone membrane and the surface are therefore perpendicular to the incoming sound of the bat. On picture 1, you can see an enlargement of a sound "ray" on its way to the microphone. Of course, such "rays" are spreading everywhere, but we only see the one that will hit the membrane of the microphone. In picture 2, from all the rays from the sound field hitting the detector, only two are drawn this time. We see the previous ray that already has been partly "absorbed" by the microphone, but we also see another ray that is about to hit the casing of the detector. Picture 3 shows what happens next: The ray reflects from the casing, but on its way back to the bat, it interferes with the sound wave that is hitting the microphone membrane. The output of the detector therefore shows the typical interference notches (oscillogram left). In reality, however, the bat doesn't produce these notches at all, of course! The oscillogram of its true emission is shown on the right.

This simple drawing proves nothing. However, when I had a D980 at my disposal I had the strong impression that this was indeed happening. I cut out a triangular shape from cardboard with a hole for the microphone and when I used it I had the impression the problem was no longer there. Simple foam with a hole also worked.

The interference problem, causing the notches, will only show up when making on-axis recordings of bats, but since this is the best way of recording bats we should never deliberately turn the microphone away from an approaching bat. I got a bit worried lately that the interference problem may really exist because I received quite a number of nearly noise-free recordings that nearly ALWAYS showed notches in the oscillogram. I realise that this may just be a coincidence. Water can act as a reflector and so can nearby walls, the ground or even trees. Still, if this problem is real and is so easily solved I would recommend to do some tests. A test one could do is pointing the D1000 (or whatever box you have) at a TV (produces 15 kHz constantly) that is on and make a recording from a distance, perfectly on-axis and also a bit off-axis. If you see an amplitude modulation on-axis, that disappears off-axis there really is an interference problem. If not, I will take this warning offline immediately. Good luck!

Technical note from Volker Runkel

There is even worse happening than the echos bouncing back from the device: the microphone area is large enough to get interferences of sound waves arriving at an angle to the microphone (and probably also from frontal) leading to extinction or amplification of frequencies (dependent on the angle of sound incidence).
We talked a couple of years ago to professor Lerch here in Erlangen, who was working on the Circe project (the roboter bat head) and he just laughed when he saw the huge mebranes of the mics (D980/D1000). He told us he would use them as a speaker only. Even so, their sensitivity is fabulous.
Below are some graphs that illustrate the problem:



Last updated 26.1.2009