Listening to Bats

© Brooke Clarke 2007 -2010

After reading Chapter 2 of The Blind Watchmaker by Richard Dawkins which goes into some depth on how he thinks bats use echolocation I'm writing this section.  I spent a couple of decades in the Radar Warning Receiver business.

There are a number of different way that RADAR can work.

Pulse Time of Flight (Wiki: RADAR)

This is the classical method.  A pulse is sent out and the time of flight is measured.  The result is you know the distance to the target.  It can be done using sound, radio waves, or when light is used it's called Laser Range Finder.  Note the result is a number, i.e. the time to the target.  To get a Plan Position Indicator (PPI) display the antenna can be rotated about a vertical axis.  This gives you a map but you don't know the elevation of the target or it's speed.

Doppler (Wiki: Doppler Radar)

This is what's used in the police radars that look for speeding motorists.  They transmit a Continuous Wave (CW) signal and the received signal has been offset by the doppler shift of aynthing moving relative to the transmitter.  If you looked at a spectrum analyzer plot (the HP 4395A would be ideal for this) of the audio output of a police radar system that was in a police car traveling down the highway you would see a signal from the road, trees and all the stationary objects at a doppler frequency related to the speed of the police car.  Note that this would not be a single narrow spike on the spectrum display but rather it will be spread because of the angular offset of things not directly in front of the transmitter antenna.  If there was an oncoming car going at the same speed as the police car there would be a signal at twice the stationary object signal.  If there was a car approaching from behind the police car (most car mounted speed radars are really two units one pointed to the front and one pointed to the back) at half it's speed there would be a signal at half the stationary object frequency.

In the case of a ground mounted search radar there are echos from nearby stationary targets that are of no interest and that can hide more interesting targets.  By using some doppler processing a Moving Target Radar can be made where the returns from stationary targets are not shown, only moving targets show up.  This can be refined by only showing targets that are moving between two specified speeds.  In the case of bats it may be that the speed of a moving target would be "seen" as a color.

So CW doppler is good for learning about the speed of things that are approaching or moving away.

FM Doppler (Wiki: FMCW)

By Frequency Modulating (FM) the CW signal, say with a linear ramp now you can determine the range to a target.  This type of system uses much less power than a pulsed system and has better suppression of background clutter.  One application is in artillery fuses that can be set for a distance above the target.  A bat that used FMCW SONAR would know how far away a target was and if it also used CW it would also know the relative velocity of the target.

These was some thought of equipping all cars with FMCW radars as the heart of a collision avoidance system.  But this did not happen once it was realized that if all cars had these systems they would interfere with each other and that would cause them to fail.

Matched Filter (Wiki: Matched Filter)

The signal to noise ratio of a bats echolocation receiving system would be improved if it used what's called a matched filter to pass the desired signals and reject the undesired signals.  It's probable that bats are very good at this.  For example they could tell the difference between echos from their transmissions from signals from other bats.  They might even be able to use the echos from other bats to better "see" targets?  In The Blind Watchmaker Dawkins mentions there's a bat that changes it's transmitter frequency so the received frequency is always the same.  That's an excellent way to make use of a band pass filter (Wiki: BPF) in the receiving system to improve the singal to noise ratio.

Signal Domains

There are a number of ways of looking at the signal coming from a bat.

3 Dimensional Image

Note that all the above RADAR (SONAR) systems need some sort to scanning to get a graphical display.  Humans have only two ears, yet we can tell where in three dimensional space a noise is coming from (that's why a Home Theater system is such a large improvement in movie watching).  At first blush it would seem that with only two ears we should not be able to have 3D hearing.  But it turns out that the external ear is part of a matched filter that colors the sound and that color adds the missing information needed to give us 3D hearing.  In a similar way a bat can color the sound two ways:  first it may be colored when it's sent (maybe related to the shape of it's head) and second by it's ears.  This would allow the bat to "see" a 3D image of it's surroundings.

This image might have features related to fixed objects and different features related to moving objects.  For example the "color" of a moving object might be related to it's relative speed and size.  I remember a TV program that said frogs can "see" small objects that have convex shapes and are moving (i.e. small flying insects) but can not see a stationary insect and will die of starvation when there are a lot of dead flies nearby.  They also see predators, not like we do, but instead by their size and shape (or by their shadow).  So not only would bats "see" their pray they would see other bats in a similar way.  Since they have eyes the echolocation signals are probably combined with the visual signals to form a composite "multi spectral" image (Wiki: MSI).   Note the a multi spectral image is classicaly only made up of light at different frequencies so this combination of a sonar image and an optical image might be called a multi sensor image.

Echolocation Links

 - Echolocation in the Bat -this is a more in depth coverage than in Dawkins book (might be where he got the info)

Hetrodyne

By mixing a local oscillator with the amplified output from a microphone the bat's ultrasound is changed to a frequency that  you can hear.  The mixing process preserves the amplitude modulation on the chip so close bats are louder than far away bats.  Once you know the frequency for your local bats you probably don't need to change it.  The bandwidth of a single bat species is typically less than 10 kHz.

Frequency Division

The idea is to amplify the bat ultrasound then use it to clock a digital counter chip.  The output can be taken from some divisor that brings the ultrasonic frequency down to into human hearing range.  The two disadvantages are that you loose the amplitude information so can't tell if a bat is close or far away and the width of the "chirp" gets compressed so the fidelity is not as good.  But the good news is that there's no tuning required.

There are some ideas floating around that would keep track of the amplitude informatin and use it to modulate the frequency divider output, but I don't know if this has been done or is available.

SDR-IQ

Board, software & cable (box was seperarate part then)	PCB

If some wide band of frequencies are recorded like 15 to 150 kHz for example then it can be played back and all viewed on a spectrum analyzer.  You can see subtile details on the spectrum analyzer that your ears can not distinguish.  The Software Defined Radio called the SDR-IQ has this capbility.  I.e. it can record 500 Hz to 190 kHz directly and can show a real time frequency spectrum for that band or any smaller band. While it's displaying the wide band you can put a cursor on a bat frequency and it will demodulate that down to baseband.  I haven't yetused the SDR-IQ for bats so don't have details about modulation type, bandwidth, etc.

It's powered by the USB2 cable and has a BNC input jack.  The DB-9 connector is so that the SDR-IQ can control radios that may be acting as RF front ends for frequencies above 30 MHz.  For frequencies of 30 MHz and lower the SDR-IQ can receive them directly.  The hardware on the board mixes the input and a Direct Digital Synthesizer supplied Local Oscillator the I & Q channels of the mixer output get sampled and the digital data stream gets decimated down to a bandwidth that the USB2 port on a computer can handle.  The Spectraview software runs inside the PC to further tune and demodulate the digital data.  The sound card might act as the audio output with the SDR-IQ, but for the stock setup is not involved with the reception.
Electronic Design -March 9 2010 Software Defined Radios Are Here Now

PC Software

Spectravue is a spectrum analysis package
Winrad is a radio interface (Yahoo Group, Winrad, Radio Portal)

Accessories

Griffin PowerMate - USB Knob
Andrea PureAudio - full duplex USB sound card
SDR-Radio - Allows making SDR-IQ internet accessable
G4HUP Reference Locking the SDR-IQ and SDR-14 receivers - Docs - from a GPS disciplined 10 MHz source
500 kHz Low Pass Filter - PAR BCST-LPF - blocks AM broadcast stations to prevent front end overload. McKay Dymec DP 40 could also be used.

Recording

A normal audio recorder can be used on the headphone output from the heterodyne or frequency divider type bat detectors.

An ultrasonic microphone could be amplified and recorded on a recorder that could handle the bandwidth (not a normal Hi-Fi audio recorder).  Such as an instrumentation recorder or a video recorder with an attachment would work.  Modern Instrumentation recorders are based on digital techonlogy related to digital TV recording and have total bandwidths similar to a TV channel.  For example the Kinetic Systems DAQ848 has 48 channels each just uder 100 kHz wide or about 4.8 MHz total bandwidth.

A more advanced approach uses direct to hard drive recording, such as done by Wide Band Systems.  The bandwith is extended by using RAID.

The SDR-IQ (see above) can record a very wide band (up to 190 kHz) directly to a PC USB2 port.

Microphones

The normal audible range electret type condenser microphone has some response above the 20 - 20,000 Hz audio range.  There are also ultrasonic speakers and sensors made to work with remote control applications but these are usually optimized to operate over a narrow band and are not suitable for the 15 to 150 kHz bat band.  There are commercial products that work very much like a bat detector called "ultrasonic leak detectors" but unless you can find some specification about frequency coverage it's hard to know what they cover.  There are also hydropones with the needed frequency coverage, but they are designed to work under water and their performance in air is rather poor.

The other type of microphone is the condenser mike specified to work in the ultrasonic range.  It would be nice if there was an electret type made for ultrasonic since then you would not need a DC bias supply, but as far as I know they don't exist or are very expensive.  That leaves the standard condenser microphone.  Probably more properly called a transducer since it can make ultrasonic sound as well as it can detect it.

The one that's probably been made in the highest quantity is the one used on a number of Polaroid cameras for the auto focus function.  Details on the Polaroid Sonar One Step Camera are on a seperate web page.

Knowles Acoustics makes a wide band ultrasonic sensor that does not reuqire bias.  But it's a surface mount technology device and needs a local op amp as well as some RFI supression.

The speed of sound in air is about 1124 feet/second, at 50 kHz a wavelength is 0.27 inches. 

If a sound wave impinges on a microphone diaphragm on it's central axis then the diaphragm can be many wavelengths in size and will work better because of the large diameter.  This is the case with the Polaroid Sonar type transducers.  They are used in an application where the outgoing and incoming wave fronts are on their central axis.  That's why in this particular application the diaphragm size can be many wavelengths in diameter, but for general purpose microphones it's much smaller.

If an omni directional microphone is used to pickup sounds from a random location and it's diaphragm is comparable to a wavelength the average sound pressure can vary because a peak may cancel out a valley.  In this case the diaphragm needs to be small compared to a wavelength to get flat response.

I expect that when the Polaroid Sonar transducer is used as a bat microphone it's response will have peaks and valleys that are quite deep (more than 10 dB) and spaced close to each other.  But as the bat moves it will be moving through a number of these and the peaks will have a much higher output level than  you would get using one of the much smaller professional type microphones.

Bertrik's bat detector page -

Intensity control during target approach in echolocating bats; stereotypical sensori-motor behaviour in Daubenton's bats, Myotis daubentonii -

DigitalBatDetector - direct to digital then process


Back to Brooke's Products for Sale, Radios, Military Information, Sundials, Time & Frequency, Personal Home page.