Build Direction Finder Apparatus Under £100

[wolram]

Hi all, i want to build an apparatus that can locate the direction an
intermitent sound came from.
I imagine if i had four microphones pointing north east south west, i could compare the signals, if say directly north the north mic will give highest reading, if south east it would me a mix of south and east.
so could i build a circuit that would be accurate to 5 degrees
and record the direction?
This is a project i would not want to spend more than £100 on.

 

[berkeman]

Will there be echoes? If there are echoes, the problem becomes more difficult.

Even if there are no echoes, it would still be best if you could digitize the sounds received at 3-4 omnidirectional microphones spaced apart, and compare the time difference between the arrival of the sound at each mic. To do that, though would require at least a 3-channel USB oscilloscope with some kind of triggering mechanism, which I don't think can be bought for £100.

What is the nature of the sound? Is it a lot louder than the ambient? How would you trigger the measurement to take place?

 

[wolram]

Will there be echoes? If there are echoes, the problem becomes more difficult.

Even if there are no echoes, it would still be best if you could digitize the sounds received at 3-4 omnidirectional microphones spaced apart, and compare the time difference between the arrival of the sound at each mic. To do that, though would require at least a 3-channel USB oscilloscope with some kind of triggering mechanism, which I don't think can be bought for £100.

What is the nature of the sound? Is it a lot louder than the ambient? How would you trigger the measurement to take place?

Thanks for reply Berkeman,
I am attempting to find the source of a noise akin to a marble being
droped onto a wooden floor, there would be little if any background noise,
so i had hoped that the apparatus could be quite crude, but i admit i have
no idea as to how to record the detected sound.

 

[rbj]

if there isn't really nasty reflections, you can locate the direction of a sound anywhere along a plane with three omni-directional microphones placed on that plane. it's probably best to position the microphones in a nice equilateral triangle.

using cross-correlation, you can compute the difference in arrival time of a sound between any pair of microphones. from that difference in arrival time, you can compute an angle of direction of the source (sorta like the Blumlien stereo patent of the 1920's), but this would really define a cone (with some angle) in which the source would lie upon hence the need for the third microphone to make it unambiuous.

this can be and has been done. but it needs a little DSP computation.

 

[light_bulb]

if you knew the frequency ahead of time you could use comparators.

 

[wolram]

if you knew the frequency ahead of time you could use comparators.

Alas, i do not know the frequncy,

I am trying to find the source of a noise that is very infrequent
and short lived, the source would be no more than three metres
from the apparatus.

 

[light_bulb]

then the only requirement would have to be that it would have to stand out from whatever background noise their is, this can still be done just with more comparators (eg. lighting leds or parallel port interfacing) or you could buy two cheap soundcards and use the trial on a multitrack recording program. if you want to get fancy use a waves 3D spatial plugin ment for surround sound mixing to pinpoint the location.

 

[wolram]

then the only requirement would have to be that it would have to stand out from whatever background noise their is, this can still be done just with more comparators (eg. lighting leds or parallel port interfacing) or you could buy two cheap soundcards and use the trial on a multitrack recording program. if you want to get fancy use a waves 3D spatial plugin ment for surround sound mixing to pinpoint the location.

Thanks Light-bulb, the led idea sounds great, is there any chance
of a cicuit diagram or rough idea where to start, i guess i could build
the circuit on vero board.

 

[light_bulb]

i just visualize an op-amp amp then add comparitors and set a voltage threshold to trigger a latch(or closed loop gate) which keeps 3 leds on that are closest to the direction of the sound.

 

[Lawst]

Could someone in layman's terms describe to me what I can use to accomplish the following two objectives:

1. Find the direction of an noticeable sound at a range of 2 or so miles
2. Recognize sound pattern algorithms and produce an analysis based on those pre-determined algorithms. ie: one bang = x two bangs = that.

cheers

 

[Bob S]

Many years ago I had to locate the position of an EM pulse to a very small fraction of the distance between two pickups (about 0.1 mm out of 35 mm separation). The pulses were about 300 nanoseconds long and modulated at 50 MHz. The amplitude dynamic range was about 100:1. The final design used the equivalent of quadrature hybrids (using delay lines) to convert the amplitude disparity to a time (phase) disparity, and then a double-balanced mixer to get a phase difference, which was independent of amplitude. It was later discovered that the sound direction finding system in the female cricket also uses AMPM (amplitude modulation to phase modulation) in a tracheal tube to find chirping male crickets. Perhaps cardioid mics could give direction-dependent amplitude. [added] A marble dropping on a hard floor is a high frequency impulse, and would make a bandbass filter (~500 Hz) ring.
Bob S

 

[Mike_In_Plano]

Cheesy++ Solution

(8) electrete microphones
(2) TL084 op amps
(2) LM339 comparitors
(1) PAL22V10
(8) LEDs
(1) 3 cell AA battery pack with batteries
(1) adjustment pot
(1) reset switch (momentary)
(misc) R's and C's

Build ring of mics, each with an LED. The op amps buffer the mics and the comparitors detect when a mic reaches a level set by the pot. The fiirst mic to trigger it's comparitor is recorded in the PLD. All of the other mics are locked out and the winning mic has a lit LED.

Move the mics around as you gain information.

 

[sophiecentaur]

DSP is so achievable nowadays that I would really go for the time delay analysis method. You could probably buy a simple three+ channel digital recording system that would produce three (or more) data streams which you could analyse on your PC. Whatever the waveform of your short burst of sound, it would be easy (well not too hard) to correlate the (possibly) three signals and to calculate the three differences in arrival times. Then it's just a matter of solving some simultaneous equations to give you direction (and range, I think, in one plane). The good thing about using timing rather than amplitude is that the method is not affected by the polar response of the microphones.
You could even look up how the Decca navigator system worked - that's the same principle but the other way around. It's based on the fact that the locus for a given time difference is a hyperbola.