help please..build a robot that can detect sound

[A-Saeed]

Hey,
i'm trying to build a robot that can detect sound and move toward it, but all sound sensors that i've found has a limited detecting range by around 3 meter away..
the problem that i want to detect a single tone sound from about 12 meters away from my kit.
Any opinions?

 

[alec_t]

A 'rifle mic' or 'shotgun mic' should be able to pick up sound directionally at 12 m. You would also need a tuned amp to filter the mic signal.

 

[A-Saeed]

thanks for replying,
is there any other solution?
or any smaller mics?
i'm using a RF link with my robot "at where the Tx" and the Rx is attached to an item. when the item goes far away "about 12 meters" from the robot the communication is lost and i use this as indication to trigger an alarm buzzer circuit.. i want my robot to detect this single tone buzzer so i would be able to direct it toward it.

 

[alec_t]

Smaller mics are less directional so may may pick up other sound sources.
You could do a test using a headset mic, your PC sound-card, and audio software such as Audacity.

 

[tomizett]

You can also get good directivity using a parabolic reflector dish with a microphone - I have a vague recollection of seeing kits for such things for sale at some electronics supplier (Rapid maybe?).
Would it be an option to spin such a dish on the top of your robot like a radar? Or maybe to rotate the entire robot to home in?
Recognising the sound you're after could be tough though, do you have access to DSP for processing your audio?

 

[cowboybob]

Welcome, A-saeed!

(post is a little old, but...)

You might consider a simple mike to LM567 circuit configured to detect the frequency(s) of the "buzzer circuit" you're using and use that negative going output pulse as a trigger for the robot's "seek" function. The LM567 will respond to very low input levels (<200mv).

I can provide a schematic, if you like.

 

[A-Saeed]

Welcome, A-saeed!

(post is a little old, but...)

You might consider a simple mike to LM567 circuit configured to detect the frequency(s) of the "buzzer circuit" you're using and use that negative going output pulse as a trigger for the robot's "seek" function. The LM567 will respond to very low input levels (<200mv).

I can provide a schematic, if you like.

Thank you!
This would be so helpful..

 

[cowboybob]

Here's a configuration I've used (Vcc of regulated 6VDC):

The datasheet is at https://html.alldatasheet.com/html-pdf/8984/NSC/LM567/236/6/LM567.html see page 4.

There is NO Spice model for this chip.

 

[4pyros]

Here's a configuration I've used (Vcc of regulated 6VDC):

The datasheet is at https://html.alldatasheet.com/html-pdf/8984/NSC/LM567/236/6/LM567.html see page 4.

There is NO Spice model for this chip.

Is the output of the chip on/off only?

 

[cowboybob]

Is the output of the chip on/off only?

Yes. Just a negative (0 VDC) trigger (from Vcc) that has a pulse width equal to the duration of the signal that matches the center freq of the VCO of the LM567 (within the bandwidth limits).

 

[large_ghostman]

Can i ask a stupid question? In this situation why wouldn't you be tempted to use a band pass filter, tuned to a narrow band, so only the homing signal is passed. Ok other things might have same frequency but wouldn't the homing sound swamp them out at that distance?

 

[4pyros]

Can i ask a stupid question? In this situation why wouldn't you be tempted to use a band pass filter, tuned to a narrow band, so only the homing signal is passed. Ok other things might have same frequency but wouldn't the homing sound swamp them out at that distance?

Smart thinking! but I think its kind of built in to the chip.

 

[cowboybob]

Can i ask a stupid question? In this situation why wouldn't you be tempted to use a band pass filter, tuned to a narrow band, so only the homing signal is passed. Ok other things might have same frequency but wouldn't the homing sound swamp them out at that distance?

Once the homing signal (assuming its not just a continuous blare) gets through the bandpass filter (BPF) you'll still need to turn that signal into a trigger of some sort.

As 4pyros points out, both functions (detection [plus variable bandwidth selection] and conversion to a usable trigger) are accomplished with the one chip.

I had initially tried a BPF method and the first problem I ran into was what sort of filter to use to get the "tightness" of the filtering I needed (1st order, 2nd order, 3rd order, etc., each successively more complex) and within an audio situation with lots of noise and harmonics.

And then, of course, I then needed a circuit to act as a trigger (simple rectifier, comparator, wharever).

I had not heard of the LM567 but a search of ETO produced the idea. I bought 10 of them (remarkably cheap) and haven't looked back.

 

[large_ghostman]

I will read the datasheet, I have never heard of the chip but sure sounds cool! So i guess in a way it is a bandpass filter but will all the ad dons, Time to go look on ebay

 

[ronsimpson]

The LM567 works but not well. But it is simple.
Band pass filter: Any time you can reduce the bandwidth you will decrease the noise.

1)Band width filter: Put you mic in a 1 foot long tube (card board).
1a) It will become directional. Using a servo motor, spin the mic+tube around and see what direction is loudest.
1b) A one foot long tube should resonate at (I don't remember 1khz or 500hz some thing like that). If you get the tube to resonate at the transmitting frequency you will reduce the noise and increase the signal.

2)Phase angle: Use two mics separated by 1/2 wave length. Add the two signals together. (again spin the mics around by a servo motor) When the mics are facing the sound the two signals will add up. At 90 degrees they will subtract to near zero. I think the mics can be wired in series using one amplifier. This should give you a very directional 'antenna'.

 

[cowboybob]

The LM567 works but not well...

Well, depends, I guess.

Worked very well for me, and good old Ma Bell as well.

And I can confirm (due to numerous Vcc and input signal level overshoot mistakes ) that it is remarkably bullet proof...

 

[ronsimpson]

The goal is to find the sound. Go to the sound. The LM567 will detect if there is that sound but not how loud or where it is located.

 

[jpanhalt]

Once you have the LM567 detecting the sound, you can localize it using two detectors. Here is a very old DPRG post that does that with infrared.

**broken link removed**

I believe there are more posts on that site along a similar vane.

John

 

[4pyros]

Thats what I was going to say. You need two ears.

 

[cowboybob]

Merry Christmas.

If this were a Lego Robotics device, I'd configure it as follows:

1. LM567 circuit's output into a touch switch input for the controller, forcing a "Stop" gosub routine.
2. "Stop" Gosub loop that was preprogrammed to force the robot to then travel at a pre-determined compass heading (Home).
2. Use distance measuring device in same loop to detect a flat board (at the final "Stop" position) to force a "quit" or "hold" commancd.