Sound robot using only amplifiers....

[Stoudemire]

:idea:Hello all,

I want to all to share your electronics knowledge and hints on how to tackle my robot project.
My group members and I have done some brain storming but we would like to here some ideas from the best.
We got to build a robot that follows sounds literally without using any code only amplifiers, capacitors and other chips for comparison.

The robot will move toward the 3KHz sound source(some 2nd order band pass filter used here)

I think its best if I use 3 microphones:
A B C (they form an equilateral triangle and each sensor captures 120 degrees of noise.
We get the comparison of AB AC AD then we see how motor 1 and motor 2 move and get the logic equation of the motors.

I want to ask you if there is a better way to do this? Any ideas that might help make the circuitry easier? Any1 knows how to get the phase change just from chips so I can use it also? Hints? Drawings? Remarks?

Thank you
Best

 

[balucyber124]

hi,

i would suggest you use op amps for comparing the signals of what range you want with a timer set up at that frequency this should be modified to your input value approx depending on the requirement. then use logic gates connected with H-BRIDGE to run the motors its just simple as it looks like i can say hardly a day job.

best regards,
bala krishna

 

[chuckey]

I only need two ears to be able to turn towards a sound. If you try and compare individual cycles of your 3 KHZ. Then if both ears are hearing the same phase signal then they are pointing towards the source (or away from it!!. If however, the ears are pointing slightly of, then the phase difference changes depending on the trigonometry of the distance between the ears and the path length to the source. So it helps if the ears are as far apart as possible (more phase shift). have you done the calculations to see what the maximum distance is before you get errors? I think you will need need ALC on your amplifiers to keep the output relatively constant over a variable signal strength. To compare phase digitally an EXOR gate is used with its output smoothed, this produces DC directly proportional to the phase error, trouble is I can't think of an easy analogue equivalent!!:-(
Frank

 

[Stoudemire]

I only need two ears to be able to turn towards a sound. If you try and compare individual cycles of your 3 KHZ. Then if both ears are hearing the same phase signal then they are pointing towards the source (or away from it!!. If however, the ears are pointing slightly of, then the phase difference changes depending on the trigonometry of the distance between the ears and the path length to the source. So it helps if the ears are as far apart as possible (more phase shift). have you done the calculations to see what the maximum distance is before you get errors? I think you will need need ALC on your amplifiers to keep the output relatively constant over a variable signal strength. To compare phase digitally an EXOR gate is used with its output smoothed, this produces DC directly proportional to the phase error, trouble is I can't think of an easy analogue equivalent!!:-(
Frank

Hey chuckey;
Im trying to mimic this video https://www.youtube.com/watch?v=OKhOtbA0b1I
What do you mean by an ALC? And to the last statement I didnt get how I use the EXOR! To get the phase shift?
Do you have any handouts that might make things clearer?
best

 

[BradtheRad]

I suppose you plan to place your mics a few inches apart at least. The further the distance, the longer the interval between a sound impulse reaching each mic, and the easier to detect which mic is getting the first sound wave.

However 3khz has a wavelength of 4.4 inches. The phase difference could be off by one cycle or more.

I would rather send audio clicks to the robot.

The strategy that comes to my mind is to detect which mic gets the first sound wave. Then which mic gets the second wave, etc. Turn the robot in the proper direction.

To detect the first sound wave, there are circuits with names like 'quiz show lockout'. It detects which contestant was first to press his button, then immediately locks out the other contestants no matter how quick they are. The circuit sends current to the appropriate LED. It will hold that state until you reset the circuit.

 

[crutschow]

The human ear also detects direction by sound level. The ear closer to the sound obviously hears a higher intensity. You might be able to use that by placing a mic on each side of the robot (as far apart as possible) with possibly some sound-absorbing foam between the two mics or on the side of each mic. Then you just compare sound intensities.

You might also need a third mic in the front to differentiate front from back.

 

[Stoudemire]

I suppose you plan to place your mics a few inches apart at least. The further the distance, the longer the interval between a sound impulse reaching each mic, and the easier to detect which mic is getting the first sound wave.

However 3khz has a wavelength of 4.4 inches. The phase difference could be off by one cycle or more.

I would rather send audio clicks to the robot.

The strategy that comes to my mind is to detect which mic gets the first sound wave. Then which mic gets the second wave, etc. Turn the robot in the proper direction.

To detect the first sound wave, there are circuits with names like 'quiz show lockout'. It detects which contestant was first to press his button, then immediately locks out the other contestants no matter how quick they are. The circuit sends current to the appropriate LED. It will hold that state until you reset the circuit.

Do you have any quiz show lockout chip in mind? Do you think I can use the "555"* chip with it to do a proper circuit.
If your idea works then your a genuis! (Y)
*I frankly don't know the name of this chip, its something that turns on off, like a pwm

- - - Updated - - -

The human ear also detects direction by sound level. The ear closer to the sound obviously hears a higher intensity. You might be able to use that by placing a mic on each side of the robot (as far apart as possible) with possibly some sound-absorbing foam between the two mics or on the side of each mic. Then you just compare sound intensities.

You might also need a third mic in the front to differentiate front from back.

That what we thought to do but I want to see if there are easier ideas

 

[crutschow]

............................

That what we thought to do but I want to see if there are easier ideas

As an alternate, here is an analog phase detector description. The second equation in the paragraph, which is for the output of a single analog multiplier, should also be relatively easy to implement. You filter the output to leave the DC term, which is the phase difference and phase direction value.

Analog Devices makes many different types of analog multipliers that should work for this. You would still need the third microphone in front to help differentiate front from back.

 

[BradtheRad]

Do you have any quiz show lockout chip in mind? Do you think I can use the "555"* chip with it to do a proper circuit.
If your idea works then your a genuis! (Y)

I'm just good at remembering catchy circuit names after seeing them a dozen times.

The circuits I've seen are constructed from logic gates. The following links might be adapted for your purpose.

'Pinewood Derby Finish Line Lamps'
and
'Game Show Indicator Lights (Who's First)'

http://www.bowdenshobbycircuits.info/page7.htm#game1.gif

---------

'Digital Game Show Timer'

http://www.techlib.com/electronics/games.html

 

[chuckey]

Automatic Level Control. The levels of the signal received by each ear will be changed more by the echos/multiple paths in the room then on their difference in path length to the sound source when not pointing in the correct direction. If you work only on the phase error, then a constant level into the phase detector will make it more accurate.
EXOR, exclusive OR gate, see post with link to Wiki.
Frank

 

[crutschow]

Automatic Level Control. The levels of the signal received by each ear will be changed more by the echos/multiple paths in the room then on their difference in path length to the sound source when not pointing in the correct direction. If you work only on the phase error, then a constant level into the phase detector will make it more accurate.
EXOR, exclusive OR gate, see post with link to Wiki.

I have doubts that a digital EXOR gate will work well with analog signals.

 

[Stoudemire]

Can you please explain how to use the EXOR and its relation with phase,
can u explain how I connect it with the microphones?
I really dont have a clue

 

[chuckey]

EXOR = Exclusive OR gate - gives a 1 when inputs are different, so amplify you 3 KHZ A & B with an open loop op amp, you will have two square waves now. Trying to do a crude wave form now:-
A 111000111000
B 111000111000
O/P 000000000000 both inputs the same
A 111000111000
B 011100011100
O/P 100100100100 Vout = 1/3 VCcc
A 111000111000
B 001110001110
O/P 110110110110 Vout = 2/3 Vcc
A 111000111000
B 000111000111
O/P 111111111111 Vout = Vcc
So as the phase changes between the two waveforms the mean DC changes, but you must smooth out the pulses.
Frank

 

[Stoudemire]

EXOR = Exclusive OR gate - gives a 1 when inputs are different, so amplify you 3 KHZ A & B with an open loop op amp, you will have two square waves now. Trying to do a crude wave form now:-
A 111000111000
B 111000111000
O/P 000000000000 both inputs the same
A 111000111000
B 011100011100
O/P 100100100100 Vout = 1/3 VCcc
A 111000111000
B 001110001110
O/P 110110110110 Vout = 2/3 Vcc
A 111000111000
B 000111000111
O/P 111111111111 Vout = Vcc
So as the phase changes between the two waveforms the mean DC changes, but you must smooth out the pulses.
Frank

Frank, I bought an exor opp amp .
*I want to ask you about how to get the boolean expression of A and B? Check the pic of the circuit that I have done in my lab;
https://i50.tinypic.com/ivbtxu.jpg
*I have done only a 1st order low pass filter for now; how can I change the circuit that I did to your explanation? A compare Opp amp? How does the phase tell me what is the location of the sound? How can I get the Vout out of the boolean expression of the O/P?

ALSO


*How can I improve the circuit above? https://i50.tinypic.com/ivbtxu.jpg The microphone is getting a voltage about 1.4V but Im afraid of the size of the gain that I use.
because I dont want to reach the Vout beyond VSS. I think we got a 60milli volt peak to peak when sound is sent to the microphone.

*Do you think I should also do a DC offset?
This is the mobile car that we will try to work on https://i50.tinypic.com/5tt6p5.jpg

P.S : I couldn't find the (Who's First) opp amp that some has described, it would be easier to find if an actual name or number.

 

[BradtheRad]

P.S : I couldn't find the (Who's First) opp amp that some has described, it would be easier to find if an actual name or number.

The 'who's first' circuit has to be custom assembled from a few IC's. I have never seen anything called a 'who's first' IC.

And now I just tried the Bowdens hobby circuits link and find it doesn't work. I'm sorry the link was wrong.

http://www.bowdenshobbycircuits.info/page7.htm

For example this is the schematic 'Pinewood Derby Finish Line Lamps' at the website above.



And the other one is 'digital game show timer' from the techlib website. (I hope I'm not infringing on copyright by posting these, considering they are for public observation at the websites anyway.)

 

[chuckey]

AND gate gives a 1 out when all inputs a ! +
A 0 0 1 1
B 0 1 0 1
C 0 0 0 1 O/P
The more narrow the bandwidth of the amplifier, the less the circuit will react to non 3KHZ pick up and the robot will work better. Also it will react less to self generated noise. If you are going to use a digital gate to do your phase comparison then it is good that the square wave goes from E to Vcc, I would run the op amp on open loop with just a tiny bit of bias on one input simply so you know that in the absence of the 3 KHZ, the output will be at earth potential. Else if one op amp sits with a earth, the other at Vcc, the phase detector will see a huge signal, causing the robot to turn at its maximum rate.
Frank

 

[Stoudemire]

AND gate gives a 1 out when all inputs a ! +
A 0 0 1 1
B 0 1 0 1
C 0 0 0 1 O/P
The more narrow the bandwidth of the amplifier, the less the circuit will react to non 3KHZ pick up and the robot will work better. Also it will react less to self generated noise. If you are going to use a digital gate to do your phase comparison then it is good that the square wave goes from E to Vcc, I would run the op amp on open loop with just a tiny bit of bias on one input simply so you know that in the absence of the 3 KHZ, the output will be at earth potential. Else if one op amp sits with a earth, the other at Vcc, the phase detector will see a huge signal, causing the robot to turn at its maximum rate.
Frank

BradtheRad, Ill see if I can implement this circuit today in the lab

Frank what do you mean by: AND gate gives true when all inputs are not positive?
Today, Ill try lower the bandwidth of the circuit.
How do I use the digital gates? I am pretty sure I get analog signal from point A in https://i50.tinypic.com/ivbtxu.jpg
How do I convert it? Schimdt trigger or is there something better?

In the last sentence you told me to use an open loop to have an output between E and Vcc, but I don't how to bias an open loop op amp? Can you a simple drawing to explain your concept. What do I connect to the negative terminal and to the positive terminal? Should I use an instrumentational amplifier/differential one? Is that your point if yes how much biasing? 1V?

You also explained in post #13 the exor; but I havent yet understood what I should do to it?

 

[chuckey]

Don't know what happened there!, finger trouble I think. it should be when all inputs are a 1.
To run the op amp on open loop, remove the 1K5 resistor, for stability it would be best to return the -ve input to the earth rail via a high value resistor, say 1M. If you want to experiment with a bit of off set to hide the noise, return the +ve input to a very small voltage - say 1 mV. If leaving the 100 ohms in, connect the +ve input to the +V line via 470K resistor. ( I 100 = 1mV/100 = 10 microamp, to drop 5v @ 10Microamp, R = 5X10^6/10= .5M ohm), this will keep the op amp output at Vcc in the absence of an input.
Frank

 

[BradtheRad]

Indications are that you're on the right track, so I don't mean to distract...

I've been reading the article linked below, on how to measure phase differences between two signals. I don't know that it can be adapted to 3 signals but it has schematics that might provide ideas.

'Phase Correlation Meter'

**broken link removed**

 

[chuckey]

The last circuit is exactly what I was thinking off, Brad !
Frank