Designing a sound detector 2

[Poubella]

Hello all!

I am trying to develop an application for deaf or hearing impaired people. The point is to light a lamp when a specific sound appears (it’s a “beep” that appears when a barrier is about to close down). I measured 3675 Hz for the frequency of this beep.

I cannot act directly on the system which produces this sound, so I have to design an independent system. I guess the best synoptic to do that is:

Microphone – amplifier – frequency detector – light

The first steps are OK for me, but I am stuck with the frequency detector… Using simulation tools I made up a basic bandpass filter centered on 3675Hz, which gives excellent simulation results (see enclosed file), but inevitably when I try to realize it “in real life”, it doesn’t work at all…!

Is that due to the tolerance of the components I used? Or is it simply impossible to realize such a perfect filter IRL?

Or maybe somebody has a better solution to make my sound detector?

Thank you very much for your help!

 

[VEBill]

Google: 567 PLL tone detector

A more modern approach would be to program a cheap, 8-pin uC to perform the same function.

 

[MacGyverS2000]

Sounds a lot like a project for school...

Dan - Owner

http://www.Hi-TecDesigns.com

 

[Skogsgurra]

Careful with your handle! I wouldn't use that in French speaking countries - it mean 'feminine waste bucket'. Not what you want to be called, is it?

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[IRstuff]

"it doesn't work at all...!"

You need to explain HOW it doesn't work? Where are you oscilloscope results?

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[benta]

IFAIK, the NE567 is obsolete since years.

But a PLL is the way to go, it will detect the tone and directly give you a "locked" output signal. CMOS 4046 or whatever...

Benta.

 

[Poubella]

Thanks for answers!

« It doesn’t work » means that all frequencies are attenuated by the filter: when I sweep the input frequency from 1 Hz to 10 kHz, the output remains at zero volts. I wonder if my filter bandwidth is too tight to implement it in reality?

Thanks for the 567 and CMOS 4046 tips, I will look at these options! But still, I would like to know if I could realize this function with a basic RLC filter? (maybe using very precise components?)


Ps: this place is very useful but a bit funny, it looks like some laws guardians are seeking for any school problem... I am sorry if I don't use the correct professional vocabulary, I'm not native English!

 

[MikeHalloran]

Non-English speakers are very welcome here.

Students are welcome, too. ... when they graduate.



Mike Halloran
Pembroke Pines, FL, USA

 

[benta]

Forget that filter and do some real circuit design on paper, your simulation software is misleading you.

Real components have tolerances.
Inductors have series resistance.
Inductors have parallel capacitance (between the windings).
Capacitors have equivalent series resistance.
Capacitors also exhibit inductance.
Resistor have inductance as well.

And so on...

Benta.

 

[Poubella]

Yeah I know simulation tools are perfidious, but they are so easy to use!

A PLL seems to be clever and efficient, I will investigate that way!

Thanks for helping

 

[MiketheEngineer]

Buy a dog??

 

[VEBill]

IFAIK, the NE567 is obsolete since years.

Agreed.

If the NE567 isn't readily available, then use the LM567 (or any other PN variation...).

 

[MacGyverS2000]

Ps: this place is very useful but a bit funny, it looks like some laws guardians are seeking for any school problem... I am sorry if I don't use the correct professional vocabulary, I'm not native English!

Your vocabulary is fine, it's the general simplicity of the problem, the apparent lack of understanding of how simulations are only a starting point, how a filter works in the real world, the expectation the frequency is very specific and won't change, etc. that raises flags. You may be a practicing EE, but the problem smells an awful lot like a student project.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[IRstuff]

You need also to make sure that the duration of the signal is longer than the acquisition time of whatever circuit you use.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize