Microphone based alarm

[piezomic]

This might seem counter intuitive but I need to make a circuit that sets off an alarm when a certain frequency or noise is not present anymore. Basically it will be required to monitor a certain sound and if that sound stops for a certain time period (about 3-5 minutes) that will cause an alarm to sound to alert to this fact. It is for a medically related research project if anyone is interested.

The problem is that I have very limited understanding of electronics outside of some highschool classes and self interest. So I would not know where to start when making an alarm circuit like this. Note that I have already got a microphone circuit and am planning to use a piezo speaker rigged as a mic to create a contact microphone.

Any help would be hugely appreciated, even simply pointing me in the right direction in terms of sites I can visit or schematics I should refer too would be fantastic.

 

[ericgibbs]

hi,
It is possible to detect a break in a continuous frequency.

More information about the signal would help.
Frequency, amplitude etc.
Are there other sounds the microphone may sense, ie: background noise

E.

 

[alec_t]

I think detecting the 'certain sound' will be the hardest bit. As Eric says, can you give us more info?
If we can come up with a circuit design would you be able to build it?

 

[piezomic]

Sorry about the delay in replying.

I'm still running samples for the frequencies as for the sound (this is a different application for the chip and thus I can approach this separately) the easiest way I can describe it would be to reference a water pipe: it would be like placing the contact mic on the drain and it being able to discern when someone pulled the plug on the sink as apposed to the plug still being in by analysing the intensity of the sound coming through the pipe. Obviously there would be background noise, as there would be in patients but that would only reflect a low voltage going through the mic while the plug being pulled and water coming down the pipe would result in high voltages.

I basically just need something to be able to discern this high voltage situation and then trip an alarm (preferably after a timer of a few minutes but this can be modded later)

Sorry if this analogy seems confusing but it's the closest I can come to the situation I will be dealing with

 

[WTP Pepper]

Microphone > amplifier > High Q band pass switched capacitor filter > level detector > alarm

https://www.electro-tech-online.com/custompdfs/2012/06/107-MF10.pdf

 

[piezomic]

Thanks WTP Pepper, I've had a read into that and understand that the capacitor filter is used to filter out a select number of frequencies but since I have no knowledge of how to integrate them into a circuit would it still work if it was left out and simply the level detector was used? Or is the filter needed to clean the signal so that the level detector can actually differentiate accurately?

Sorry if this is a very noobish response

 

[alec_t]

A filter is almost certainly necessary, otherwise the circuit will give false alarms for any noise source (passing car, cough, phone ringing ....) loud enough to trip it .

 

[WTP Pepper]

What alec_t says is correct. To me, your idea is sound in theory but you do need a little more experience in the building blocks of a system. That takes experience and alot of bookwork.

Start by looking how stable your signal is you are trying to detect. Design a band pass filter to cope with any drift. Determine after the filter what threshold voltage would be a positive trigger, design a comparator circuit to trip just above that level.

That's about as much detail I can give without a one on one session....There are a lot of variables only you can determine and measure.

 

[colin55]

Just use an 8 pin microcontroller to do all the work. In the following project I used a PIC12F629 to detect a whistle. **broken link removed**

 

[DerStrom8]

Just use an 8 pin microcontroller to do all the work. In the following project I used a PIC12F629 to detect a whistle. **broken link removed**

That circuit is rather useless without the code. Also, it can't be very precise. It'll most likely be set off by a wide range of noises. And if you try telling me it's tuned to the frequency of a whistle, that's also a load of bull. Everyone whistles at a different frequency, and a lot of other variables come into play. I'd like to see the code to see how you expect this thing to work

 

[RCinFLA]

Look up NE567.



**broken link removed**

 

[WTP Pepper]

It would work probably better than most of the duff key finders produced in the early 90's with the right code. They would respond to a TV sound IIRC.

Just a simple prog to detect sharp changes in input frequency using timer interrupts.

Not something I would use on a critical system like whistle to drop the Nuke