Small cheap simple microphone needed.

[cupoftea]

Hi,
We have an electrical product which when it breaks it gives out a constant tone at approximately 1kHz to 4kHz.
So we want to detect this with a microphone and have a circuit to take the output of the microphone and
signal to us that the sound is being emitted.
So we need a really cheap microphone that consumes very little bias current.

Cheapest I can find is this piezo microphone...

https://www.farnell.com/datasheets/2877014.pdf?_gl=1*wd1u56*_gcl_au*MTI5OTAzODk1Mi4xNzQxMDMzNjAz

Do you know of any cheaper?

 

[D.A.(Tony)Stewart]

your phone is free

 

[cupoftea]

Thanks but we need to permanently install this in a box near the instrument that may fail. -So that it can "Listen" for the 1khz to 4khz tone.
BTW the 1khz to 4khz tone is actually loud...but we need to detect it because often the kit kit will fail when no-one is nearby....hence we need the cheap microphone to sense the sound.

 

[D.A.(Tony)Stewart]

These cost pennies to make so the cost to you depends on your buying power/skills.

I can get for $0.18c with a wired plug JST 2.0, but will sell for $0.50. 1k MOQ

 

[betwixt]

Consider using a piezo disc, it is probably what makes the sound in the first instance. They work quite well as narrow bandwidth microphones and will produce many mV into a high impedance. They are cheap and often sold as drum beat sensors as they can be glued to the sides of drums an the output used as a trigger for other music instruments or as a pad to be directly hit by a drum stick. Cost (imported from you know where) is about UK 20p each.

Brian.

 

[Easy peasy]

https://www.aliexpress.com/item/1005005849864178.html?src=google&src=google&albch=shopping&acnt=298-731-3000&isdl=y&slnk=&plac=&mtctp=&albbt=Google_7_shopping&gclsrc=aw.ds&albagn=888888&ds_e_adid=&ds_e_matchtype=&ds_e_device=c&ds_e_network=x&ds_e_product_group_id=&ds_e_product_id=en1005005849864178&ds_e_product_merchant_id=767508601&ds_e_product_country=NZ&ds_e_product_language=en&ds_e_product_channel=online&ds_e_product_store_id=&ds_url_v=2&albcp=21627925371&albag=&isSmbAutoCall=false&needSmbHouyi=false&gad_source=1#nav-specification

 

[cupoftea]

Thanks and i assume that Piezo disc microphones, and electret microphones give out a very small signal, so i am going to use possibly a diff amp to amplify it. This is just to detect if the noise is there or not, then "inform" the microcontroller.....so any old 'low as possible' component count opamp circuit will do.
Ill just adjust the gain so that only the faulty machine's noise is detectable to the micro that will be reading this (and not the background noise) ....i won't even bother with AC coupling of the signal....just a single supply opamp and shovel the piezo-disc or electret microphone into it......ill probably end up with a rail to rail oscillation at the output of the opamp but thats fine...i can convert that to a form that the micro can read.

If the opamp is giving rail to rail oscillation when the faulty machine is buzzing, and is lodged at a rail for when the machine is not buzzing, then thats fine.
Crude-as-you-like, cheap-as-it-comes opamp circuit coming up!........with a cheap-as-chips opamp too!

No need for an InAmp here!.....just want to know..."is the machine making the fault noise or is it quiet?"

 

[betwixt]

Both electret and piezo discs can produce over 100mV easily so you can almost feed them to a comparator directly. I very strongly advise you add some frequency filtering though or it will detect any ambient sounds too. The advantage of piezo over electret is they are already resonant in the frequency range you are interested in and they need no power source so its easier to mount them remote from your electronics. It is quite likely the alarm sound itself comes from a piezo device.

I have seen applications using piezo discs to detect broken glass if windows are broken to gain entry to buildings because the sound of falling glass falls largely in the piezo bandwidth.

Brian.

 

[cupoftea]

Thanks i see what you mean, it would be handy to have the frequency filtering in there...but to be honest i suspect we may be able to get away without that, as usually the place is quiet...(well, no continuously loud noise, just the occasional loud bang)......and to filter down to 1kHz to 4kHz is going to need a relatively big filter, so i am not so sure if our cheap-as-chips mindset can extend to that. Thought to use a cheap opamp as will need amplification.

--- Updated Yesterday at 12:38 PM ---

Hi,
Please advise if you know of a cheaper/simple way to detect the continuous
tone at 1khz to 4khz which comes out of the machine continuously
when its failed? The three option circuits attached (A,B,C) seem to do the job. (LTspice and .PNG attached)
However, as betwixt kindly says, it would be nice to get some bit of frequency
discrimination in there aswell if possible.
I used a piezo disc microphone for cheapness. (thanks to betwixt for putting it forward)
Please entertain the most crude-as-you-like circuits of simplicity as yet untold,
for us, where "yeah, does the job" is far more important than "quality way of doing it".

 

[D.A.(Tony)Stewart]

@Easypeasy found 2x more expensive ones but also without wired connector for a 2 pin header.

@cupoftea you said "i won't even bother with AC coupling of the signal.." is a big misteak Since you have no idea exactly what IDSS is on the FET, nor the DC output bias voltage, any substantial DC gain will saturate the Op Amp to one DC rail or the other.

A more prudent design choice is to provide 20 to 26 dB gain between 1k and 4k with a 4th order BPF then detect peak amplitudes sustained for 1 second above 1Vpk.
This can all be done using a single quad OA with a logic level out. The detector ought to be an AC coupled diode-detected integrator with a suitable Vref.

i won't even bother with AC coupling of the signal..
No. It'll only be reliable to condition the signal to avoid false positives and true negatives. (Although it might be possible to avoid the fault in the 1st place. It depends on the cost of failure)

Use a (Vcc-1V)/2 for Vin+ ref on each stage for use on a single supply such as +5V on an LM324B.


The LM324B quad OA is cheap and has PNP inputs thus the Vin+ optimum is less than Vcc/2 and price is $0.053usd | 1u
The LM1369 quad OA is not cheap but is R2R.
e.g. Mouser non-stocked

Quantity: 2,500 MOQ​

Unit Price $11.17
Ext. Price $27,925.00

 

[betwixt]

.. and none of the three schematics would work anyway because piezo discs don't conduct DC so the op-amp inputs would 'float'. You would have to add some voltage to the non-inverting inputs.

If you are sure the background would be quiet and can tolerate false alarms if it isn't, even momentarily, use a piezo (or electret) followed by a high gain op-amp stage, it doesn't matter if the output clips. Follow that with your rectifier or a triggerable timer.

If you know the frequency you are detecting or are willing to 'tune' the detector with a trimpot, you could consider a PLL like the NE567 which could be driven directly from the pick-up and give you a simple high or low output logic signal. It would do the amplifying and filtering for you in just a few components.


Brian.

 

[D.A.(Tony)Stewart]

piezo discs don't conduct DC so the op-amp inputs would 'float'. Brian.

The FET conducts DC and its gain is proportional to the DC voltage drop.

The BP filter that I made are impedance scalable to 10xR and C/10 and AC coupled so the FET pullup you choose is smaller than the load, and the 5 uF is not required.
. Any questions? BOM cost 25 cents, incl mic and logic level out. MOQ + shipping & tax

 

[betwixt]

Very true Tony but the comment was about the piezo discs shown in the schematics, they are more like capacitors and have no FET in them.

Brian.

 

[cupoftea]

.. and none of the three schematics would work anyway because piezo discs don't conduct DC so the op-amp inputs would 'float'. You would have to add some voltage to the non-inverting inputs.

Thanks very much Brian....and may i please ask do you think that the attached three (adjusted) solutions would now be better? Which is best and do you have a cheaper /simpler one?
(LTspice and PNG schems attached )
Thanks to Tony also for the frequency windowed solution , which we will bear in mind

 

[D.A.(Tony)Stewart]

Very true Tony but the comment was about the piezo discs shown in the schematics, they are more like capacitors and have no FET in them.

Brian.

But those are not the cheapest nor best mics with wire & connector... that cost $0.18 /ea

 

[Easy peasy]

10.5 cents each USD is not to bad ( 10 in a bag ), and yes you need a DC blocking cap due to the way the electret works with the internal fet.