Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

I need a circuit to detect and filter 10 KHZ sound

Status
Not open for further replies.

Navid

New Member
Hi friends
I need a way to detect 10 KHZ sound, I use simple electret microphone with pair of 2N2222 transistors as aplifier, the problem is with other frequency noises on it, i want to filter other sounds below 8 khz and upper than 11 khz, can someone suggest me a filter or circuit?
I use electret microphone because of it's low price, can i use anything instead of microphone?
Regards
 
I don't know what detect means. Do you want to filter out other noise, leaving just the 10khz? OR Do you want a on/off output?
The LM567 is a tone detector that will be of interest to you. With it you should be able to set frequency and width and all is well. BUT Many people have struggled with it. If the noise is too loud it will have problems.
 
UPDATE 2016_07_02. The bandpass filter described below does not have sufficient out of band rejection to meet the OP's requirements.

Hi friends
I need a way to detect 10 KHZ sound, I use simple electret microphone with pair of 2N2222 transistors as aplifier, the problem is with other frequency noises on it, i want to filter other sounds below 8 khz and upper than 11 khz, can someone suggest me a filter or circuit?
I use electret microphone because of it's low price, can i use anything instead of microphone?
Regards

Hy Navid,

Welcome to ETO- where are you? Care to tell us a little about yourself- your age, the electronic tools and test equipment you have, your level of expertise in electronics?

As the other members have said, you haven't stated your requirement in sufficient depth for us to do a definitive design, but I have made a few assumptions and posted the circuit below for discussion. The circuit is a band pass filter, which will pass 10KHz but attenuate higher and lower frequencies. Thanks to Alexander C. Frank, aka อ.ช้างเผือก (Ref 1) for the use of his excellent bandpass filter calculator.

If you need a visual indication that 10KHz is present at the output of the opamp (U1), like a meter or a LED, we can do that.

2016_06_25_iss01_ETO_MFB_BABDPASS_FILTER.png
COMPONENTS
(1) C1, C2: 2n2 (2.2 nano Farads)
(2) R1: 361K72 (361.72 Kilo Ohms)
(3) R2: 723K43 (723.43 Kilo Ohms)
(4) R3: 72R4 (72.4 Ohms)
(5) U1: Texas Instruments OPA192 operational amplifier (opamp)
(6) Power lines: +-15V

DATA SHEETS
(1) OPA192: https://www.ti.com/lit/ds/symlink/opa192.pdf

BAND PASS FILTER CHARACTERISTICS
(1) Center Frequency (Fc): 10KHz
(2) Bandwidth (BW) @ -3dB: 200Hz
(3) Quality Factor (Q): 50
(4) Gain at Fc: 1

REFERENCES
(1) https://www.changpuak.ch/electronics/Narrow_Bandpass_2.php
(2) **broken link removed** (paragraph 16.5)
 
Last edited:
Yes the NE567, Touch Tone Decoder is probably your Best solution.
I have used it in Many applications.
 
What produces the 10kHz sound and why do you want to detect it?
 
Here is how I would do it:

First is a frequency response of the 2-pole Butterworth Filter showing a gain of 10, centered on 10kHz.
375f.png


Second is a time domain sim showing the decoder rise time and decay time.
375t.png
 
Your filter is not a second-order Butterworth, it is Multiple Feedback with a Q much higher than a Butterworth. Its Q is probably not high enough because it has plenty of output (-20dB) between 2.2kHz and 40kHz.
Its output between 8kHz and about 13kHz is almost the same (it is only -3dB) as the output peak at 10kHz.
 
Your filter is not a second-order Butterworth, ...

I beg to differ. It is a 2nd order Butterworth with a gain of 10. I specifically choose the -3db points to be 8kHz and 12kHz. I plotted V(a) in volts; not in db...

It uses the MFB topology as this is required if you want a gain higher than 1. (instead of Sallen-Key). I choose a gain of ten to overcome the loss in the detector diodes. Note that the input is 250mV. If the TS needs less gain, I can either redesign the filter, or he can put a pot on its input...
 
With the Values on your Schematic, My Filter Program Shows it as a (Multi Feedback Filter" and Graphs it this way.
im_grph.png
 
If you define your frequency tolerance of 10kHz, this would define the maximum Q of a simple 2nd order BPF toplogy shown. If you want a variable Q variable F and fixed gain and variable or fixed threshold, you better define it first, before we get another batch of almost solutions.

Goods specs is crucial to any design. Include everything relevant, such as say data rate and group delay distortion.

40 yrs ago my grad thesis was the above filter for middle C D E F so a quadriplegic who could whistle a tune, could whistle 2 successive notes and t0ggle a light switch or dial a phone with 16 switches for any purpose.

The variable Q with constant gain used a pot in the feedback loop in a topology where one R could tune the Q while another could tune f but the threshold was fixed and adjacent note rejection could be 30 dB as I recall so the Q could be between 20 and 100.mThe higher the Q the smaller your f error tolerance will be and the more stable the components must be.

... so you must give better specs for passband gain, BW, band reject levels , threshold and duration levels for triggering etc. You may also want to specify error rate for true and false detection rates as anything can go wrong, it will. This will determine your minimum SNR of the signal & Note, that a loud snap is broad band noise may trigger it.
 
I showed my teacher my project that used "voice recognition". I said OPEN and the door opened. I said CLOSE and the door closed. Then somebody clapped twice and the door opened and clapped one time and the door closed. My circuit cheated.
In this thread we are still guessing about what produces the 10kHz. Nobody can sing or whistle that high.
 
My computer uses the program called Audacity to make tones. I use it and my hearing as an RPM meter for my model airplane motors.
 
I have faced this problem before.
Harmonics and other frequencies cause aliasing and false output.
The LM567 was not able to do what I wanted to.

In the end, the solution was a switched capacitor filter from Maxim. (MAX267 I think it was)
 
Navid ? Please help answer questions about accuracy of frequency and noise interference levels
 
I have faced this problem before.
Harmonics and other frequencies cause aliasing and false output.
The LM567 was not able to do what I wanted to.

In the end, the solution was a switched capacitor filter from Maxim. (MAX267 I think it was)

The LM567 should work, If you put an input resistor and 2 diodes (one Forward and one Reverse Biased) across the input to Limit the input voltage to .6v or lower.
 
Continuing further from chems input, the '567's bandwidth isnt constant unless the i/p signal is over 200mV, so it can be beneficial to ensure the i/p voltage is allways above this.
This can be done with an op amp with enough gain on the mic i/p, a crackly old 741 ought to do the job.

Edit: Actually a 741 might not be a good choice, the Gbp and slew rate of a 741 might not be good enough, esp in a single stage, a ne5532 might be better.
 
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top