PCG signal display issue

[MD_SHAHRUKH]

Hello everyone,
I have been working on an electronic stethoscope in which I have implemented one piezoelectric senor to hear the heartbeat, I get the heartbeat sound very clear to my earphone, but when acquiring that signal to ADC and try to plot the same, I am facing the issue of not so good signal, that means the plot rarely visualize the lub dub signals.

I have made many changes to the circuit, it enhances my sound signal but it is not at all improving my plot.

Should I get some digital IC into it? or analog circuit with ADC works fine?

 

[andre_luis]

I get the heartbeat sound very clear to my earphone, but when acquiring that signal to ADC and try to plot the same, I am facing the issue of not so good signal

Your headphone is perhaps filtering high frequency noises above the audio range, reason for which heartbeat sounds good and scope seems noisly. Think of making some kind of spectral analysis ( e.g DFT, available on digital osciloscopes ) in order to determine what kind of components are in the noise. In general muscular movements lies in the range of 30-40Hz, so noch filter for this range in addition with 50/60Hz would improve the acquired signal.

 

[c_mitra]

I have made many changes to the circuit, it enhances my sound signal but it is not at all improving my plot.

Of course you have a computer (a laptop or a desktop) and you have a mike input. Piezo transducers are high voltage and you can feed the output of the sensor into the audio in of the sound card and save the audio.

Using some software you can see the voltage waveform. I use audacity. You can use your own.

This signal now need signal processing to be used in a plotter. You can use many different type but the idea is simple: detect abnormal sounds.

Another interesting thing I have noticed that many old doctors cannot hear low frequency properly. Using software the frequency can be multiplied by 2 or 5 times and the low frequency sounds can have much information.

You should work systematically and use tools that are simple and common.

 

[KlausST]

Hi,

There are many issues:
* A linear regulator needs input_capacitor and output_capacitor to work properly. --> Read it´s datasheet

* Never leave any (unused) inputs floating. Connect all TL072 inputs to get a non saturated output.
* A TL072 needs a power supply bypass capacitor.
* Use standard OPAMP symbol
* Use GND symbols.
* Use HPF and LPF to focus on the heartbeat frequency bandwidth.

* Q2 is not properly biased (thus you get distorted output. Better directly use the TL072 output)
* R7 value makes no sense.

***
There are many heartbeat preamplifier circuit examples/descriptions in the internet. Why don´t you read/use them?

Klaus

 

[danadakk]

Unused opamp connections -

https://www.ti.com/lit/an/sboa204a/sboa204a.pdf

Regards, Dana.

 

[Audioguru]

The TL072 has two opamps. Your unused opamp is not disabled then it might be oscillating. Use a TL071 single opamp instead.
The minimum supply for a TL07x is 7V where it barely works. You have only 5V but use 9V instead.
The opamp amplifies noise from the REG1117 because you are missing a filter capacitor from the junction of R3 and R4 to ground.

Why is the opamp an inverting type with an extremely low input resistance of only 470 ohms? Its very low input resistance is killing the output level from the piezo sensor. Make the opamp non-inverting with a 220k input resistance then you do not need such a high amount of amplification.

Why is the transistor with NO base biasing and NO negative feedback? Then its output is extremely distorted and is missing low levels. If the opamp is non-inverting with a high input resistance then the transistor is not needed.

Why are you using a REG1117 to power the opamp? The opamp does not need it. The REG1117 makes noises because you are missing the very important input and output capacitors shown in its datasheet.
What voltage is VDD? It is missing an important capacitor to ground.

Why do you have R5 and R6?

 

[MD_SHAHRUKH]

Extremely sorry for the late reply. I was trying few more changes. The above image is the working circuit for the heartsound. It is working perfectly now. I am getting the graph as well, thank you so much for suggestions, I have implemented proper filters after voltage regulator, for 5V I have used LM7805 and for 12V I have used LM7812. The circuit above is drawn in LTspice and due to symbol for opamp I have used AD8674, but TL072 is used. Due to unavailability of TL071 I was not able to use it in the circuit.

The proble is arrised here!
Now that I am getting heart sound and able to plot it, I missed out the breathing sound and its respective waveform which is also necessary to identify few more diseases.
Can you able to suggest me few more corrections so to get the breating sound as well, or refer me any circuit which will help me doing the same.

Thank you so much. I am attaching the waveorm I have got with this circuit very soon.

--- Updated Apr 29, 2021 ---

I have also changed the sensor.

Here is the datasheet link for the microphone.

Thank you.

 

[KlausST]

Hi,

The given circuit makes no sense.
C4, C2, C3...are very suspect

Let me guess: This was a trial and error simulation approach...without prior calcualtions...without knowing why to chose those values.
Don't get me wrong: it may work, but it's far from being properly designed.
Don't be surprised if you kill the Opamp by the huge capacitive load. Oscillation may occur, overheat.

It's like driving a car with full throttle and slowing it down with continously pressing the brake.

Sorry for my honest words.
There are so many online tutorials, even proven circuits with description and calculation examples...why don't you read/use them?

But for sure you are free to go your own way.

Klaus

 

[MD_SHAHRUKH]

I am really clueless about this approach as you said, but I have a deadline to meet. I have not much time to read or study the calculation. I am doing the 3D modeling for the product enclosure also and simultaneously working on the circuit, I have not enough time to make PCB as well, this is done on vero board itself. This one is a prototype model very much needed to raise funds. I hope you understand the urgency. EDAboard is my final resort because of so many brilliant people to give me suggestions.

I know many of you will not agree on my proceedings, and I also agree that I am not good in circuit analysis, but please help me out. I have to finish this whatever it takes.

--- Updated Apr 29, 2021 ---

Please help me with hyperlinks or pdf so that I can directly visit those pages or pdfs or videos so to make this system working. I have only 2 days left.

 

[KlausST]

Hi,

I usually don´t do others circuit design in the forum....

I did some internet search for heartbeat signals to get a clue about the frequency range.
This absolutely is your first job when doing a design: get the requirements.

Also I don´t know what gain you need. (again your job to get the requirements)
I just guessed a gain of 11 in the frequency range of interest could be suitable.
Gain is set with R1 and R4

I don´t know what information you are interested in or what your application is expected to do. Thus I just tried to reperesent the waveform:
I decided the frequency range of interest to be 1Hz ... 100Hz (while the heartbeat frequency usually is 0.5Hz .. 2Hz I guess you are more interested in it´s overtones.)

Since your mic has an internal series capacitor of 100n ... you need an input resistance of at least 1.5MOhms to get 1Hz lower cutoff. (stage 1)
--> this is made by R5 & R3 which also act as DC bias.

C3 is for DC decoupling. But it´s dominated by the 100n microphone internal cap C5. Don´t use C5 in your circuit.

R2 and C2 are for upper frequency limit ( stage 1)
R1 and C1 are for upper frequency limit (stage 2)

R4 and C4 are for lower frequency limit /stage 2)

C6 is just to protect your earphones from DC.
(Again not knowing any information about your earphones, thus just guessing)

****
All this is far from being exact...

Klaus

 

[MD_SHAHRUKH]

Thank you so much. . I am doing it right now on my breadboard. I don't have OP07, I have 741 and TL072. I will use both of them to get the result. I'll let you know very soon.

Thank you so much. And I promise you I'll not repeat this kind of approach again, I'll work harder from now on.

 

[KlausST]

I'll work harder from now on.

No need to work harder. I´d rather think it´s the other way round.

It´s like a holiday tour.
You surely don´t buy a random bus ticket and a random ferry ticket .. without knowing if you ever need them.
Instead I guess you first find out where you want to go to. Then find out whether you need a ferry at all and need a bus at all. Then select the route, then buy the ticket(s) you need.

Klaus

 

[Audioguru]

Yikes! Your new circuit has many errors:
1) C4 will overload and probably cause oscillation in any opamp because it is an AC short circuit, not an RC filter.
2) The inverting input of your opamp is also a dead short to the microphone signal. The opamp should be non-inverting with a high input resistance to pass low frequencies from the output capacitor in the microphone.
3) The new microphone has an internal preamp that must be powered.
4) An antique 741 opamp will not work with a supply that is only 5V and the minimum supply for a TL072 is 7V.
5) The unused opamp in a TL072 must be properly disabled.
6) You should never bias a transistor like that since the hFE could be high causing saturation or the hFE could be low causing cutoff.
7) The output of the transistor has DC that will fry your earphone. Use a coupling capacitor.

 

[MD_SHAHRUKH]

Hi,

I usually don´t do others circuit design in the forum....

I did some internet search for heartbeat signals to get a clue about the frequency range.
This absolutely is your first job when doing a design: get the requirements.

Also I don´t know what gain you need. (again your job to get the requirements)
I just guessed a gain of 11 in the frequency range of interest could be suitable.
Gain is set with R1 and R4

I don´t know what information you are interested in or what your application is expected to do. Thus I just tried to reperesent the waveform:
I decided the frequency range of interest to be 1Hz ... 100Hz (while the heartbeat frequency usually is 0.5Hz .. 2Hz I guess you are more interested in it´s overtones.)

Since your mic has an internal series capacitor of 100n ... you need an input resistance of at least 1.5MOhms to get 1Hz lower cutoff. (stage 1)
--> this is made by R5 & R3 which also act as DC bias.

C3 is for DC decoupling. But it´s dominated by the 100n microphone internal cap C5. Don´t use C5 in your circuit.

R2 and C2 are for upper frequency limit ( stage 1)
R1 and C1 are for upper frequency limit (stage 2)

R4 and C4 are for lower frequency limit /stage 2)

C6 is just to protect your earphones from DC.
(Again not knowing any information about your earphones, thus just guessing)

View attachment 169140
****
All this is far from being exact...

Klaus

Thank you for the circuit you have shared, It is working perfectly fine for heartbeat sound, but it is not giving the sound of breathing, I am placing a sensor to my neck to hear the sound, a very small sound of breathing is coming. Unfortunately, I have few components with me which not include the exact value of cap and resistor. But the I am able hear the heartbeat with values of cpacitor and resistors below in the circuit.

These are the values I have used instead. I am currently simulating the circuit in TINA and using TL072 IC.


Thank you

 

[KlausST]

Hi,

you used C5 = 100n?

I don´t know how "breathing sound" is defined in technical terms, especially frequency range.
If you are too lazy to find out, I surely will not do your job again.

I will help you if you show your own effort.

Klaus

 

[MD_SHAHRUKH]

you used C5 = 100n?

No, I didn't use it in the circuit. It is only in the simulation.

I have placed one 10k ohm variable resistor in series with 5 k ohm in the position of R1 in the circuit. By rotating the potentiometer the breathing sound has improved. I am now working on plotting the signal properly. And breathing sound is in the range of 60Hz to 600Hz. Some diseases can push the frequency further to 2000Hz also. Heartbeat sound frequency comes under 15Hz to 140Hz and in few diseases, there are 700Hhz to 1Khz frequency sound comes from the chest including the heartbeat sound.

--- Updated Apr 30, 2021 ---

The voltage applied here is 11.1 V from a Li-ion battery, This is a battery application, I don't have a boost converter right now, so later I'll add to it for the power supply.

--- Updated Apr 30, 2021 ---

I have now changed the C2 cap to 1nF, and able to plot and hear the breathing sound. Few adjustments I think I need to do because of the amplification there is some sound break while hearing.

 

[andre_luis]

No, I didn't use it in the circuit. It is only in the simulation.

Most Spice-based simulators have easy to use built-in analysis in the frequency domain, so I would expect to see how did you infer the above values.

 

[KlausST]

Hi,

reduce the value of C6, too.

One could emphase the breathing or the heartbeat .. as you like. Either permanently or with keypress.
Show scope pictures of a typical heartbeat signal and a typical breathing signal at output of the circuit. Both with information about amplitude and timing.

****
I´m a measurement enthusiast. So if I had to design a high quality type of this, then maybe I´d use some kind of audio processor, like ADAU1701.
It may automatically adjust on frequency range and dynamically adjust volume. No soldering of capacitors to adjust frequency response. ;-)

Klaus

 

[danadakk]

Breathing rates/spectral analysis in the second range.



**broken link removed**


Your filter response should look something like

Signal in >> LPF >> 50/60 Hz Notch >>HPF or BPF >> Notch

DSP an excellent way of doing this.


Regards, Dana.

 

[MD_SHAHRUKH]

Here is the breathing graph, the sensor placed on the neck.

The leftmost to 3 boxes right is my inhalation and next 3 box are my relaxing and after that is my exhalation.
The plot is not final, I am working on it.

The Signal is being overlapped over the heartbeat sound coming from the vein near my neck.

S1 and S2 are close as I have a little higher BPM. The above pic is my heartbeat plot taken when sensor placed on chest.