Electret mic and phone for heart and lung sound for telemedicine

[Nigel Goodwin]

R6 in your 1st schematic is 100k and is 47k in your 2nd schematic which explains the difference in gain.

Two 10uF capacitors in series produce 5uF. Then the -3dB cutoff frequency into 5.1k is 6.3Hz.
Every jumper wire is an antenna that picks up interference. All the rows of contacts and long wires on a breadboard are antennas.

The circuit has many RC low frequency cutoffs. Maybe their total cuts heatbeats.

They certainly do, particularly C4, C5, and C8 - but also C9 and C10 (why two?). Healthy heartbeats can be well below 1Hz, particularly in super fit people - and also unhealthy heartbeats could be uncommonly slow as well - mines the opposite, uncommonly fast and variable.

A young friend and workmate was in hospital many years ago, and they wouldn't let him out - so he asked why, to be told his heartbeat was far too slow and they were concerned with it. He pointed out that was perfectly normal, and due to his level of fitness (we called him Arnie, for obvious reasons - even though he was only short).

 

[rjenkinsgb]

but also C9 and C10 (why two?).

My addition, as both the opamp and phone mic connection can have voltages; the two back to back with a ground bias eliminate any possibility of reverse bias, no matter which side has the higher voltage.

Bigger caps certainly would not hurt - but remember many electret capsules are not good at low frequencies.

 

[audioguru]

Here is a typical electret mic response that will vibrate your bones:

 

[rjenkinsgb]

but remember many electret capsules are not good at low frequencies.

Relates to:

Two 10uF capacitors in series produce 5uF. Then the -3dB cutoff frequency into 5.1k is 6.3Hz.

Healthy heartbeats can be well below 1Hz, particularly in super fit people

Very different from audio frequencies!

That frequency response plot down to 20Hz is hardly typical. I've spent hours in the past trying to find wide response ones from major electronics suppliers.

The first two capsules I looked at, at random, on Mouser
frequency (f) 100 20,000 Hz
Frequency Range 50 ~ 16,000 (Hz)

So arguing or criticising 6.3Hz response without considering the exact mic response seems a bit silly.

And:
While the repeat rate may be below 1Hz, the frequency components of the pulse waveform are obviously at rather higher frequencies or would be inaudible.


It got me curious.. I found an online reference library of heart & lung recordings:

[Note - turn the volume down before trying any of these, especially if you have a sub - it just about bounced mine off the floor].
**broken link removed**

Using the "First heart sound" example:
**broken link removed**

Zoomed in to a single pulse (first half) the longest interval between major peaks is roughly 25mS, so the lowest audible component is somewhere around 40Hz. Any cutoff frequency reasonably below that should be fine.



On the high frequency side, looking at a spectral analysis, that recording is approximately 30db below peak at 400Hz & 60db below peak at 1000Hz but has content to 7KHz. The spectrum plot also puts the lowest component at 43Hz.

 

[audioguru]

I looked at hundreds of electret mics at Digikey.com and found only one more that shows a frequency response graph below 100Hz. Most of them "say" they have a 20Hz to 20khz response without a graph and without saying the +/- deviation of dB:

 

[rjenkinsgb]

Are there two digikey.com sites?

Looking at electret mics and then the selection "filters", the "Frequency range" box has 73 entries, if I've counted correctly.
Only five of those lines start at 20Hz or below; 55 start at 100Hz or above.
The actual item count for 20Hz or below is 52 out of 840. 100Hz or above, 521.

That's under 10% of their range that they specify as having a response down to 20Hz.

So I was wrong saying many electret capsules are not good at low frequencies.
It should say most are not good at low frequencies.

The mic capsule selection is very important, to get good response below 40Hz. A random one is likely to be a bad one.

 

[audioguru]

I used PRO electret mics from Audio Technica and they are good. I have never bought an electret mic cartridge. Mine came from toys and cell phones.

 

[doldett]

I was trying to measure the cutoff frequency using the function generator.

At point C, the low and high cutoff frequencies are 14 Hz and 3.1 kHz at the


The cutoff frequencies of point A, B, and C are almost identical.


Do you think frequency ranges obtained from this circuit is ok for capturing lung and heart sound?

So I was wrong saying many electret capsules are not good at low frequencies.
It should say most are not good at low frequencies.

The mic capsule selection is very important, to get good response below 40Hz. A random one is likely to be a bad one.

Could it be the mic that i bought is not good at low frequencies? I purchased the cheap ones because I want all covid patients to be able to afford it. That is also why I have not used opa2134.

I will build the circuit on the perfboard next as all the components have arrived.

 

[audioguru]

Why is the load R13 a resistor and why is its value only 5.1k ohms? The output capacitors are polarized electrolytic and the high value of 10uF each because R13 is such a low value. If the load is 100k then the output capacitors can be one 330nF film capacitor.

You said that R13 is the input of your phone. If it is for a microphone then this circuit has too much gain and too much maximum output level.

An OPA2134 is a dual audio opamp that still works when a 9V battery has dropped to only 5V. It has Jfet inputs that provide an extremely high input resistance so that the 1M feedback resistor gives you the gain that is calculated. A cheap old NJM5532 dual audio opamp was designed for low resistance dynamic mics and its input resistance is a wide range of fairly low values which affects its gain in this circuit.

Will your Covid patients use this heartbeat and breathing sounds device to play to a far away doctor on their phone?
But doesn't your phone cut frequencies below 300Hz like for the last 150 years??

 

[rjenkinsgb]

Why is the load R13 a resistor and why is its value only 5.1k ohms?

Because that was determined early on as being appropriate to trigger the mic select circuit in the various phones!

low and high cutoff frequencies are 14 Hz and 3.1 kHz

That's fine; the professional example heart recordings I linked to above are almost entirely within 40Hz - 1KHz, above that it's over 60db below peak level; just noise, really.

Higher frequencies may be relevant to such as lung sounds, so I'd not try to limit the response any further.

It may be work buying a known-spec mic capsule; Digikey list quite a few with response claimed to 20Hz or lower, many for less than one dollar.
eg.

CMA-4544PF-W CUI Devices | Audio Products | DigiKey

Order today, ships today. CMA-4544PF-W – 20 Hz ~ 20 kHz Analog Microphone Electret Condenser 3 V ~ 10 V Omnidirectional (-44dB ±2dB) PC Pins from CUI Devices. Pricing and Availability on millions of electronic components from Digi-Key Electronics.

www.digikey.co.uk

CMA-6542PF CUI Devices | Audio Products | DigiKey

Order today, ships today. CMA-6542PF – 20 Hz ~ 20 kHz Analog Microphone Electret Condenser 4.5 V ~ 10 V Omnidirectional (-42dB ±3dB @ 94dB SPL) PC Pins from CUI Devices. Pricing and Availability on millions of electronic components from Digi-Key Electronics.

www.digikey.co.uk

PMOF-9745P-39UQ Mallory Sonalert Products Inc. | Audio Products | DigiKey

Order today, ships today. PMOF-9745P-39UQ – 20 Hz ~ 16 kHz Analog Microphone Electret Condenser 1.1 V ~ 10 V Omnidirectional (-39dB ±2dB @ 94dB SPL) PC Pins from Mallory Sonalert Products Inc.. Pricing and Availability on millions of electronic components from Digi-Key Electronics.

www.digikey.co.uk

Or an expensive? one claiming 15Hz:

POM-2738L-LW100-R PUI Audio, Inc. | Audio Products | DigiKey

Order today, ships today. POM-2738L-LW100-R – 15 Hz ~ 16 kHz Analog Microphone Electret Condenser 2 V ~ 10 V Omnidirectional (-38dB ±3dB) Wire Leads from PUI Audio, Inc.. Pricing and Availability on millions of electronic components from Digi-Key Electronics.

www.digikey.co.uk

 

[doldett]

Why is the load R13 a resistor and why is its value only 5.1k ohms?

These values can deceive 3 mobile devices (xiaomi, apple, and samsung) I had tested they are connected to the mic. Hopefully, it would work for several mobile devices.

An OPA2134 is a dual audio opamp that still works when a 9V battery has dropped to only 5V. It has Jfet inputs that provide an extremely high input resistance so that the 1M feedback resistor gives you the gain that is calculated. A cheap old NJM5532 dual audio opamp was designed for low resistance dynamic mics and its input resistance is a wide range of fairly low values which affects its gain in this circuit.

I am gonna order opa2134 to try and also mic rjenkinsgb recommended to see whether the overall performance could be increased. Anyway, the mic i bought, rjenkinsgb, only cost 0.09$ a piece . I think I will try the first one as it has wider operating voltage range. Many thanks for your recommendation!!

 

[Buk]

Fascinating discussion on the electronics to which I have nothing to contribute; but given you can buy a real stethescope for under £30; which comes with interchangeable heads bells:

The notion of mounting the mic (or even 2 for stereo) at an appropriate point inside the bell seems it might be beneficial.

You could go further and link it to a phone via bluetooth as with the commercial digital steths.

Two interesting notes from the specs of that Eko Digital:

1. It has a selectable 50/60Hz mains filter; which is probably a simple and useful addition to your circuit.
2. It has a builtin 0.01 Hz high pass filter. (Not sure what this is for?)

 

[audioguru]

This circuit picks up 50Hz or 60Hz only if its wiring is not shielded audio cables. A 0.01Hz highpass filter is used to block DC.
The above frequencies are in an ECG circuit that connects directly to the patient whose body is an antenna to pickup 50Hz and 60Hz and whose muscles activate with DC. An ECG circuit cancels these frequencies with common-mode negative feedback.

Edit: A telephone call does not play the very low frequencies of heartbeats.

 

[Buk]

Edit: A telephone call does not play the very low frequencies of heartbeats.

Cellular and the packet oriented BT ACL(async) channels may be limited to 4kHz bandwidth; but BT 2.1 SCO(synch) channels can stream upto 2.1 Mb/s.

From here:

The synchronous links support two bidirectional connections at a rate of 64 kbps. The data rates are adequate for audio and most file transfers. With the Bluetooth v2 enhanced data rate, data rates of 2.1 Mbps may be achieved with QoS enabled on asynchronous links by setting the appropriate channel parameters.

The SCO link is used for streaming rather than framed-formatted data.

 

[Buk]

The above frequencies are in an ECG circuit that connects directly to the patient whose body is an antenna to pickup 50Hz and 60Hz and whose muscles activate with DC. An ECG circuit cancels these frequencies with common-mode negative feedback.

Thanks for this. With further reading I've almost understood it.

 

[doldett]

Fascinating discussion on the electronics to which I have nothing to contribute; but given you can buy a real stethescope for under £30

Thanks Buk for suggestion. We can also find cheap stethoscope at around $24 (800 Baht). My friend and I, however, thought the price could be too much for majority of people who are losing their jobs as covid is rampaging our country. That's why we go with kind of diy stethoscope so that we can allocate more cost towards electronics.

 

[doldett]

Have some updates on the project today. I've finished putting the circuit on the perfboard and also have tried some recording of heart and lung.

Here are the recording. heart sound and lung sound

Recordings still seem to have some noise, but i think they are better than breadboard-recording. Anyway, the bass sound producing by the heartbeat (right?) is far weaker than **broken link removed** (thanks rjenkinsgb). Although I know 3m produces professional device, I am hoping that our little circuit would get at least 1/4 of that bass. Is there anything I can do to improve guys?

I am wondering if i could add more capacitors in parallel with c10 to make overall capacitance of 40 uF and get cutoff frequency at 0.78 Hz. Does back-to-back capacitors (c9 and c10) have to be equal?

 

[rjenkinsgb]

I am wondering if i could add more capacitors in parallel with c10 to make overall capacitance of 40 uF and get cutoff frequency at 0.78 Hz. Does back-to-back capacitors (c9 and c10) have to be equal?

That professional recording has no content below 43Hz, so it's not a low cutoff problem in the amplifier. If the amp cutoff is under 20Hz it should be fine at the needed frequencies.

It has to be either the mic / pickup assembly, or the phone itself has a high pass on the mic input.

Try different phones / tablets, and try adding a diaphragm across the open face of the pickup - even just strips of tape, as a test, to make it airtight.

 

[Buk]

try adding a diaphragm across the open face of the pickup - even just strips of tape, as a test, to make it airtight.

A balloon or condom might work well.

And maybe a smaller diameter (and deeper?) cap would localise the pickup somewhat.

Forget that. Looking at **broken link removed**, the head diameter is given as 48mm, and they look pretty shallow. And in this vid:

they clearly have a diaphram covering the bell.

 

[doldett]

That professional recording has no content below 43Hz, so it's not a low cutoff problem in the amplifier. If the amp cutoff is under 20Hz it should be fine at the needed frequencies.

It has to be either the mic / pickup assembly, or the phone itself has a high pass on the mic input.

Try different phones / tablets, and try adding a diaphragm across the open face of the pickup - even just strips of tape, as a test, to make it airtight.

A balloon or condom might work well.

Forget that. Looking at **broken link removed**, the head diameter is given as 48mm, and they look pretty shallow. And in this vid: they clearly have a diaphram covering the bell.

Thanks rjenkinsgb and Buk.

I am still waiting for the ones ordered from digikey to arrive. Will try that when they do arrive. I will next try using balloon or food wrap as diaphragm to see if i can notice any differences on sound quality.