[SOLVED] Noise and inproving electret preamplifier

[djdaga]

Greetings.

I made this little preamplifier for electret microphone, but I have a problem. Sometimes background noise is quite annoying, as well as high frequency noise (aka the sssssssssssssss). I've worked with some filters and software level and I don't get acceptable results. The microphone picks up on conversations from neighbours, and all the noise mixes with the voice.

I need some advice on how to improve this circuit. I have 12V regulated. If you see the cable is well done and also good quality. Operational *RC4558*/JRC4558/LM4558... I don't have any others on hand.

Here a demo: https://www.youtube.com/watch?v=mSlaXgVjXWY

 

[Audioguru]

The 4558 dual opamp is very old. It has fairly low noise when it is fed from a low impedance dynamic mic, not a higher impedance electret mic.
Modern audio opamps have lower noise, especially when the mic is electret.

Your mic is probably a cheap omni-directional electret that picks up sounds all around it from the neighbours. A cardioid (directional) mic is used in a gooseneck mic and has a vent at the rear to cancel rear and side noises.

You forgot to post your schematic for us to see if resistor values cause some noise or maybe the supply voltage for the electret mic is too high.
How are you disabling the second unused opamp? Maybe it is oscillating and causing noise.
How much voltage gain for the opamp and how close to your mouth is the mic??

 

[djdaga]

The 4558 dual opamp is very old. It has fairly low noise when it is fed from a low impedance dynamic mic, not a higher impedance electret mic.
Modern audio opamps have lower noise, especially when the mic is electret.

Your mic is probably a cheap omni-directional electret that picks up sounds all around it from the neighbours. A cardioid (directional) mic is used in a gooseneck mic and has a vent at the rear to cancel rear and side noises.

You forgot to post your schematic for us to see if resistor values cause some noise or maybe the supply voltage for the electret mic is too high.
How are you disabling the second unused opamp? Maybe it is oscillating and causing noise.
How much voltage gain for the opamp and how close to your mouth is the mic??

Sorry. Pleas see the image.

 

[KlausST]

Hi,

This circuit simply is bad.

* You need a bulk capacitor (let's say 100uF) at the power supply. (To stabilize power supply)
* add a ceramics 100nF across the Opamp's supply legs (to avoid Opamp ringing)
* add a 10uF capacitor across R5 (reduces high frequency noise)
* replace R3 with 12k (drops gain and reduces noise)
* connect a 15nF across R3 (reduces high frequency noise)
* connect C2 to the output of A1 (instead of A2. A2 is useless now. It is not needed)

I wonder why A2 is used at all:
A1 amplifies the signal, while A2 attenuates this amplified signal.
This combination
* adds noise
* reduces dynamic range
* adds distortion
* consumes power

I assume you picked this circuit from an unreliably hobby circuits site.
I recommend to look for such circuits at Opamp manufacturers site.

For an example:
An internet search with "texas instruments electret amplifier circuit"
Brings you to this document: https://www.ti.com/lit/ug/tidu765/tidu765.pdf
It tells a lot of detail information, so that you are able to optimise the circuit for your requirements.

Try other manufacturers, too.

Klaus

 

[Audioguru]

I agree that the new circuit is very bad. It also has R1 connected directly to the power supply so that it amplifies supply noise. Instead, R1 should connect to an RC filter. Also the value of R1 is too high since most electret mics draw 0.5mA leaving almost no voltage on the mic.

The fix is also bad:
If you change the value of R3 to only 12k then the voltage gain of the first opamp becomes about 1 (the impedance of the electret mic is about 2k ohms in series with R2). A mic preamp usually has a gain of 100 to 200.
When 15nF is added across R3 then all audio frequencies above 890Hz are reduced to make very muffled sounds.

 

[KlausST]

Hi,

The fix is also bad:
If you change the value of R3 to only 12k then the voltage gain of the first opamp becomes about 1 (the impedance of the electret mic is about 2k ohms in series with R2). A mic preamp usually has a gain of 100 to 200.
When 15nF is added across R3 then all audio frequencies above 890Hz are reduced to make very muffled sounds.

True, it may be bad in the meaning of audio usability.
I´ve simply tried to maintain the input impedance (10k) and the overall gain (about 1.2) of the original circuit.

The 15nF is of my mind calculation was wrong. I tried to get fg of about 3500Hz. This should be enough for speech, while it may reduce the hissing noise.
Now that I have a calculator by hand: 3.9nF is the more correct value.

Klaus

 

[Audioguru]

What did you say? What did you say? Did you say "A" or "say"?
A 3500Hz cutoff is OK for vowels but is horrible for speech. But speech is not just grunts and groans, it has many important high frequency consonants that easily reach 14kHz. The TI circuit has a cutoff of 133kHz so that 20kHz is reduced only -0.1dB.

Also, the TI circuit does not have R2, the opamp is not a voltage amplifier instead it is a transimpedance current to voltage converter.

 

[djdaga]

Hi!. I'm going to run some tests and let you know.

 

[KlausST]

Hi,

Audioguru, I know what you mean.
So it's indeed on the OP what bandwidth to choose.
I don't know what signals need to be amplified and what frequencies need to be suppressed.
Because the OP talked about "conversation" in post#1...I assumed he is interested in speech audio bandwidth.
The frequencies are from: https://en.m.wikipedia.org/wiki/Voice_frequency

And the cutoff is just the -3dB point even at double the freqency you still get about -8dB. This is not "nothing"

I hope you agree:
* the higher the cutoff frequency the better the audio quality
* the lower the cutoff frequency, the better the suppression of high frequency hiss.

Klaus

 

[Audioguru]

FM radio stations have an upper frequency limit of 15kHz so that no audio interferes with the 19kHz pilot tone.
CD and MP3 players have an upper limit of 22kHz so that sounds are normal.

Polycom make some high tech speakerphones and tele-conferencing systems that use two telephone lines for double the bandwidth so that voices sound clearer.
When I worked with speakerphones and tele-conferencing systems I sent tones out on one phone line and measured the response received on the next phone line and 3kHz was down at -12dB so I complained to Bell. They said it is fine because their limit is -15dB. Then I made equalizers that boosted frequencies above 2.5kHz peaking at 10kHz. When I demo'd each one in a tele-conference system, they liked it and many were sold.

 

[djdaga]

Greetings. I followed everyone's suggestions, facts and considerations.

I did some tests and ended up doing the "Single-Supply, Electret Microphone Pre-Amplifier of John Caldwell"... Very good results. I decided to set a potentiometer to avoid the tangling with the resistors, which I don't have on hand. I was reading the operational data sheet set out in document (172) and comparing it with 4558.

It works very well at first, and that's what you'd expect. Although in the insulation box the voice is boxed in. In any case, it is a matter of filtering and/or post-processing. Of course, the components need to be placed on a pcb, metal insulation, and good grounding..., this will surely improve performance. The components are subject to change, all depends on further testing. For example the 13k resistance, I don't see any use for it...

Listen to the audio on youtube. Although there are background sounds, I can control that. https://www.youtube.com/watch?v=4bsCmVZbenY

 

[Audioguru]

For example the 13k resistance, I don't see any use for it...

His 13.7k resistor R1 powers the Jfet inside his electret mic. The mic will not work without it. If your mic works without this important resistor then it is not an electret type.

 

[djdaga]

Ño is the 20k resistor for the micro... 13k is other.