Maybe some of you knowledgeable EE forum members can help me to reduce the ADC noise of my set-up?

Set-up:

1. 5V DC wall-PSU MEANWELL GS12E05-P1I PSU powers a 5V Adafruit Metro Mini microcontroller
2. Adafruit MAX4466 board powered by the microcontroller's 3.3V output pin
3. Adafruit MAX4466 gain pot centered so at around 75 x gain (25 x to 125 x possible)
4. All grounds connected
5. Tested in very silent room with good sound proofing

What I tried:

1. No fluorescent lamps or other electronic devices are present
2. Collecting 512 samples with ADC default speed and double speed
3. Standard 5V to the board, recommended 3.3V to the board with 3.3V analog reference voltage, recommended 3.3V to the board with 2.5V analog reference voltage
4. No capacitors, or 0.1uF ceramic capacitors from ADC input to GND and analog reference voltage input pin to GND as recommended by Adafruit
5. Used 3 Adafruit Metro Mini microcontrollers and 2 Adafruit MAX4466 boards, symptoms always identical

Results:

Signal always shows strong downward spikes and is sort of wavy in layman's terms, also when microphone is exposed to 1 kHz sine wave from a tone generator; please see attached image below.

Is there anything that can be done, or is this low-end board just no good? Thanks!

 

Just to add, when I use the same set-up, in silence, with 3.3V analog reference voltage as per the product tutorial, with no capacitors or else added, and then rectify in code (bias is Vcc/2 so 512 in ADC units), the noise is strong. Of course I don't expect a steady 0, but still. That much noise will also be in the signal, when the set-up is exposed to street noise, music, etc.

 

A low pass filter will reduce the levels of the noise spikes and will reduce the levels of high audio frequencies.

 

Looks like the Nyquist noise E=sqrt(4⋅R⋅k⋅T⋅ΔF), thus You have three ways to act: 1) to immerse the amplifier into liquid helium temperature; 2) to restrict the dF to the minimally necessay value, via some sort of bandpass filter; 3) to sharply decrease the input impedance of that amplifier. In to adittion, should assess the transistor noise graph from the datasheet, there always are certain i(c) and i(b) values what gives a best result and vice versa. Some very the few best transistors sorting the over full box may have Nyquist noise factor so low as 0.4 dB while 3 dB is count good, and 12-15 dB is count average. Thus, see You may got the decade better at need. Start Your tournee with BC546, 547, 549 and then look about 2222 and Riga semiconductor product kT3107 (1.5 and 3 dB).

 

Maybe some of you knowledgeable EE forum members can help me to reduce the ADC noise of my set-up?

Set-up:

1. 5V DC wall-PSU MEANWELL GS12E05-P1I PSU powers a 5V Adafruit Metro Mini microcontroller
2. Adafruit MAX4466 board powered by the microcontroller's 3.3V output pin
3. Adafruit MAX4466 gain pot centered so at around 75 x gain (25 x to 125 x possible)
4. All grounds connected
5. Tested in very silent room with good sound proofing

What I tried:

1. No fluorescent lamps or other electronic devices are present
2. Collecting 512 samples with ADC default speed and double speed
3. Standard 5V to the board, recommended 3.3V to the board with 3.3V analog reference voltage, recommended 3.3V to the board with 2.5V analog reference voltage
4. No capacitors, or 0.1uF ceramic capacitors from ADC input to GND and analog reference voltage input pin to GND as recommended by Adafruit
5. Used 3 Adafruit Metro Mini microcontrollers and 2 Adafruit MAX4466 boards, symptoms always identical

Results:

Signal always shows strong downward spikes and is sort of wavy in layman's terms, also when microphone is exposed to 1 kHz sine wave from a tone generator; please see attached image below.

Is there anything that can be done, or is this low-end board just no good? Thanks!

View attachment 319036

What you have to determine is if the ADC is loading down the mic board. But your method of changing the offset in programing isn't correct.
If the source of ripple is just the power supply, you need to see if 1/2 B+ ripple is in phase to the ripple in the signal.
This is accomplished easily by using two .1 to 1 uf (film preferred) in series across 3.3V to gnd and look at the ripple signal from this signal voltage divider. If its the same ripple in phase, then you use that as your Vref. This will subtract the ripple from center crossing. If out of phase, you use that point as the signal gnd.

But first you need to find out if you are loading down the input of the ADC. Because you might need to insert a 470 ohm to 1K resistor in series on the ADC input, and a 5K to gnd on the output of the mic. I think the ripple is from loading effects instead of just a poor quality power supply.

 

A low pass filter will reduce the levels of the noise spikes and will reduce the levels of high audio frequencies.

You mean, a low-pass filter between the 5V PSU and DC-jack? A resistor and a capacitor? An inductor and a capacitor? What values should I try (unfortunately, I have no oscilloscope at my disposal)? I don't have any inductors here, but could mail-order a set

 

changing the offset in programing isn't correct

I am just reading the signal at the ADC's analog input pin 0 at default speed and double speed. No offset is applied as far as the original image in post #1 is concerned. Without 3.3V Aref the signal sits at around 337. With 3.3V Aref around 512.

If the source of ripple is just the power supply, you need to see if 1/2 B+ ripple is in phase to the ripple in the signal.
This is accomplished easily by using two .1 to 1 uf (film preferred) in series across 3.3V to gnd and look at the ripple signal from this signal voltage divider. If its the same ripple in phase, then you use that as your Vref. This will subtract the ripple from center crossing. If out of phase, you use that point as the signal gnd.

Unfortunately, I don't own an oscilloscope, so I can't measure phase, ripple, etc.

But first you need to find out if you are loading down the input of the ADC. Because you might need to insert a 470 ohm to 1K resistor in series on the ADC input, and a 5K to gnd on the output of the mic. I think the ripple is from loading effects instead of just a poor quality power supply.

I've tried 3 different 5V wall-PSUs, no change.

 

Looks like the Nyquist noise

Thanks, Janis59, but I don't own a lab and can only use the MAX4466 board as it is, comparing it to the other hobbyist's adjustable gain electret microphone boards. Interestingly, the comparable Sparkfun SEN-12642 is nearly noise free, as was a rather costly Chinese clone, but with most Chinese Adafruit, Sparkfun, et al. product clones, the noise, or how what I see is called, is even worse by a wide margin.

 

Four fresh AA batteries, with 3.3V to the board from the microcontroller. Fist image shows no more wave, but still very noisy. So all three PSUs played a part. Then, the same set-up "out of the box" without using 3.3V Aref. No more wave, but the noise persists.

 

The noise spec for the low quality opamps (LMV324) on the cheap Sparkfun SEN-12642 and MAX4466 preamp boards is poor, producing a high amount of noise. Sparkfun talks about additional noise from a USB power supply.

The numbers on your oscilloscope photos do not say volts, ms, DBs or frequency and do not say where is the signal in the circuit.

 

The noise spec for the low quality opamps (LMV324) on the cheap Sparkfun SEN-12642 and MAX4466 preamp boards is poor, producing a high amount of noise. Sparkfun talks about additional noise from a USB power supply.

Good to hear from someone who knows. That means very little can be done other than finding a better 5V wall-PSU to at least get rid of the wave in the signal? Can you recommend a brand?

The numbers on your oscilloscope photos do not say volts, ms, DBs or frequency and do not say where is the signal in the circuit.

Values from the Metro Mini's 10-bit ADC obtained with analogRead() at default (prescaler 128) and double (prescaler 64) speed.

 

I have many poor quality 5V wall-PSUs but the circuits they are used with are well filtered and produce no problems.

 

Where is the power supply filtering? lol

 

I have many poor quality 5V wall-PSUs but the circuits they are used with are well filtered and produce no problems.

That sounds promising... how should the filtering be applied in this case?

 

Where is the power supply filtering? lol

Hence my initial question...

 

The main power supply connection is Vcc that has NO filtering.
R2 feeds most Vcc noise to the input of the preamp with C1 filtering frequencies above 1600Hz. If C1 is 10uF then all Vcc audio frequencies will be filtered away from the input.

 

If C1 is 10uF then all Vcc audio frequencies will be filtered away from the input.

All right, so the conclusion is that Adafruit relies on users either 1. using a battery to avoid the wavy output, or relies on users 2. believing the claim that the 3.3V microcontroller output is clean from interferences and noise when powered from an external PSU, while this is obviously not the case.

I can't exchange SMD parts for more suitable ones. In a nutshell, this product is inferior to Sparkfun's. Good to know. Thanks for all the input!

 

The main power supply connection is Vcc that has NO filtering.
R2 feeds most Vcc noise to the input of the preamp with C1 filtering frequencies above 1600Hz. If C1 is 10uF then all Vcc audio frequencies will be filtered away from the input.

lol, C1 should be a 10 uF and the Vcc 5V since that should be the 2.5V for the microphone. Who made this wonder board up?

 

The digital microcontroller is a very powerful noise generator and so that3.3 volt output will have a whole lot of noise on it. Does that microphone amplifier need to run on 3,3 volts, or could it run on a higher voltage? Use a separate linear voltage regulator to supply the microphone amplifier, with separate power wire connections to the 12 volt source, and be sure to install the recommended capacitors on that regulator. That should stop the noise problem.

 

Does that microphone amplifier need to run on 3,3 volts, or could it run on a higher voltage? Use a separate linear voltage regulator to supply the microphone amplifier, with separate power wire connections to the 12 volt source, and be sure to install the recommended capacitors on that regulator.

Thanks, there's only a 5V external power supply, see the initial post.

After very much ado, testing countless combinations, the best ADC signal output I obtained was like so:

1. 5V DC wall-PSU MEANWELL GS12E05-P1I PSU powers the 5V Adafruit Metro Mini microcontroller
2. Adafruit MAX4466 board powered by the microcontroller's 3.3V output pin via a 100 Ω/0.01 µF resistor/capacitor
3. 2,5V from a 1.5 kΩ/1.5 kΩ voltage divider from the 5V wall-PSU fed to the microcontroller's analog reference voltage pin, with a 0.1µF capacitor to ground
4. Microcontroller's ADC prescaler set to 32 instead of default 128, thus sampling at around 19 kHz instead of around 9 kHz
5. All grounds connected
6. Keeping all wires as short as possibly possible

The signal quality is not good compared to what I obtain from sampling from the output of a laptop's, smartphone's or quality podcaster's microphone headphone jack (with the necessary additional circuitry for the ADC input), but then that was what I expected. This is probably as good as it gets. At least quite a lot better than what you get following Adafruit's or other tutorials.

One would have to design one's own solution, with a quality capsule, quality amplifier components, etc. to achieve much better results.