1

I have 4 electret microphones for a project, each going to an AC channel. On the general case each mic has to be processed individually but in one particular case, I would like to have an image of a sound being made in front of all the microphones, as faithfully as possible.

Would it make sense in this latter scenario to average electrically the amplified output of all the 4 mics, to reduce the noise of each of them (at least white noise, I'm not talking about for example crosstalk from the ADC clock or this kind of correlated noise)?

Something like that for example :

schematic

simulate this circuit – Schematic created using CircuitLab

Florent
  • 370
  • 3
  • 18
  • Be aware that even if you properly implement your averaging, it will only be of use if the sound source is at the same distance from ALL of the microphones. Basically, straight ahead of the middle point of all the mics, with all of them pointed in the same direction. If the distances are different, then you will get summing and cancellation of sound depending on frequency and the distance differences. – JRE Aug 23 '17 at 09:29
  • @JRE very true, I haven't thought of it this way. So using only 1 of the microphones would be the best solution then? I was hoping to be able to improve the recording quality by maing use of all the hardware there is (sorry if this is outside the scope of the question and the SE). – Florent Aug 24 '17 at 23:51
  • related: [Can I connect multiple electret microphones for better gain?](https://electronics.stackexchange.com/questions/324450/can-i-connect-multiple-electret-microphones-for-better-gain) – handle Jan 13 '21 at 08:59
  • I suppose you can try to calculate and correct for the different delays between the source and the microphones. Simple cross correlations should give you the delays – Scott Seidman Oct 03 '22 at 18:26
  • Stupid question -- is there any reason to not try this with the sampled mic signals to see if it helps?? – Scott Seidman Oct 03 '22 at 18:28

2 Answers2

1

The output of an electret microphone element is a small (millivolts) AC voltage riding on a large (volts) DC voltage. Because no two elements have the exact same DC rest voltage value, a better way is to capacitor-couple each of the three signals to a 2-resistor voltage divider between Vcc and GND. This is then connected to the ADC input. In this way the three signals are summed to a single signal that is the algebraic average, and that signal is biased to the middle of the ADC input voltage range.

Are the microphone elements 2-lead or 3-lead?

AnalogKid
  • 16,865
  • 1
  • 13
  • 25
  • I have already biased the output of the mics (2 leads electret mics) with my amplifiers (simple single supply op amp schematic, I can add it to my OP if necessary). When you say riding on a DC voltage, do you mean that the electret's output is not centered around 0? – Florent Aug 23 '17 at 03:44
  • Pullup R biases open drain output Vdc near V+/2 which also sets the gain. – Tony Stewart EE75 Aug 23 '17 at 05:42
0

Electret microphone capsules in 2-wire configuration are current sources, so just wire all of them in parallel and adapt the bias resistor to a 1/4 of the value for a single mic. Your output current will not be an average as much as a sum of the partial currents, so expect the output voltage to be comparable in spite of the quite higher load.

As long as you stay in the saturation region of the capsule FETs (namely within the specified operating voltage range), the main sources of sample variation (quiescent current, pinchoff voltage) of the capsule FETs don't come into play and the principal variable is its transmittance factor. But that's directly responsible for capsule sensitivity, so for identical capsule types (which tend to use selected FETs) you cannot really do better even with more complicated circuitry.

user107063
  • 2,507
  • 1
  • 13