Does a differential electret microphone configuration provide a better S/N ratio?

Discussion in 'The Projects Forum' started by skip.ele, Jun 30, 2013.

  1. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    I've been playing around with electret microphone configurations and I was wondering if (instead of the typical single ended configuration) a differential configuration would provide a better S/N ratio. My thinking here is that the differential signal now provides twice the voltage differential and the noise is common mode rejected. I don't have test equipment good enough to measure the actual levels here but to my ear it does sound like the S/N has improved over my previous single ended configuration.
    There is another trick to improve the S/N ratio by paralleling two microphones such that the signal is increased by 6db but the noise being uncorrelated only increases by 3db.
    Is this differential configuration doing the same thing or is it doubling the signal while keeping the noise at the inherent source level?

    skip
     
  2. #12

    Expert

    Nov 30, 2010
    16,257
    6,763
    Only one error. The differential connection diminishes the far field sounds quite a bit. It also diminishes the near field sounds, but not as much. They don't add together for near field sounds, they just have a slightly different sound level input for near field sounds, so the differential connection has a lot better signal to noise ratio, and that is very important. With modern amplifiers at about 79 cents each, millivolt signal levels are no problem. It's the signal to noise ratio that matters.
     
  3. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    Thanks for the response. Although I'm not sure I understand. When I say "differential connection" I'm referring to the wiring in my picture. In that configuration are you saying that sounds are "diminished" meaning that somehow this configuration reduces the effective sensitivity of the microphone? If that is the case can you explain?

    Thanks

    skip
     
  4. #12

    Expert

    Nov 30, 2010
    16,257
    6,763
    Complete misunderstanding. I thought your diagram was irrelevant because it only has one microphone, therefore the two microphones can not be connected in opposition with each other.

    You are talking about whether the amplifier is single ended or differential. Generally speaking, a single ended amplifier has one stage active at the original signal to noise ratio. That's one less than 2 stages contributing noise. Notice in LB-52, Fig 1, the equation for noise includes the current in Q2.

    Then, in, "Design of Ultra-low Noise Amplifiers", there is a way to parallel 2 transistors to get the 3db difference of which you spoke, but they are still not in a differential relationship.

    "Low Noise TUT.pdf" confirms this.

    Pulling from memory, 2 microphones in parallel cause a 6db increase in signal, but the "uncorrelated" noise voltages only add to a 3 db gain in noise. Still not a differential connection or a differential amplifier.

    In your drawing, C1 and R1 make no sense to me except as a low pass filter which partially grounds some of the signal into the 5 volt supply with a -3db point at 106Hz. That is, frequencies above 106 Hz are diminished. Re-draw it like this and you see that you have a switchable, low pass, filter. Turn the filter off and see that you have 2 of exactly the same signal going to your differential amplifier. This does nothing for the signal to noise ratio.

    And then, there is crutschow. He always knows ways to improve on my concepts. Maybe he will contribute.
     
    Last edited: Jul 1, 2013
  5. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    Sorry for the confusion. The attached diagram is the subject of this thread. Just trying to learn the pros and cons of this type of configuration as compared to the more typical single ended configuration where the microphone is connected to only one input of the amp. Pros and cons with respect to noise level, S/N ratio, common mode rejection, and so on...

    thanks.

    skip
     
  6. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    Thanks again for the feedback. I'll have to digest this tonight. R1 and C1 on the left side are there to inject the same noise from the 5v supply that the mic bias resistor R2 and C1 (forgive the mistaken duplicate use of numbers here) inject on the right side. In both this differential configuration and in the original single ended configuration injecting the same noise from the 5v supply to both inputs significantly reduced the noise on the output. It looks less complicated in the singled ended case (not shown) but even in the case per my picture it still significantly reduces the noise on the output. Maybe there is a better way to do this?

    Maybe I'm using the term incorrectly but when I say "differential" in this case I'm just referring to the opposing phase of the two signals from the top and bottom of the microphone.

    thanks
    skip
     
  7. #12

    Expert

    Nov 30, 2010
    16,257
    6,763
    Maybe I'm wrong. That's why we have about a hundred people contributing here.
    I've always gone for the single ended configuration because of the noise contribution of the second transistor in a differential pair.
    Injecting "equal" noise from the supply voltage kind of makes my brain sizzle. It's so easy to filter the noise out of the power supply, especially with modern voltage regulators, and 10uf of capacitor can be bought for about 10 cents. I haven't even explored that idea.
     
    Last edited: Jul 1, 2013
  8. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    well, it is funny and maybe there is something else going on here but my mic supply is coming from a 78L05 regulator. You can see the entire original single ended schematic attached here. If the left side noise path (for lack of a better term) from the 5v supply on the preamp input is removed then the output is far more noisy. By adding R1 and C1 on the left I was just trying to counter balance the mic side. Maybe there is some other effect here that is faking me out. I'll try to analyze this further. Need to think outside the box...

    If I understand you correctly you are saying that the general use of a two input differential amplifier like this is likely to be more noisy than say a single input non-inverting type of configuration? If the op-amp is much less noisy than the signal source then is a differential input not better because it takes advantage of the op-amp's common mode rejection to block noise from the source?

    What I'm after here is a way to minimize the noise from the microphone. My op-amps are pretty low noise I think (5nv/Hz) and I'm assuming that what I perceive as a high noise level in my output is coming from the microphone and possibly the high value feedback resistors. If I ground my preamp inputs without the mic then the amp is much quieter So I'm focusing on the mic at this point as the source of noise. I'm trying to figure out the best configuration for the lowest noise.

    thanks
    skip

    P.S. Scematic Correction: The resistors R11, R12 and R13 are shown with incorrect values for a 300hz high pass filter. The correct values are R11 = 1k, R12 = 390 and R13 = 6.8k.
     
    Last edited: Jul 9, 2013
  9. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    Yes, I thought the same thing. You could eliminate C1 and R1.

    Just wondering, what is your purpose for the AGC circuit ?

    EDIT: in Amp rev 3 schematic, you've, got a mic drawn single-ended and a connector for audio in balanced ??
    Could you clarify :rolleyes:
    Also, you might want to pull the LP-HP filters out of the AGC loop put them before the AGC stage, to condition the signal before that stage.
     
    Last edited: Jul 1, 2013
  10. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    thanks for the feedback tubeguy. Removing R1 and C1 on the left side makes the amplifier noisier. I can't see removing it unless I can somehow get rid of the noise on the mic bias side. I'll experiment with that some more.
    The purpose of the AGC is.....AGC. :) With an overall gain of over 1000 close loud sounds and physical handling noise drives a nail in my ears with headphones on. My goal is to pick up very faint sounds. The AGC is there to protect against the inevitable.
    The reason for the input being like that is because I wanted to experiment with different microphone configurations. I wanted to be able to try one or two mics to play around with noise cancelation. I have done some of that and it has been interesting for sure but turned out not to work very well. Even with two mics glued together shock induced signals did not cancel out. The balanced input with R1 and C1 first served to lessen the noise when using just one mic but later served to bias the second mic as well.
    Your comment about putting the filters inside/outside the AGC loop has me confused a bit. Could you please clarify the need to change that? Also I have heard some folks talk about putting the filters in between the preamp and main amp (maybe this is what you meant?). What are your thoughts on that too?
    In general this project was started just as a learning experience. I spent quite a lot of time researching and designing this just for the opportunity to learn about analog circuits and op-amps. All the stages here pretty much account for the typical parts of an amplifier. It has been a fun project up to this point and now I am trying to take it one step further in refinement with the help of the kind folks on this forum. :)

    thanks
    skip
     
    Last edited: Jul 2, 2013
  11. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    Getting late, couple things. Putting the filters before the AGC stage may reduce the noise because the bandwidth will be reduced before the added gain of the AGC stage.

    The C1 R1 we were talking about were in the first circuit posted. If removing those help lessen the noise, then it appears the noise is coming from the mic.
    Have you tried other microphones?
    Does the noise lessen dramatically when the mic is removed?

    Differential input circuits are used to cancel unwanted noise signals which are picked up from outside sources on both inputs equally and in phase - like hum from 60 cycle AC. Internal Mic noise developed between the mic hot and gnd won't be cancelled.

    Gotta go for now.
     
  12. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    Continued thanks for staying with me on this. I am willing to relocate the filters just to get to the bottom of the mystery around this. For every case I hear in favor of putting them in late I hear a case for putting them in early. Seems to be a lot of controversy around this but nothing beats good old experimentation. :)
    I'm pretty sure that removing R1 and C1 from the first circuit posted increases the noise on the output. I'll remove them again and confirm this.
    I have tried many different mics but all the same model number.
    The noise gets far less when the mic is removed. Much much less.
    I understand the common mode noise rejection of external sourced "pick up" but I was hoping to find a configuration that would also reject the noise from the mic. Is the noise across the mic as seen at the top and bottom points of the mic always out of phase and therefore it does not get rejected by the diff amp? If that is the case then the differential mic configuration would be worse rather than better?

    thanks
    skip
     
    Last edited: Jul 2, 2013
  13. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    My opinion is to put the bandpass filter section early in the signal chain to remove unwanted/un-needed frequencies before further processing.
    But try filters in different positions in the signal chain. I've learned a LOT by experimenting.

    If you can, try some different brands of mic capsules.

    You could try a battery supply to bias the condensor to make sure you aren't getting some unexpected noise from the 5 volt regulator.

    I think single-ended is the way to go for the pre, because the noise floor of the mic is what it is.
     
  14. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    Ok, I made some modifications to the schematic per our conversations here. I have replaced the preamp and microphone configuration with a single ended non-inverting stage. This allows for higher input impedance and lower resistor values overall. Both of which I understand should reduce noise. I moved the filters out of the AGC loop and placed them before the final amp.
    I have a question about this. Tubeguy, you mentioned that placing the filters before the main amp might reduce noise because of the reduced bandwidth going into the main amp. Are you talking about true noise reduction here or just improved reduction of unwanted frequencies that the filters are rolling off?
    Also, should the AGC output tie in before or after the filters? Is it possible that the AGC circuit could introduce noise that I would want the filters to remove?
    I also put a heavy low pass filter on the mic bias.

    It will take me a couple of weeks to turn out a board so I don't have any results right now. I just wanted to post the changes for comments.

    Thanks
    skip

    P.S. Scematic Correction: The resistors R11, R12 and R13 are shown with incorrect values for a 300hz high pass filter. The correct values are R11 = 1k, R12 = 390 and R13 = 6.8k.
     
    Last edited: Jul 9, 2013
  15. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    I answered below in blue . (another did this -liked it :))
     
  16. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    If you want to experiment with using 2 or more mics at the same time, you should use a mixer circuit to properly combine the signals. I would recommend using another mic preamp stage for each additional mic to properly buffer the mic(s) before mixing.

    You could also then easily invert signals, add more EQ etc. More fun.

    Happy 4th, gotta go.
     
  17. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    Happy 4th to you to.

    Still toying around with the differential mic configuration that I started this thread with and the corresponding differential preamp since that is what I currently have built up I decided to check your advice and the advice of #12, by looking at the regulated 5v microphone supply. My scope doesn't go low enough on volts/div to be able to see much below a few millivolts. So I started by adding a 10ma load to the 78L05 regulator thinking that maybe the .5ma max current being drawn by the mic might not be sufficient load to provide the best regulation from the regulator. This resulted in no perceivable improvement. So I then started bypassing the 5v regulated mic supply with larger electrolytic caps. Immediately at 100uf bypass I noticed a drop in high frequency noise on the output. I then tried 470uf and the noise was reduced even further. Still not happy I grabbed a 2200uf cap and tried that. It made a huge difference. The amplifier with the mic installed and with the 2200uf bypass cap is very quiet.
    Now, can someone tell me why it takes 2 2 0 0 uf to settle the noise on a 5v regulator with a .5ma load??? I realize that this noise is apparently very low amplitude since I can't see it on my scope. Its high enough for the amplifier gain to make it audible though. I am attaching another pic for the purpose of discussion on this post.

    Thanks again for letting me drag you through the mud on this one, Tubeguy.

    skip
     
    Last edited: Jul 4, 2013
  18. skip.ele

    Thread Starter Member

    Nov 27, 2011
    40
    0
    Ok, I'm going to take a stab at answering my own question. At 20Hz a 2200uf capacitor has a capacitive reactance of about 3.6ohms. Let's say for example that the noise on the 5v regulator produces even as little as .00005amps - that's 50uA. 50uA through 3.6ohms produces .00018v or 180microvolts. 180microvolts x 1000 (amplifier gain is 1000) = 180mv or .180v which is easily detectable in headphones. These numbers are just pulled out of the air but my point is that even a very tiny noise on the input can be a problem on the output. It takes a capacitor this big to get the reactance low enough to get the noise level down low enough for it not to be a problem on the microphone bias.
    Does that sound right?

    thanks
    skip
     
    Last edited: Jul 9, 2013
Loading...