I'm trying to fix a preamp (made by a colleague) that's noisy and somehow not amplifying enough. I should say right away that this project is pushing the edge of my electronics knowledge so I have a lot of questions. There's actually 2 copies on 2 boards, for one electret and one dynamic mics, but the design was the same except for the "plug-in power" for the electret (not phantom power, this expects about 5V, though I guess it's currently getting 12V). This is the current circuit (only R3 is different between the 2 boards).


simulate this circuit – Schematic created using CircuitLab

From my research I found a few things maybe should be done differently. First, it seems it's mostly power line noise that I'm getting. (I tried getting cleaner power supplies but on a small budget that didn't work out.) Instead, I could add some filtering on the voltage divider, and on the electret bias circuit. I also don't like the idea of attenuating after amplifying so I'd move the potentiometer to control the gain instead, as such:


simulate this circuit

But these changes mean I have to pretty much redo the whole thing, and the product is in a "finished" state (in a box with connectors and everything), so I'd like to minimize the changes if possible.
1) (main question) I was wondering if "pre-filtering" the power supply would be equivalent/better/worse to having the filters in the preamp circuits?
I was thinking for example adding a 2nd order RC filter on a separate little board. It would reduce the voltage a bit, but I can use small resistors and then do one good filter instead of 3?

The other issue is the gain. Note that the goal is to get line level audio, so that voice from the mics is at comparable volume to other sounds coming from a computer. The outputs goes to speakers (this is part of an intercom/mixer). The gain is already at 2k and still pretty weak. I'm thinking maybe the electret mic isn't getting the appropriate power? I think it's supposed to get around 5V, but I read in another thread something I didn't quite understand about the electret inner workings and how the resistor value would change how much amplitude it would output.
2) Any tips on what value to use for the electret bias?
Note that the dynamic mic board (without R3) also sounds weak, but I'll have to verify, its R5 might have a different value.

Finally, I also read about resistor noise and how I'm better off not going too high (and use metal film) with such high gains. But I'm not sure which resistors would be most critical. In the second schematic, I'm guessing R8, R4... Not sure if the voltage divider also, R3, R9/11? Related to that, I'm unsure what values to use for the voltage divider.
3) Other than not wasting too much power (prefer large values), or resistor noise (prefer low values), are there other factors to consider?
Note that the dynamic mic has output impedance of 600Ω but the electret, 5kΩ.
4) Should I "tune" the circuits differently, or just go with the highest one and give them about 50kΩ input impedance (R8)?

Any other comments or tips on the circuit would be appreciated!

5) In the second circuit, should the output after Cout be left "floating" or add a R to neutral, and if so what R value?

  • 1
    \$\begingroup\$ What type of noise is it? What does it sound like, and is there any way to bandpass and find the frequency. This could help in localizing the problem in the circuit or possible within your full setup. \$\endgroup\$
    – mcmiln
    Commented Feb 3, 2016 at 23:49
  • 1
    \$\begingroup\$ The noise actually sounds different with different power supplies, buzzes of various frequencies mostly, and goes away with a big (expensive) desktop regulated supply. I also started testing adding a filter before the preamps with encouraging results, but some other parts in the same box didn't work with the filter. I'm now getting ready to disassemble some more parts hence my post. \$\endgroup\$
    – zorgkang
    Commented Feb 3, 2016 at 23:58
  • \$\begingroup\$ What type of condenser microphone are you using? Is the gain adequate when used with a dynamic microphone? \$\endgroup\$ Commented Feb 4, 2016 at 0:07
  • \$\begingroup\$ To carry on with @ThreePhaseEel, also check the expected input on whatever equipment you're using to see what it can take. Also double check the preamp to make sure it can deal with 48V phantom power. I actually don't see phantom power. Can you verify? \$\endgroup\$
    – mcmiln
    Commented Feb 4, 2016 at 0:25
  • 2
    \$\begingroup\$ Oh, are you sure you have the polarity correct on the microphone? Condenser microphones have DC polarity. \$\endgroup\$
    – user57037
    Commented Feb 4, 2016 at 7:06

1 Answer 1


Your question is too long to read, so I'm only responding to the schematic. It's not clear which schematic you are asking about, so I grabbed the first one. To keep this answer consistent if the question is edited, here is the schematic:

  1. 12 V is quite high for driving a electret microphone. 3-5 V is more normal. Check the datasheet.
  2. 22 kΩ is quite high for driving a electret microphone. Did you even look at the datasheet?
  3. Noise on the supply will be directly coupled onto the input via R3.
  4. C4 and R5 form a high pass filter at 1.6 kHz that affects the overall frequency response of the amp. Frequencies below 1.6 kHz will have less gain. This makes no sense. Rolling off gain below 20 Hz is fine and is still considered "HiFi" audio. Rolling off at 100 Hz or so can be acceptable in some cases. Rolling of at 1.6 kHz is just bad.
  5. You are asking for way too much gain from a single stage. The gain you are trying to achieve (above the 1.6 kHz high pass) is (R4+R5)/R5 = 2.2k, which is absurd for a single stage. Let's say you only care about frequencies up to 10 kHz (HiFi goes to twice that), and you want 10x gain headroom for the feedback to work well. You are expecting the gain bandwidth product of the opamp to be 2,200*(10 kHz)*10 = 220 MHz. Even without the factor of 10 for the feedback that's way out of line.
  6. R6 makes non sense at all. I can't even guess what you think it does, but what it actually does is waste 1/2 the gain.
  7. The volume control on the output isn't a good idea, especially when you are expecting such a very large gain. Loud signals will clip before you can attenuate them with the volume control.

To fix points 1-3, use a resistor divider to make the right voltage for the electret. 20 kΩ on top and 10 kΩ at bottom will divide the supply by 3 to make 4 V, which is probably in the intended range of the electret. To filter out supply noise, break the top 20 kΩ resistor into two 10 kΩ resistors and put a cap to ground between them. The divider impedance at the cap is 10 kΩ//20 kΩ = 6.7 kΩ. That requires at least 1.2 µF for the rolloff frequency to be 20 Hz or less. You seem to have a bunch of 10 µF caps around, which would work well.

Connect the + side of the electret to the junction of the bottom two resistors. This drives the electret with 4 V at a dynamic impedance of 5 kΩ, which is a lot better than the existing circuit.

To fix the gain, it's good to start with what the opamp can do. The TL071 has a typical gain*bandwidth of 3 MHz. Let's say we want a gain of 10 headroom for the feedback to work well, so that leaves 300 kHz. Assuming HiFi audio, the gain should be flat to 20 kHz. That leaves a gain of (300 kHz)/(20 kHz) = 15. This was based on the typical, not minimum guaranteed gain. However, the factor of 10 for the feedback isn't exact. If it's only 5 at 20 kHz the closed loop gain will still be reasonably flat, so lets aim for 15x voltage gain.

15x voltage gain means R4/R5 = 14. Keeping the existing R4 means R5 should be 1.57 kΩ, so 1.6 kΩ it is. The C4-R5 filter should roll off at 20 Hz or lower, which means C4 must be 5 µF or more. Keep the 10 µF.

With a sane gain, you need extra stages to get line level and beyond. One advantage of this is that you can leave the volume control where it is, immediately after the first stage. The signals in the first stage won't be large enough to cause problems, even when very loud. 10 mV from a microphone would be a lot, which times 15 is only 150 mV, so that's all fine.

Or work it backwards. The TL071 might be able to swing 8 Vpp in this setup. That divided by 15 means it won't clip as long as the input is 500 mVpp. No electret or dynamic mic is going to put out that much.

  • \$\begingroup\$ Thanks @Olin. Your answer is as long as my question, only took 2 minutes to read carefully ;). A previous comment did point the gain bandwidth issue. I think I understand all you've said, but wouldn't the the power supply noise be also coming through R1/R2? That's why I had added C5 (and R8) in the second diagram. \$\endgroup\$
    – zorgkang
    Commented Feb 4, 2016 at 22:21
  • \$\begingroup\$ @zorg: If done right, R1//R2 should be high enough impedance so that whatever is coming thru Cin swamps any power supply noise. Doing what you did in the second schematic doesn't hurt other than add complexity, might help a little. I'd fix this by making R1 and R2 larger. After all, they only need to overcome the leakage current of the opamp input, which is quite small for this amp. 1 MOhm each sounds about right. Noise is definitely coming from the supply thru R3 though, which you need to fix. That's why I said to use a divider and break the top resistor in two to allow for filtering. \$\endgroup\$ Commented Feb 4, 2016 at 22:38
  • \$\begingroup\$ Regarding 12V on the electret, to make a long story short, the voltage divider idea doesn't work well. The mic itself makes a divider with R3. I found through testing that I get the best output from the mic with 12V supply (tested 3 to 12 with clean supply) and 43k (tested 0 to 50k), which gives 4V across the mic. I'll have to test further to figure out how much gain I need and where filtering is required, based on your and other's suggestions. (And I don't have a datasheet, it's a Sony mic, I found few specs for it. Had to resort to testing.) \$\endgroup\$
    – zorgkang
    Commented Feb 5, 2016 at 22:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.