I'm still a newbie with electronics beyond basic microcontroller stuff, so sorry if the answer's obvious.

I have this circuit, just an electret and a simple amplifier stage. I've built it on a breadboard, but I get some weird noise. Then if I touch the microphone it disappears enough that the circuit becomes actually usable.

The electret is the ABM-709-RC from Multicomp, the opamp is the OPA344PA from Texas Instruments.

Here's the output with the noise (and some sound in the middle):

And when I touch the microphone:

So it's not completely gone but more manageable. I think I got it to be less than that once.

My breadboard setup:

breadboard setup

(Yeah I should cut all those legs, I didn't because I was trying multiple values. Can that be the cause?)

I know I should probably do some filtering too, I tried that with the few parts I have and it didn't make it better. The power rails seem fine on the scope, though I suspect the power supply board is not very good quality because there's low frequency oscillations if I select 5V.

Thank you in advance!

  • 1
    \$\begingroup\$ Did you do any sound calculations (your desired pascal range as input to the process) through the electret and then when considering gain? Have you ever read anything like this before on stability? Do you know what voltage gain you are asking for with the circuit? \$\endgroup\$
    – jonk
    Commented Nov 28, 2022 at 2:28

1 Answer 1

  1. R5 should be placed directly across the IC without any jumpers. Leads as short as possible.
  2. Decouple caps: a) 100nF from pin 7 to pin 4 directly across the IC. No jumpers. b) R1 should be spit into two series resistors with a bypass capacitor to ground from their junction.
  3. C1 electrolytic is very nearly reverse biased. a) Make certain the voltage across C1 does not reverse bias even with signals from the microphone. b) OR, use a ceramic or film capacitor.
  4. U?-pin 3: Place a 100nF from there to ground.
  5. Place a small 5pF to 10pF capacitor across R5. This improves stability.
  6. The high gain will turn microvolts of interference into tens of millivolts. Clean up your layout. Wires as short as possible. minimize the area of current loops. As few jumpers as possible. None preferably.
  7. Keep power and ground connections as short as possible. Don't use the rails that go all around the board. Connect power and ground directly to the op-amp first then route to R1.
  8. Twisted pair out to the microphone.

These are speculations that can be verified by testing. Usually this kind of noise can be reduced by proper layout.

Incidentally, the GBW for this amp is 1MHz. So for a gain of 1000, the bandwidth is only 1KHz.

  • \$\begingroup\$ Just saw this. +1 definitely. You've covered all of my initial thoughts last night (#1, #2, #4, #5, and #6), and a few more I should have thought about. Nice. (At that voltage gain I did also want to see where the phase and/or gain margins were at but was too sleepy to bother.) The OP should follow this and select it as an answer, I believe. \$\endgroup\$
    – jonk
    Commented Nov 28, 2022 at 19:14
  • \$\begingroup\$ Thanks @jonk! Whether PCB or breadboard, layout is everything… well almost. \$\endgroup\$
    – RussellH
    Commented Nov 28, 2022 at 20:13
  • \$\begingroup\$ Sorry I didn't have much time to test yesterday, I'll do that by tomorrow. Thanks already for the detailed answer! \$\endgroup\$
    – Pik'
    Commented Nov 29, 2022 at 15:16
  • \$\begingroup\$ The mess of far apart parts and rows of contacts on a breadboard guarantees the oscillation. Make a compact pcb for it. \$\endgroup\$
    – Audioguru
    Commented Nov 29, 2022 at 18:18
  • \$\begingroup\$ Alright I've got time to implement #1 and #6 through #8, the weird oscillation has been replaced with a clean 50 Hz sine wave that's coming from the input signal. I'll investigate some more in a few hours but it's sufficient to mark as answer. Thank you! \$\endgroup\$
    – Pik'
    Commented Nov 30, 2022 at 17:21

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.