1
\$\begingroup\$

I built a simple circuit (included below) on a breadboard and have been having some trouble with noise. Generally speaking, I get what seems like random noise averaging at about 1.5Vpp. Otherwise, I'm happy with the output and the amount of gain present.

I chose the ne5532 specifically for its low noise properties, especially with the amount of gain I am using, so this amount of noise was unexpected for me. I am using metal film resistors and ceramic caps (except for c1 and c2). I have tried to shorten wire as much as possible to reduce their parasitic capacitance. What should I be looking for to reduce the noise? Are breadboards just this noisy? If more information is needed, I can certainly provide it.

EDIT (MORE INFO): My input is a low Z dynamic mic measuring at around 3mV/Pa. Desired gain for each gain stage is 40 or around 32dB.

EDIT 2: I have included a clip of my oscilloscope reading (I know it's not a very good one.) Zooming in, I wonder if this is actually not noise but an oscillation? https://vimeo.com/872686526?share=copy

Simple Preamp

\$\endgroup\$
16
  • 1
    \$\begingroup\$ When presenting your design, it's helpful to share how much gain you designed for in each stage (saves us a calculation, or else lets us spot if there's a reason that gain can't be achieved). An important consideration here will be the magnitude of the noise when referred to the input (I presume your "1.5Vpp" is at the output). \$\endgroup\$
    – Ben Voigt
    Oct 9, 2023 at 17:33
  • 1
    \$\begingroup\$ It's also somewhat difficult to read the component values because of the dot grid and how close they are to the designators. I think that there are no decimal points in any of the feedback resistances and that both feedback dividers are 470k and 12k, so AC gain of 40 each. \$\endgroup\$
    – Ben Voigt
    Oct 9, 2023 at 17:35
  • 2
    \$\begingroup\$ Are the decoupling caps placed as far away from the OP amp in reality as they are in the schematic? \$\endgroup\$
    – Klas-Kenny
    Oct 9, 2023 at 17:41
  • 1
    \$\begingroup\$ Voltage noise from large-value resistors (especially R3=330k) contributes noise. With bipolar supplies, do you need C4, C5? If you ground the input, does noise go down? 1.5V p-p noise seems excessive. Transferring to PCB can reduce noise a little. \$\endgroup\$
    – glen_geek
    Oct 9, 2023 at 17:45
  • 3
    \$\begingroup\$ @EliY That's not good. The decoupling caps should be as close to the OP amp power pins as humanly possible. I've seen oscillating op amps on breadboards with bad decoupling before. Try placing 100nF straight across the power pins on the OP amp. \$\endgroup\$
    – Klas-Kenny
    Oct 9, 2023 at 19:34

3 Answers 3

1
\$\begingroup\$

Your first opamp has about 12k load impedance on the negative input. That is good for about 16nV/√Hz. The opamp noise is just 5nV/√Hz in contrast, about 10dB less noise than the resistor. The current noise of the opamp is about 0.7 pA/√Hz at 1kHz. On 12k resistance, that gets you another 8nV/√Hz.

Amplified by 40, it's more or less irrelevant how noisy the second stage is. Divide R1 and R8 by 5, and you'll get into a ballpark where the NE5532 starts making sense. C4 will still be ok for most signals (making for a corner frequency of 8Hz or so).

R3 is good for 25nV/√Hz but only if the input is left open. With a low-impedance mic connected, that will instead determine the input noise. So you should be fine there.

As a rule of thumb, at room temperature, 50ohms is good for 1nV/√Hz, and for every factor of 4 in resistance, the voltage noise doubles.

\$\endgroup\$
6
  • \$\begingroup\$ This is incredibly helpful. Thank you so much. Do you know of any books/materials where I can learn more about the kind of measurements you have provided? I haven't seen anything this detailed regarding amps before, and I would really appreciate learning more \$\endgroup\$
    – Eli Y
    Oct 11, 2023 at 16:41
  • 1
    \$\begingroup\$ Well, there is noise and noise. Johnson-Nyquist noise is inherent to resistance: you cannot make it go away by choosing "better" parts. You need different values. With the low voltages of a dynamic microphone, the saving grace is the low thermal noise due to low impedance. If you then bring in the noise via the other input, you are hosed. Of course breadboard, grounding, hum, contacts and so on can also impact the results, but the thermal noise is a given. You need to control it with circuit design: the execution cannot fix it. It's more of a preamp detail than an amp detail. \$\endgroup\$
    – user107063
    Oct 11, 2023 at 18:05
  • 1
    \$\begingroup\$ Oh, by the way: I didn't measure a thing. That's all theory. Learnt because I needed to figure out why my low-noise opamps were not producing low-noise results... \$\endgroup\$
    – user107063
    Oct 11, 2023 at 18:07
  • 1
    \$\begingroup\$ @EliY If you want to learn about noise, "Low-Noise Electronic Design" by Motchenbacher & Fitchen is a book used by many. However, the noise you are seeing is caused by some external device that is switching. It is not Johnson noise as that would look like fuzz, not blips. \$\endgroup\$
    – qrk
    Oct 11, 2023 at 20:29
  • 1
    \$\begingroup\$ @qrk Well, "of course breadboard, grounding, hum, contacts and so on can also impact the results, but the thermal noise is a given". You may be right that the thermal noise is not the elephant in the room on the scope. But when the potential elephant is gone, the Johnson-Nyquist terrier will still be biting the NE5534's legs. \$\endgroup\$
    – user107063
    Oct 11, 2023 at 20:53
1
\$\begingroup\$

Are breadboards just this noisy?

Breadboards are unsuitable for an amplifier that has a net gain of about 1600. I mean, 1 millivolt p-p of noise will produce 1.6 volts p-p of noise on the output. Apart from slight intermittent connections there are ground loop inconsistencies that just cannot be resolved with breadboard.

Then there's the inherent noise from your signal source to factor into the situation. And, your power supply noise and mid-rail generator noise could be significant too.

\$\endgroup\$
4
  • \$\begingroup\$ Thank you for your response. My input is a low Z dynamic mic with 3mV @ 1Pa. I don't believe it should be having nearly as much noise. Looking at my ground and supply rails in an oscilloscope, I don't see any noise. Is the best way to fix my problems just to transition to a pcb? \$\endgroup\$
    – Eli Y
    Oct 9, 2023 at 19:21
  • \$\begingroup\$ You won't see that sort of noise on a standard oscilloscope. Get your circuit design checked in a brand new question after firstly simulating it is my advice. \$\endgroup\$
    – Andy aka
    Oct 9, 2023 at 20:05
  • 1
    \$\begingroup\$ You can certainly stagger lower gain amps on a breadboard effectively, as long as you remember to low pass filter every now and again. The askers circuit does not have the low pass filtering, though. \$\endgroup\$ Oct 9, 2023 at 20:50
  • 1
    \$\begingroup\$ @ScottSeidman Well if a simple low pass filter (possibly between the two stages?) is all I need here, that is a very easy fix. Can you elaborate? \$\endgroup\$
    – Eli Y
    Oct 9, 2023 at 21:12
1
\$\begingroup\$

The noise could be coming from any number of sources like your test equipment, lighting, consumer devices, chargers, and other things can cause this sort of interference. If you have an anti-static mat on your table top or a metal table top, this can couple interfering signals in to the breadboard or input leads.

It's hard to read the resistor values due to the grid. From comments, I'll assume you have 470k, 8.12k (non-standard resistor value), 9.12k (non-standard resistor value) resistors. What worries me is the non-standard resistor values which means you didn't take care to create a proper schematic.

The oscilloscope picture shows AC coupling which means you don't know where the DC output level is. It it near zero volts, or is it close to the power supply rails? AC coupling should be avoided at first to determine if the biasing is correct.

You also need to increase the sweep speed of the oscilloscope to determine if the scope is aliasing the glitch signal. Details of the glitch signal may be interesting.
Digital oscilloscopes are full of pitfalls that aren't found in good ol' analog scopes, one of the reasons why I use an analog scope when dealing with low-noise, high-gain small signal amplifiers.

You are probably picking up some external signal that is switching around 1.25 kHz if the sweep speed is 100 us/div. If I read the resistor values correctly, looks like you have close to 70 dB of gain which is a lot of gain to have on a breadboard unless you are well practiced in dealing with high-gain amplifiers.

A picture of your breadboard and setup would also be helpful.

\$\endgroup\$

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.