0
\$\begingroup\$

I'm having issues with the amplified noise my op-amp is generating, and also I don't fully understand its behavior when a speaker is introduced.

First issue: in the first Picoscope image, you can see the signal and FFT in the "white noise" node when a speaker is connected and disconnected in parallel to R10. The white noise's FFT has a peak at about 10.8 MHz and some other peaks at ~980 kHz, ~1.1 MHz, and 1.22 MHz when the speaker is not connected; and those peaks appear at that node when I connect it to the NE5532P op-amp. They go away if I introduce the speaker in parallel to R10, which is not a bad thing but I wonder why is it happening and if it could work without the speaker.

Second issue: as you can see in the second Piscoscope image, I'm not able to amplify the signal reliably, even though the op-amp has a very low noise of its own and a fairly wide gain bandwidth for this purpose (~10 MHz). But the amplification is completely distorted (it's not even white noise!). That worsens if the speaker is introduced in parallel, as you can also check on the second Picoscope image. I wonder why am I doing wrong with the noise's amplification, why those periods appear (they seem to be 40-50 kHz apart) and how can I deal with them to get a true amplification of the signal.

Thank you very much!

A brief explanation of the circuit: I have a buffer stage that allows me to create a virtual ground to have a "-12V" into the op-amp. Then Q2 is reverse-biased and produces an avalanche into the base of Q1. That is the source of the white noise. Then I'm trying to amplify that white noise signal through an op-amp with a non-inverting setup.

Datasheets: LM358 (buffer), NE5532P (amplifier), 2N2222A (transistor).

EDIT: Following Andy Aka's suggestion I've added power supply FFT (last picture) as the periodic peaks seem to be contaminated from it. But I still don't know how to improve the design to have reliable amplified white noise.

schematic

simulate this circuit – Schematic created using CircuitLab

enter image description here

enter image description here

enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ You’re running in circles because you never told us what you intend to do. Each time someone solves your immediate problem, you change it so much it doesn’t work anymore. This will cause you lot and lot of grief. Please describe the entire application, so that every bit of it can be scrutinized. You don’t know enough what’s important and what isn’t yet - that’s a common problem for beginners, and nothing wrong with that. To make things really simple: this is a circuit that normally takes an hour to put together and get working well. Ask with this in mind, otherwise you’ll spend ages on it. \$\endgroup\$ Commented Apr 9, 2023 at 19:43
  • \$\begingroup\$ To make your life easy, I’ll gladly build the thing and show a working schematic, but you absolutely must say everything you expect this thing to do (and not do). Is this meant to be a white noise generator with a speaker? As an alternative that would take 30 minutes: an Arduino with a LFSR controlling four mosfets driving the speaker directly. Maybe 10 components total (Arduino board counts as one), and software defines what it does. Just so you know there are many simple ways of doing it. \$\endgroup\$ Commented Apr 9, 2023 at 19:52
  • \$\begingroup\$ @Kubahasn'tforgottenMonica thank you very much! You’re right, I’ve been driving in circles for three weeks now and meeting dead end after dead end. So I really appreciate your help! I’m trying to create a noise generator that can switch between white, pink and red noise. Fo that I’m using a reverse biased transistor. I should be able to listen to it and adjust the volume through a potentiometer. I’d like to make it with simple electronic components such as transistors, op-amps, ICs (for learning purposes) \$\endgroup\$
    – ludicrous
    Commented Apr 9, 2023 at 22:16
  • \$\begingroup\$ I’ve been facing multiple colateral challenges. For example, not having a negative voltage. I need the op-amp because the output I’m getting at the output of the second transistor is too slim (10Vpp) and not audible. The circuit should connect through a barrel jack to a DC power supply of 9V/12V. I have tried multiple ways of getting negative voltage to supply the VEE of the op-amp. Also have been centering the noise close to zero and flatten the FFT for low frequencies. I even bought low noise/ wide bandwidth op-amps I thought would help, but each solution seems to be a dead end \$\endgroup\$
    – ludicrous
    Commented Apr 9, 2023 at 22:41
  • \$\begingroup\$ I also need the op-amp to produce the pink and red noise (I thought of producing the red noise by integrating the with noise with an op-amp). Perhaps I could do this with transistors (and this way don’t have a need for a negative power supply) but the gain seems to be very slim through the transistor method as far as my circuit goes \$\endgroup\$
    – ludicrous
    Commented Apr 9, 2023 at 22:45

1 Answer 1

2
\$\begingroup\$

First issue: It's probably power supply noise. Try measuring the power supply and do an FFT.

Second issue: There are two problems here. Firstly the most serious.

You are using an LM358 as a mid-rail splitter. This means that you are then expecting the LM358 to handle the heavy-duty currents fed into the 8 Ω speaker that route back to mid-rail i.e. you are expecting the LM358 to be able to handle 8 Ω speaker currents and still make a nice clean mid-rail voltage.

This won't happen. Try using a 1,000 μF capacitor to feed your speaker and have the speaker return wire connected to your -12 volt rail.

The second problem is that you are expecting the NE5532 op-amp to drive an 8 Ω speaker. This won't happen either.

\$\endgroup\$
10
  • \$\begingroup\$ Oh, I see the issue with the splitter. But if it is the speaker that is causing the issue, why does that period in the amplified noise appear at the output pin of the NE5532 even when the speaker is not connected (second Picoscope image)? Regardless of the speaker, do you think the LM358 a bad decision to use as splitter? \$\endgroup\$
    – ludicrous
    Commented Apr 9, 2023 at 16:55
  • 1
    \$\begingroup\$ You have two issues; one is likely to be power supply noise (you should check this as I suggest) and, the second is a poor design that just cannot handle speaker currents in the mid-rail splitter (solution offered in my answer). \$\endgroup\$
    – Andy aka
    Commented Apr 9, 2023 at 17:12
  • \$\begingroup\$ The last image I presented has the speaker disconnected, it's just the amplifier NE5532 and the load R10 (connected to the virtual ground*), but the peaks and periods are there so I wonder if there's something else off besides the speaker, since this behavior happens even when the speaker is not connected (meanwhile I made the correction you wisely pointed out in my circuit and in the schematics above, but the periodic peak remains) \$\endgroup\$
    – ludicrous
    Commented Apr 9, 2023 at 17:28
  • 1
    \$\begingroup\$ Well, I guessed you were using a wall-wart or other form of switched mode power supply and, if you are, this is the source of all the noise you are seeing. So, @ludicrous either change the power supply or add linear regulators to buffer the power supply noise from getting on the output. \$\endgroup\$
    – Andy aka
    Commented Apr 9, 2023 at 19:11
  • 1
    \$\begingroup\$ You should probably get rid of the rail splitter and throw the NE5532 back into a draw and replace the power section with an LM386 audio power amplifier. you could then power the noise generator from a regulated supply and use a linear 12 volt supply. \$\endgroup\$
    – Andy aka
    Commented Apr 9, 2023 at 19:58

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.