0
\$\begingroup\$

I'm building a guitar audio processor. I inherited this circuit from someone else's project. I am seeing a significant amount of noise on the input. I digitize the input using a 16b ADC and clip to 12b to eliminate quantization noise. However, I was still seeing a lot of noise (+/- 7mV) even when the input signal prior to the preamp is clamped to 0V. If I bypass the preamp and clamp the input to the ADC to 0V I get a clean 0V signal. So, I know the source of the noise is the preamp. The schematic is attached. As you can see the analog input (SW_JACK_IN) is passed through a CAP to remove DC and then a 1.65V DC is added to put the signal mid-rail of a 3.3V range. The input is a basic unamplified audio signal from a guitar, so not much more than a few 10's of mV. It then passes through a low-pass filter set for about 20KHz, a opAMP amplifier and a second low-pass filter also at about 20KHz. The final analog signal is then digitized through an ADC at 96Khz.

When I do a SPICE simulation of this circuit I see a huge sensitivity to noise. I'm trying to figure out how to reduce the noise. I suspect the problem is in the opAmp amplifier. Would I be better off just doing a voltage follower prior to the amplier to get a high-impedance input prior to amplifying it?

enter image description here

\$\endgroup\$
3
  • \$\begingroup\$ C20 would do more good on 1V65 than 3V3. Your C4 says 68nF but you meant 68pF. TL972 has a rather large input bias current - even nominal 200nA will generate 0.2V offset after flowing through R11 and then that gets amplified by your gain. What gain do you have (1 + R4/(RV1+R5) when you observe this 7mV noise? Look for a rail-to-rail opamp with much lower input bias current. \$\endgroup\$
    – td127
    Commented Apr 6, 2023 at 2:34
  • \$\begingroup\$ Do you get mains hum or what kind of noise? Can you show a trace? And also a sidenote: "I digitize the input using a 16b ADC and clip to 12b to eliminate quantization noise." - What ? If you clip to 12 bit, you get 12 bit with the quantization noise of 12 bit. \$\endgroup\$
    – tobalt
    Commented Apr 6, 2023 at 7:23
  • \$\begingroup\$ Do not use a ceramic cap like a X7R for C1 as hi-k capacitors are generally piezo-electric. This means they can work as microphones. \$\endgroup\$
    – Kartman
    Commented Apr 7, 2023 at 16:13

1 Answer 1

0
\$\begingroup\$
  1. If you look into the data-sheet, then you can see that the noise is going up very fast if the output voltage is higher. With a load resistor of 10k, the noise is lower over a wider voltage area. The best voltage area is around 0.1V till 0.9V with a load of 10k Ohm.

Page 6 - https://www.ti.com/lit/ds/symlink/tl972.pdf - Figure 4. Total Harmonic Distortion + Noise vs Output Voltage

  1. The resistor values are very high and your circuit is possibly not shielded. You have magnetic fields in your environment and they can create a current in the traces of the PCB or breadboard. You can try to use a shield with a metal canister or a mesh net what is on ground potential.
\$\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.