0
\$\begingroup\$

First, read this other question of mine. I have modified the amplifier, as suggested by τεκ to run on the full rail.

This is the modified circuit.

enter image description here

Notice how the amplifier ground is tied to -5 and the power to 5, meaning that it is powered by the 12 VDC coming from the external adapter... see schematics.

After turning the device on, I hear a hum and buzz sound that is bigger than the sound. What I did was this: after the capacitors C25 and C26, I have changed all 0V references to -Vcc (C21, C22, C23), pins 4, 5, 9, 12 and 13 of TEA2025, C27 and C28. But this is my problems:

  1. the final output, jack U2 0V. If I change that to -Vcc, this jack's ground will be 6V lower compared to the input jacks' grounds that are on the virtual 0. I don't think this is good.

  2. The potentiometers are grounded to 0V. I think this should stay at 0V right? because everything before C25 and C26 are between -5 and +5 and those caps are insulating this part of the circuit and the amplifier part, right? Or should I connect the potentiometers to -5V?

\$\endgroup\$

1 Answer 1

1
\$\begingroup\$

The TEA2025 is designed for single supply operation only, and has internal components connected its ground pins which will be referenced to the negative rail when powered by a split supply.

The MAX359 only draws 200uA from its negative supply, so I suggest you forget about trying to create a virtual ground and use a negative voltage converter (eg. ICL7660) instead. The LM317 can then provide a regulated voltage of 6 to 9V, to power the voltage converter and switching control circuit.

\$\endgroup\$
6
  • \$\begingroup\$ ok but see my schematics. Forget the virtual ground for a moment and focus on the DC external adapter. This adapter delivers 12 VDC, two wires, 0 and 12 VDC, right? I am using this to power the TEA2025. The virtual ground splits the 12 VDC, so we will have the new ZERO at 6V, that is on the middle of the rail. The problem is: the audio needs this virtual 0 because it is in the middle of the rail. Taking that into account the input jacks have their ground at this virtual 0 (that is 6VDC compared to the input adapter). Now I inject the audio into an amplifier. This amplifier ... (continues) \$\endgroup\$
    – Duck
    Apr 4, 2018 at 1:30
  • \$\begingroup\$ ... (continued)... is powered by the full rail, meaning that its "ground" is not at the virtual zero but at the adapter's zero, that is -6V compared to the virtual zero. So, I have the amplifier section supplied with -6V and 6V, or seeing from another perspective, adapter's 0V and 12V. At the output of the amplifier, I will see that its ground is at -6V. Remember that the input's ground is at virtual zero. I don't know if I was able to explain that... \$\endgroup\$
    – Duck
    Apr 4, 2018 at 1:33
  • \$\begingroup\$ ... (continued) I have uploaded a schematics of the modified circuit. See the amplifier, from the potentiometer to the final jack and how it is supplied by the full rail. And see how the final jack "ground" will not have the same potential as the input jacks' ground... \$\endgroup\$
    – Duck
    Apr 4, 2018 at 1:42
  • \$\begingroup\$ Even if it works, having different input and output ground potentials is a bad idea. C15 and C17 put half the power supply noise onto the virtual ground. You might be able to reduce this noise by removing C15 so the virtual ground is bypassed to the negative supply only. You should also remove C1 and C14, which are coupling power supply noise into the LM336. However this could make the regulators go unstable due to insufficient input bypassing. All-in-all this is a bad design that probably isn't fixable. \$\endgroup\$ Apr 4, 2018 at 6:44
  • \$\begingroup\$ do you think all problems would be solved if this was a real symmetric power supply, without using virtual ground? If so, how do I solve the difference of potential between the input and output grounds? \$\endgroup\$
    – Duck
    Apr 4, 2018 at 19:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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