# AC coupling capacitor in the ground to break the ground loop?

I'm implementing the following audio single-ended to differential converter for an ADC:

This is the reference design for the PCM1804 that I intend to use.

What would happen if I do this:

Would it still work?

The reason I'm interested in doing this is demonstrated in the following diagram:

simulate this circuit – Schematic created using CircuitLab

Even though I'm using an isolated power supply for my DSP, my amplifier internally connects the signal ground to the chassis. There's clearly a huge ground loop, consisting of the ground shield in the RCA cables and the car chassis that is a big source of noise.

I thought inserting a capacitor in the way of the ground as shown above would break this ground loop. But apparently, this is not a good idea. Any other suggestions are much appreciated.

I am aware of ground loop isolator transformers, but that's a band-aid fix, and they create a large amount of distortion.

The whole circuit you provided is a single ended AC coupled to the differential output.

1st stage buffers (x -1) then the output is AC coupled on its output and AC coupled input for the noninverted on the bottom.

The output is a differential analog signal. But the input impedance is unbalanced and does not convert stray noise current or ground shift to a differential noise voltage such that the Common Mode noise cancels out.

Maybe what you want for better immunity is a differential amplifier with precision matched resistor ratios or an Instrument Amp (INA).

Your suggestions won't work at all as others have explained.

The cct that works has an input impedance of 4.7k//{3.3k+Z(10uF)}

There are other ways to do this but that was not your question.

• Thanks. But I did not understand Audioguru's answer. Can you please explain why it won't work? Is there an alternative to feed in single ended input without the grounds directly connected? Sep 29 '18 at 3:38
• Both inputs to Op Amps ideally use the same DC resistance for low input offset V from input bias current. But your 1st cct has no R.on Vin+ so Iio*R= an integrator to saturation when R is open Sep 29 '18 at 3:40
• I have to be honest, I'm totally lost. When you say my first circuit has no R on its Vin+, what does that mean? What is lio? Sep 29 '18 at 3:53
• Reworded my question and removed my first circuit to make more clear what I'm asking. Sep 29 '18 at 4:10
• I don't know the output impedance of my head unit, but it's probably in the 50 - 500 Ohm range. The problem I want to solve here is not directly connect the grounds of the audio signal to my ADC ground. I would really appreciate if you can draw any sort of diagram of what you mean. Sep 29 '18 at 4:28

Your circuit will not work because the (+) input of the opamp is missing a DC bias voltage that is at half the supply voltage. A simulation guesses that the opamp has no input bias current and the capacitor on the (+) input is already charged to half the supply voltage.

Your first circuit had its opamp (+) input floating without the COM voltage you show in your added full circuit. An opamp input must NEVER be floating.

• I am confused. Can you please explain your answer in more detail? "DC bias voltage that is at half the supply voltage"? I'm connecting two supplies to this opamp: +10V and -10V, and the grounds of these supplies are tied to the same ground in the circuit. Sep 29 '18 at 2:41
• See edit to my question for further clarification of the components in the LPF. The plus sides of the opamps are tied together and fed to the ADC as the common mode voltage. Sep 29 '18 at 2:56

It's common to solve this problem using a transformer in the audio signal line. Most places that sell car audio equipment will also sell the audio isolating transformers. usually a device than handles two channels for use with sterio equipment.

What you propose will not work because all it does is defeat the shield of the signal cable. the amplifiers are still referenced to ground.