• Resolved

OPA1642: Application Report SBOA237–August 2017 HiFi Audio Circuit Design 4.1.3. "Good" grounding

Prodigy 50 points

Replies: 4

Views: 82

Part Number: OPA1642

The a.m. design guide mentiones "good" grounding , without being more specific.

I wonder , if the following would be reasonable to apply:

In an active XO/EQ design, we use OPA1642 for Sallen Key filters.

Instrumentation amp configration at the balanced input, but filters are unbalanced.

Supply via SMPS +/-20V , followed by ferrites and LDOs TPS74... 33.. to reach +/-16V.

Another set of ferrites at the LDOs outputs in all three lines, V+/GND/V-

Three different "GNDs" forseen.

Chassis GND, which is connected to PE, Y caps of SMPS and to balanced cable shields, suspected to be potentially noisy.

Power GND is layer 2 of the pcb and connects to chassis GND via ferrite i.o. to reduce noise injection from chassis to Power GND.Power GND connects to all bypass caps near the ICs and to the GND output of the SMPS (with another ferrite in series)

Audio GND is layer 3 of the pcb and connects to power GND via ferrite. Audio GND is the GND reference for the signal path and should be as quiet as possible

With this separation of GNDs , the goal is to separate ground return currents and keep audio ground as quiet as possible.

Does this make sense or am I missing something ?

  • Hi Frank,

    We have attempted to visualize your 3 ground approach for your audio application on the white board. Generally, it looks okay but we are missing all the insights that a schematic provides. The area that wasn't clear to us is the instrumentation amplifier differential inputs whose ground is the chassis ground, and the its power supply ground return. If not set up correctly the instrumentation amplifier common-mode performance can be degraded reducing its effectiveness on noise and hum.

    I suggest having a look at the IN1650 High Common-mode Line Receivers applications information, Section 8.1. It has some helpful information about ground loop noise reduction.


    The ferrite inductances that you are placing between grounds will only be effective within certain frequency ranges for which they are optimized. The ferrite material would need to be effective at frequencies such as those generated by the SMPS switching.

    Regards, Thomas
    Precision Amplifiers Applications Engineering
  • In reply to Thomas Kuehl:

    Dear Thomas,

    thanks for the quick reply and for pointing to the INA datasheet.

    In our schematic, the input EMI capacitors connect to chassis GND (HF-path) i.o. not to "contaminate"  audio GND.

    Input biasing resistors connect to audio GND.

    When considering

    a) Chassis GND to be "contaminated" by HF from cable shields and possibly by noise on PE,

    b) Power GND from SMPS to be "contaminated" by residues of switching frequency

    the goal is to keep audio GND separated from both. As much as reasonably possible:

    by executing these GNDs as separate GND layers on pcb and by separating the GND layers with effective ferrites.

    Considering the OPamps ground return path (dual supply, +/-15V):

    I assume, the GND return of an Opamp output in a Sallen Kay filter scheme is audio GND.

    So, the return current would run from audio GND via ferrite to the power GND plane and back to SMPS.

    "Contaminated" chassis GND would be kept out of the direct current return path.

    Would you suspect issues?




  • In reply to Frank Brenner:

    Hi Frank,

    Most of my experience with ground issues has been associated with minimizing EMI problems. In those applications, and I believe for audio applications as well, the goal is to make the ground impedance as close to zero as possible i.e. ZGND = RGND + jωLGND = 0 Ω. In a zero impedance system interfering voltage sources and their fields can't be supported. However, I don't think that is always achieved, or as straightforward to implement as one might think. That is why different grounding solutions have been developed, and applied with varying degrees of success.

    I understand your reasoning for separating the grounds as you have. The inductances are intended to isolate the grounds from each other localizing the different noise sources to their particular ground while maintaining them at the same ground potential. Like anything of this complexity the reasoning behind the practices and the proof of performance will be revealed once the actual system is evaluated. Nothing comes to mind that I think you haven't considered or addressed with your solutions.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

  • In reply to Thomas Kuehl:

    Thomas, thanks