Noise problems on Active Speaker build


I'm currently building an active speaker and have run into a noise issue. Here's a brief overview of the components I'm using:
  • Two TPA3255 amplifiers (Chinese boards)
  • An ADAU1452, primarily as a crossover
  • Two PCM5102 to bridge the ADAU1452 and the TPA3255
  • A PCM1802 as the analog input for the ADAU1452
I'm powering the amplifiers with a 38V LLC power supply and using an isolated power supply for the digital components. [You can view the setup in the attached image.]


The main issue I'm encountering is that the setup is quite noisy. I observed that if I maintain the isolation on the power supply, I experience an intensely loud noise that has a significant swing. To address this, I connected the grounds of the isolated power supply, making it non-isolated. My assumption was that the noise might be due to the ground connection between the PCM5102A and the TPA3255. With the grounds connected, the only grounding occurs between these two components, potentially leading to the noise swing. However, this is just a hypothesis, and I'd appreciate any corrections if I'm mistaken.

This leads me to my main question: While I've seen numerous recommendations to use isolated power supplies, especially for separating components like Bluetooth modules from amplifiers, I've noticed that the isolation often gets compromised due to the common ground between the pre-amplifier and amplifier. In my case, using the isolated power supply seemed to introduce more problems. Can anyone share insights or solutions based on their experiences with similar setups?

P.S. I am an electrical engineer with access to measurement equipment. If specific measurements would help diagnose the issue, I can provide them. My expertise leans more towards digital systems, so I'm speculating that the noise problem might be related to grounding or another analog-related challenge.
Noise due to grounding issues can be a b*tch to solve, especially for a complex setup like yours. The noise could be coming from several places, including the I2S wiring. I would attack the problem using a bottoms up approach:

Start with the simplest configuration possible: just the ADAU1452 board, one DAC and a 5V battery supply. I’d use a pair of headphones with a battery powered headphone amp (the PCM5102A needs an input impedance of at least 1 Kohm) to listen to the output. Or use a known good pre/integrated amp. Then create a simple SigmaStudio project with just the sine tone generator (with phase and gain control) connected to the DAC.

If you still have noise it’s most likely a 1452/DAC configuration error or an I2s wiring problem. For more information on the latter see this LINK.

If you don’t have noise simply add one block at a time. I’d personally start by replacing the 5V battery with your power supplies (without adding the TPA3255 amps).

Getting additional on-line help will require a lot more information on your part. For staters, a detailed list (model number and/or link) of every component in your setup. Then a very detailed schematic of each and every connection is needed. And finally, one or more photos of your setup. This might sound overbearing, but it’s nearly impossible to diagnose grounding problems remotely without it. Who knows, your problem could be using non-isolated ADC RCA jacks that are mounted on an earth grounded chassis.

I included a couple of grounding articles I found helpful.


  • Grounding Article.pdf
    4.2 MB · Views: 61
  • Ground-Loops.pdf
    2.8 MB · Views: 54
Thank you for your thorough explanation. I'll be sure to go over those two PDFs you mentioned.

I conducted multiple tests today and found some interesting results:

  1. When operated on its own with direct analog input, the amplifier doesn't produce any noise.
  2. The PCM5102A doesn't generate any measurable noise when no amplifier is connected to it. In that case only one amp was connected to the LLC.
  3. Separating the digital circuits from the amplifier by powering them from a lab power supply eliminates noise.
  4. However, when both amplifiers are powered by the LLC, the noise returns, even if the digital circuits are still connected to the lab power supply.
  5. Using different power supplies for all three components (one amplifier to the LLC, another to a lab power supply, and the digital circuits also to a lab power supply) resolves the noise issue.
  6. Oddly enough, if one amplifier and the digital circuits share the LLC for power, while the other amplifier is powered by a lab supply, noise is present in the speaker connected to the LLC-powered amplifier. The other speaker remains silent, and I'm not sure why this happens.
Based on these observations, it appears that the issue likely stems from interference in the DC power supply line among the components. Initially, I thought the problem might involve current flowing through the ground pin of the PCM5102A's analog output. However, the findings from test number 4 indicate that the issue may lie elsewhere.

One thing which is still weird is that in my previous setup, I used an ADAU1701 where I connected both Amplifiers to the same LLC and didn't have those noise issues.