4
\$\begingroup\$

I build a kind of audio amplifier (it's actually a mixer). The problem is that the output gives a high frequency pitch noisy sound as soon as I touch the front panel which is part of the chassis. It may be capacitive coupling?

This problem only occurs if a source device is connected which is not connected to mains earth ground in any way, for example a turntable. As soon as I connect a digital media player, which has a 3-prong mains cable, the high pitch sound is gone and the chassis does not seem to react to my body near the chassis anymore.

So how is grounding done right with a SMPS which only has floating output?

I tried to draw a simplified diagram:

Simplified diagram

What I tried already:

a) DC socket bond to chassis, SMPS ground and DC/DC converter ground connected

b) DC socket bond to chassis, SPMS ground and DC/DC converter ground isolated

c) DC socket isolated from chassis, SMPS ground and DC/DC converter ground isolated, DC/DC ground connected to chassis directly

d) DC socket isolated from chassis, SMPS ground and DC/DC converter ground isolated, DC/DC ground connected to chassis via RC network (100ohm, 100nF)

No option really gives a satisfying result. The high pitch is either always there or just when my body get's close to the chassis.

Happy to hear any suggestions!


(Additional information added after the first answer was written)

I analyzed the problem further and the noise I am getting does not seem to by only high frequency. In fact it is in the full audio range, with a big peak at 50 Hz (mains frequency)

Here are two distortion spectrums: earth grounded (top) and floating (bottom):

grounded

floating

This is more than 15 dB more noise when floating. I am sure there has to be a way to solve this issue, as there are so many devices which are powered by regular floating desktop SMPS.

\$\endgroup\$
3
  • 1
    \$\begingroup\$ If the chassis itself isn't grounded to an actual ground, then no manner of tweaking inside will do much difference. Ideally you would have a safety ground there but I guess that's not an option. \$\endgroup\$
    – Lundin
    Nov 14, 2023 at 10:37
  • \$\begingroup\$ linsensuppe0815 - You wrote an "answer" but it was responding to an answer, which is the wrong process here. || Since you asked the question, please don't use the box labeled "Your Answer" below unless you are writing the full & final answer to your own question (i.e. unless you don't need further help & you solved the problem yourself using a different solution than any other answer). Instead, to add more info, please edit the question. (Make sure not to invalidate existing answers or extend the question.) Or write a comment to respond to an answer (e.g. to ask for clarification). \$\endgroup\$
    – SamGibson
    Nov 16, 2023 at 16:07
  • \$\begingroup\$ (continued) I moved your post into comments and in this case, as the new info was IMHO important for other people who may want to answer the question, I copied it into the question too, explaining it was added after the first answer was written. || FYI If you want to add image links in a comment, you can use this technique. Note that the draft answer (or question) must not be submitted. The draft answer (or question) is used only to upload the image & is then discarded. Thanks. \$\endgroup\$
    – SamGibson
    Nov 16, 2023 at 16:11

2 Answers 2

2
\$\begingroup\$

When you use a floating-DC supply, your device will literally float in the winds of EMI. If you connect more floating devices with the signal lines, then you form a large messy monster with long tentacles that picks up various EMI frequencies and cable resonances in the 10-1000 MHz range. This EMI energy has nowhere to go, so it will wobble to and fro along the arms of your monster.

Unbalanced signalling means that this interference will find its way into the audio signal and, inside the electronics, it will be rectified to become a low frequency audible signal, creating a hissing or whining sound.

When you connect a grounded device to the signal lines, you anchor one end of your wobbly tentacle monster, so it can move less. I.e. your construct will pick up fewer frequencies and the resonances will be a bit more damped, especially at the lower end of the EMI frequency range. The higher up end is slightly less of an issue, because the higher up the frequency, the more natural damping and dissipation is present.

If you do not use any earth-grounded equipment (i.e. if your circuit is not connected to a very large charge capacitance, such as the earth), you still have two options:

  1. You can still dampen the cable resonances, by using ferrite beads on the cables.
  2. You can cut your monster into pieces by using audio transformers to transmit your signals and thus break the cables. This makes the cables much worse at picking up EMI. The remaining smaller partial monsters will still pick up EMI, but at higher frequencies where the natural dissipation and damping is higher, so it becomes less of an issue.
\$\endgroup\$
5
  • \$\begingroup\$ Thank you for your answer! I analyzed the problem further and the noise I am getting does not seem to by only high frequency. In fact it is in the full audio range, with a big peak at 50 Hz (mains frequency) Here are two distortion spectrums: earth grounded and floating: grounded - floating. This is more than 15 dB more noise when floating. I am sure there has to be a way to solve this issue, as there are so many devices which which are powered by regular floating desktop SMPS. \$\endgroup\$ Nov 16, 2023 at 13:55
  • \$\begingroup\$ Can ferrite beads really help in this case? Using signal transformers is no option really. \$\endgroup\$ Nov 16, 2023 at 13:55
  • \$\begingroup\$ If you get mains hum, you should have said so. This is a different problem from high frequency noise. Much of my answer remains relevant however. 50 Hz interference is a classical issue and flows via the power supplies Y caps. Signal transformers are a classical solution. Or you could use a medical grade ACDC supply which has less capacitance to the mains. \$\endgroup\$
    – tobalt
    Nov 16, 2023 at 15:20
  • \$\begingroup\$ @tobalt - Hi, I didn't want to invalidate your answer in any way (of course), however the new distortion spectrums graphs from the OP seem too important to leave only in their comments here (which other people might not read). || As you said, mains hum should be mentioned in the question. Therefore I copied the new information into the question too, but I added that it was only provided after your answer was written, to try to avoid people thinking that it was already known when your answer was written. I hope that's OK. If you have a better approach, please feel free to edit. Thanks. \$\endgroup\$
    – SamGibson
    Nov 16, 2023 at 16:26
  • \$\begingroup\$ I'm sorry for answering my own question with additional information. I am still new to the forum. So as it seems it is a mains hum problem instead of EMC, does this make a difference to finding possible solutions? how do commercially available products solve this problem? i just measured another DJ mixer which also only has floating DC supply and there is way less of the low frequency noise. why? \$\endgroup\$ Nov 17, 2023 at 14:17
1
\$\begingroup\$

I just solved the problem. There are, which I did not know before, AC/DC power supplies which have the safety earth (FG) connected to V-. The model I am now using is the Mean Well GSM40A18 (Medical PSU). Before I used the GST40A18 (Desktop PSU), which does not connect FG to V-.

\$\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.