I have an AC/DC converter (220 to 24) inside an electrical cabinet. The 0VDC is referenced to earth, like this (couldn't find a symbol for an AC/DC converter in the schematic editor so I made a dirty sketch):

enter image description here

I've noticed there's a bit of noise on the output line:

enter image description here

I assume it's not coming from the AC input side because the converter should have a big capacitor there. I assume it would suppress all this high-frequency noise? So could the noise be coming from the earth connection?

If so, can the 0 still be referenced to earth, but without receiving all the noise from it? Would a ferrite bead help? I don't know if they make them for DIN rails, though.


The voltage between the cabinet chasis and 0 output of the converter produced the same image. Measuring between two small bus bars for zero and earth created a different image though:

enter image description here

  • 2
    \$\begingroup\$ And if you connected the probe tip to 0 volts (this may appear to be a strange thing to do), what noise do you see? You still need to have the earth clip of the probe connected. \$\endgroup\$
    – Andy aka
    Mar 26, 2019 at 12:58
  • \$\begingroup\$ @Andyaka Tried it, added more info. \$\endgroup\$ Mar 26, 2019 at 13:43
  • \$\begingroup\$ And, if you connected probe tip to clip without making a ground connection but held this "magnetic loop antenna" close to where you made the original measurement, what do you see? BTW the new picture you added is not bringing anything to the party. Then, try and make the loop antenna have a smaller loop area by spiraling the earth clip wire around the probe before connecting it. Now what do you see? \$\endgroup\$
    – Andy aka
    Mar 26, 2019 at 14:55
  • \$\begingroup\$ Related Q&A that shows how your o-scope probe can act like a loop antenna (amongst other things). \$\endgroup\$
    – Andy aka
    Mar 26, 2019 at 15:03

2 Answers 2


All power supplies (AC mains to DC converters) generate noise at their output due to the fact that they use switching circuits inside to do the power conversion.

Some power supplies output less noise than others.

The large smoothing capacitor at the AC side of the power supply circuit does remove most of the noise of the AC mains but it will not and cannot remove all of it. Large capacitors always have inductive behavior above a certain frequency so filtering only happens up to a certain frequency.

Some power supplies have additional filtering at the AC side using capacitors and/or common mode chokes. This also helps to decrease noise being injected into the AC mains by the power supply!

The way how you measure can also have a significant influence on the amount of noise you measure. Is the power supply loaded or not? That might have an influence on how much noise the supply puts out.

How are you probing, via the ground clip on the scope probe? Note that the scope also has a mains earth connection which may form a loop and pickup noise.

Of course the design of the power supply matters, if it does not have a class Y capacitor in the circuit, the high frequency switching noise might "loop around" via the ground connection instead of taking the shortcut that the capacitor should have provided.


Switching supplies have 200 volt in 200 nanosecond slewrates, approximately.

Heatsinks, or the metal tabs on power transistors, couple electric fields to all surrounding structures, and induce currents that seek return paths for the charge. This is what you are seeing. Set the scope on 10uS or 1uS per division, to learn the actual waveforms.

These waveforms are that 1 volt per nanosecond risetime waveform, pushing charges everywhere, with some of the paths being quite resonant and ringing.

  • \$\begingroup\$ The top image is 1us/div. Did you mean 1ns/div? \$\endgroup\$ Mar 27, 2019 at 8:18
  • \$\begingroup\$ Just ONE acquisition? or an average of several? Then go for 50 nanoSeconds per division. We need to examine the ringing frequency, the ringing waveforms. \$\endgroup\$ Mar 27, 2019 at 11:24
  • \$\begingroup\$ An average of several, you're right. I have a more urgent issue right now. Hopefully I'll have time to come back to this tomorrow or next week, with your and Andyaka's comments about the probe in mind. \$\endgroup\$ Mar 28, 2019 at 11:48

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.