I'm planning to convert an ATX power supply to a bench powersupply. It came to my knowledge that the DC ground is actually connected to earth. As I understand, this might become a major issue if I were to connect an oscilloscope to a device with a earth ground.

So I opened the case of my power supply and I see that the ground is directly connected to the case. It seems the board has the DC ground connected through the screws to the case then from the case to earth.

Is there a particular reason why I should not isolate the DC ground from earth?

Some solution would be to use non conductive screws. Use a file to remove conductive material around the holes and use some non conductive washer to add some space between the board and the case.

  • 1
    \$\begingroup\$ Using an ATX PSU as a bench supply is generally a bad idea as the current these supplies can deliver can be huge and might melt your wires and cause fire when the output is shorted. So add a fuse. Also the ground connection is needed as a return path for EMI emissions so I would expect that you get EMI/noise issues when you convert it so a floating supply. Also the ground connection is needed for safety reasons, there can easily develop 100 V AC on the outputs when not grounded. Not a lot of current can flow but it is still best avoided. \$\endgroup\$ – Bimpelrekkie Dec 12 '17 at 22:20
  • \$\begingroup\$ I do not see why connecting an oscilloscope would be an issue, a scope is also mains grounded. As long as you use the ATX supply as a positive supply relative to ground there should be no issue really. But if you want to measure small signals and need to avoid noise, then you need a proper lab supply, not an ATX PSU. \$\endgroup\$ – Bimpelrekkie Dec 12 '17 at 22:23
  • \$\begingroup\$ @Bimpelrekkie Having both unit grounded make the system vulnerable to ground loops if one of the ground probe the scope was to ever touch by accident a positive voltage wire. I'm only looking to isolate the DC ground from the case so it doesn't go through the earth pin of the AC outlet. The ground wire connected to the case would remain connected to the case in case of failure. As for current limit I'm not worried much about current as the device I'll be powering are going to require at least 10-15amps. \$\endgroup\$ – Loïc Faure-Lacroix Dec 13 '17 at 7:34
  • \$\begingroup\$ if one of the ground probe the scope was to ever touch by accident a positive voltage wire. Hmm, you can simply call that a short circuit. With an unfused (at the low voltage side) ATX supply that will fry your wires. Ground loops are loops involving ground only. You will already have that with a grounded ATX supply and a scope. No need to short circuit to positive rail. 10-15A at what voltage? Does it need to be noise free? Because ATX supplies are noisy. They can be as PC components either don't care or have local regulators and decoupling. \$\endgroup\$ – Bimpelrekkie Dec 13 '17 at 8:34
  • \$\begingroup\$ Regarding un-grounding only the DC side: that might work. There will be (should be) a blue (X-rated) capacitor between AC ground and DC ground. You should not remove it as it provides a return path for high-frequency signals coming through the transformer. My advise would be to place a 1 M ohm resistor in paralel with this blue capacitor, that will prevent charge-buildup at the DC side. Without this resistor you could get 100 V AC or more at the DC side, that's not dangerous for you but it can be for sensitive equipment/components. \$\endgroup\$ – Bimpelrekkie Dec 13 '17 at 8:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.