I have a system were we wish to place a USB charger outlet. The output is placed in a bench made of metal, so even if the output is isolated and low voltage, it would be a good idea to ground the common output (V-) in case of failures of the transformer isolation or the power supply shorting in a bad way. We plan to ground the metal surface of the bench, but in case of this failure, we still have the bonding of ground between the shield or casing of USB connectors (V-/shield). So a double layer of protection.

I can ground the V (-) just fine, and the device works perfectly.

If I attach the multimeter in AC mode between building GND (AC input GND) and the V(-), I read 44V AC.

This happened with my USB charger and a couple of 12V SMPSs that I have laying around from different manufacturers. It seems like this is normal behaviour for this type of devices in my installation.

What could be the reason of this? Could this be caused by a problem with my grounding scheme? (I plan to test in other installations.)

  • The USB charger its the Leviton T5632-W, but also tested an Eaton one.
  • Also tested with a generic SMPS for an LED strip, two different models with 12V output.
  • My ground seems to be all right, I have 120V between phasee and GND, and less than 1V between neutral and GND.
  • This a USA type installation, with a grounding rod near the main panel, bonded with neutral only there.

After some reading yesterday, I came to some ill informed conclusions.

  • The output of an SMPS its almost "isolated", and you can leave a floating circuit just OK. It might be safer in some cases. NEC doesn't require grounding <50V, and systems normally don't require it. Some people mention that it could be detrimental if your ground has a lot of noise or undesired voltages. But normally it is ok and it is good to have a decent 0v reference in your circuit.
  • In most USB chargers (that I tested) the V(-) and shield are bonded together.
  • USB chargers that I opened or tested don't have continuity between the AC input ground and the V (-) or V (+). They are effectively almost isolated (Did some reading about in reality the input / output of SMPS having a capacitive coupling, but this is still out of my understanding).
  • Some people recommend attaching V(-) or reference voltage of an SMPS output to ground, some other people don't. NEC doesn't recommend anything here (I think,) so it seems like both options are ok.

PD1: Just read this:

Did you measure using DC on the multimeter? You need to measure the AC voltage that might be present between output and ground. Then you could see several tens rising to over one hundred volts AC - this of course doesn't mean it's dangerous - it's just the EMI decoupling capacitors (if any) on the output. For DC measurements your meter's input impedance will likely reduce the voltage to near zero.

So that's exactly the situation. Seems like my plan of grounding the DC output its fine and safe then.

  • 3
    \$\begingroup\$ The 44 VAC is probably a phantom voltage. Your meter impedance is so high (10Meg or so is common nowadays) that it doesn't load down the output at all. Put a 10k resistor from output to GND and see that 44 VAC likely disappears. Google "Phantom Voltage", you'll find it quick \$\endgroup\$
    – Kyle B
    Commented Apr 8, 2021 at 14:59
  • 3
    \$\begingroup\$ 44V and what current? Probably microamps, equating to the series impedance of a small RF noise suppression capacitor. \$\endgroup\$
    – user16324
    Commented Apr 8, 2021 at 15:00
  • \$\begingroup\$ I'm also not sure why you think you should ground that output.... There's nothing (usually) wrong with leaving a USB floating. Alot of reasons NOT to connect it actually.. \$\endgroup\$
    – Kyle B
    Commented Apr 8, 2021 at 15:01
  • \$\begingroup\$ Does this answer your question? What does the Y capacitor in a SMPS do? \$\endgroup\$
    – winny
    Commented Apr 8, 2021 at 18:46
  • \$\begingroup\$ @KyleB maybe you can expand about the reasons against it?. There is a lot of contradictory information and people advising about grouding SMP's, so i think your reply could be very helpful for future reference. I have other very similar question were i explain why i want to do this. \$\endgroup\$
    – Dvidd
    Commented Apr 8, 2021 at 19:13

3 Answers 3


What you have is a power supply with floating output.

Since the output is floating, it is basically not referenced to anything in DC voltage terms - not even the mains earth - but there is stray capacitance of the switch mode transformer so for high frequencies there is an AC impedance between mains side and isolated side.

Even if mains neutral is connected to mains earth potential at some point, the switch mode supply basically has a bridge rectifier to rectify the mains AC into a capacitor. But since 120V AC mains live is has +/- 137V peaks compared to mains neutral, while the capacitor itself has always the rectified 137 VDC voltage over it, but the capacitor negative terminal will alternate between about 0V for positive peak on mains live and about -137V for negative peak on mains live. Therefore the mains side rectified DC has a 60Hz common mode voltage component compared to mains earth.

That 60Hz common mode component may pass on via transformer stray capacitance to isolated output side.

Also the isolation transformer is driven with a high frequency (eg. 20 kHz) 137Vpp square wave on the primary side, and high frequencies pass the transformer stray capacitance much better, the capacitance may couple a high frequency common mode on the output.

Therefore to combat the high frequency AC common mode on the output, there could be a so called Y capacitor between mains-rectified DC side and the isolated output side to reduce the AC ripple via transformer stray capacitance. The Y capacitor is therefore used for EMI reasons, and will shunt the stray current.

Even if the EMI capacitor is only few nanofarads, it also causes the output to wave along with the common mode of the input rectified mains voltage.

And so, basically, when you are seeing say 44 VAC from isolated supply output to mains earth, you are actually measuring the common mode voltage which the Y capacitor will couple from mains side to output side.

The multimeter you are using has extremely high input impedance (think in the order of 1 to 20 megaohms), so the small capacitor is able to push and pull the voltage over that impedance quite well. If you connect a smaller impedance (like a 1 kilo-ohm resistor or 100nF capacitor) from output to ground, the voltage measured will be much lower.

Sometimes power supplies that have earthed mains input, will also have earthed low voltage output, but if it is floating, the output will still most likely have a built-in EMI capacitor from the isolated output to earth, so the common mode voltage is already shunted inside the power supply to earth.


The voltage you're seeing is leakage past the isolation transformer and the EMI-suppressing Y cap in the supply. It's not harmful for a compliant supply, but the leakage can disturb some circuits due to AC coupling to the environment.

Go ahead and ground your stuff. This will shunt the the leakage and kill the AC voltage you're measuring.



There is stray capacitance between the mains voltage and the USB voltage, this is typical of unloaded SMPS and should not be worried about, if you wish to make it much lower, you can attach a high value resistor between V- to ground. But leaving it floating is fine, there is no downside unless you are doing ground referenced measurements.


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.