I have a somewhat complex circuit controlled by an Arduino, which is supplied by a self powered USB hub (earth grounded 5V supply). A part of the circuit has to be powered by 12-20V DC.

When I plug in a earth-grounded supply to provide 12+V, everything works as expected. When I provide the 12+V through a floating supply, instead, and then connect the Arduino's USB to the hub, sometimes the hub 'bricks' and I have to disconnect and reconnect it to make it work again.

Arduino's ground (hence the hub's ground) and the ground of the other supply are joined in a single ground point. I am absolutely sure I'm not exceeding any of the Arduino/power supplies/hub current ratings.

So my question is, is it possible that connecting a floating supply to an earth grounded circuit, when the former is already powered on, could cause some voltage spike/imbalance or wreak havoc in the entire circuit?

P.S. A super-simplified schematic to show the situation:


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ Can you add a small diagram of the connections you're making? \$\endgroup\$
    – clabacchio
    Oct 14, 2014 at 12:19
  • \$\begingroup\$ I have the setup you've described fairly regularly. On the handful of times the usb side has yelled at me (os x shuts down the usb port and makes you reconnect), it's been because I caused a short somewhere. \$\endgroup\$
    – kolosy
    Oct 14, 2014 at 13:13
  • \$\begingroup\$ @kolosy The first time it happened I immediately thought of a short. But then, after checking the connections and disconnecting/reconnecting the hub, I plugged the usb again and it worked, with no changes. Maybe something to do with the way I plug the usb connector? It sounds weird, though, since the first contact to connect should be ground. \$\endgroup\$
    – Cerv
    Oct 14, 2014 at 13:22
  • \$\begingroup\$ I've seen kernel log messages indicated that a USB hub shut a port down due to EMI-like electrical noise on it, so I do believe this is possible. Try establishing a common ground before connecting the USB cable, but in general be very careful using USB between powered experimental equipment. If you continue to have trouble you might consider switching to opto-isolated serial, or an isolated ethernet or wifi. \$\endgroup\$ Oct 14, 2014 at 15:56

2 Answers 2


From my knowledge:

Floating power supplies can be measured with a multimeter in AC mode in regards to ground, in my case, axp power wall plugin. It will measure 130 Vac, it will give out 12 V dc, but remember this is not entirely true. Instead, you will see a sinusoidal signal on both the + and - output. This is perfectly fine, as you put the minus on the PCB ground and the plus on your Vcc, so the voltage in between those two will always measure 12V dc. Now what happens if an external grounded device, like usually USB comes into play.

You will connect a real 0V GND to the 130Vac gnd of your PCB, now if you hit the zero crossing, of course nothing is going to happen. If you hit the 130V point, you will actually shorten 130V to gnd, since there is not a lot of power behind those 130V. Most of the time it simply turn to 0V without nothing, but sometimes, USB will be reconnected or µCs will restart or or or...

So I am also curious on how other engineers found a solution to this problem, as I am constantly struggling with this.


I can't pretend this is a real answer so please don't give me a down vote if its not helpful. I had a similar situation with a Wixel (Pololu corp) device. Granted, there were even more "grounds" involved in my case, because my MCU (the wixel) was controlling some digital potentiometers that in turn controlled some audio levels, and between the amplifiers and sources connected, the ground paths were a bit messy when I'd plug the USB into the wixel. When I did, sometimes it would re-boot, sometimes it would simply recognize the USB as it was supposed to. Well a dmm showed a small amount of AC between the ground of my wixel based device and the 0V supplied at the USB port. So I tied them together with a test lead, and having jo ill results from that proceeded to plug the USB port into the wixel. Now it consistently worked without hiccups. So maybe a separate ground link will help you too.

  • \$\begingroup\$ Well, @Randy, you seem to have experienced exactly the same issue. This confirms my suspicion: as reasonable, connecting grounds before applying other voltages could be considered 'good practice', but appears to be a somewhat overlooked or taken for granted topic. \$\endgroup\$
    – Cerv
    Oct 14, 2014 at 17:56
  • \$\begingroup\$ The usb port casing/shield, were they not tied to ground? \$\endgroup\$
    – Passerby
    Oct 14, 2014 at 19:08
  • \$\begingroup\$ @Passerby Sure, the usb port casing is connected to ground. And, obviously, the usb plug frame is connected to Arduino's ground. So grounds should be connected together before other signals 'appear'. \$\endgroup\$
    – Cerv
    Oct 15, 2014 at 13:18

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.