Arduino Uno, USB port case is under electricity?

I noticed the strangest thing today.

I had my hand leaning on the central heating radiator (which is grounded) and touched the metal case of the USB connector on the Arduino...my hand was slightly zapped. Not painful, but slight burning sensation if i kept touching the corners. First reaction...something is terribly wrong.

Situation on the arduino is as following:

An externally powered temperature probe has the + and - lead connected in the Analog 0 and GND pin on the arduino. The arduino is being powered by an external 12v Switching power supply that does not have the ground line (only hot/neutral). There are some additional digital out pins connected to a breadboard, but it's a dead end since there's no electricity there.

I did some tests. The multimeter reads 0,5 V between the USB metal case and the radiator. If i power from the USB port there is no issue. Here's the main culprit...if i disconnect the temp sensor from Analog 0 AND GND, the issue is solved. If either of the two is connected the problem persists. If powering from usb there is also no issue at all.

During operation the analog input receives a healthy 2,2 V. If i disconnect the probe power supply it stops zapping. This arduino did however experience some abuse, but strangely everything works, or at least i can't find anything wrong.

I can't even begin to think what would be the cause. Any ideas?

To what is the arduino USB case connected to?

Thanks for any comments.

P.S. Slight update...i unplugged the arduino 12v supply...but the case still zaps....even worse actually.

If you look at the schematic http://arduino.cc/en/uploads/Main/Arduino_Uno_Rev3-schematic.pdf at the lower left you can see the case is separated from Gnd by inductor L1, BLM21. Digikey shows this as a ferrite bead with numerous values available. http://www.digikey.com/product-search/en?x=0&y=0&lang=en&site=us&keywords=blm21

The FTDI datasheet, used in the Duemilanove, does not show the shield being connected to anything http://www.ftdichip.com/Support/Documents/DataSheets/ICs/DS_FT232R.pdf nor does the pdf schematic in this example design file http://www.ftdichip.com/Support/Documents/Schematics/FT232R/USB232R1.zip

One would expect the shield to be grounded only at the PC end to prevent ground loops. In my designs with MIKROE483 modules for USB interface, the shield is connected to Gnd: http://www.mouser.com/ds/2/272/usb_uart_manual_v100-32094.pdf

Why did Arduino put an ferrite chip there? I don't know.

Hmm. So i'm guessing this is normal?

I have a switching power supply that 'leaks' high voltage into the lower voltage, it builds up static electricity. And I have a TV decoder that does the same. You can not use a multimeter, if the multimeter is for example 10M resistance, the 10M will lower the voltage.

I think that your externally powered temperature probe introduces some kind of static electricity into the circuit. In the same way as my power supply with the 'leak'.

No, that is not normal. My TV decoder is a cheap and bad, and that switching power supply must have a design error inside.

Hmm. I definitely have to try another one to see if this might be the case. Might be seriously dangerous for the board if this is the case. I can confirm that the supply i bought does make the arduino input seriously fluctuate, so it's quite possible. Thanks.

Changed the power supply. I'm not getting zapped anymore, however the multimeter still gives a readout, albeit somewhat lower.

I would agree that it is some kind of power supply issue, and voltage leaking trough.

A good computer won't do that, I never had trouble with a computer. But a small switching power supply with a small transformer and perhaps some dust and dirt on the pcb, they might 'leak' something.

A readout on your multimeter is normal. The 50Hz/60Hz mains noise introduces voltages, and if you measure between two (not connected) switching power supplies, I'm sure you can measure a few voltages.

Perhaps rotating the plug in the mains 180 degrees will help ?

I can almost 99% confirm that it is a power supply thing. I found a good quality one and plugged it up. The leak is practically gone. The multimeter does read sporadically from 0.02 to 0.1 volts, which is as you said probably due to noise. But everything seems ok now. The cheap power supply really shouldn't be used for anything more than running toys. Rotating the plug...might try it.

Thanks for the help.

I’m glad I could help. Perhaps I should get rid of my bad power supply and tv decoder… better safe than sorry.

I told to myself i got a good deal when i bought the adapter. We have a saying here..."i'm not rich enough to buy cheap stuff". :) Thanks again. Finally my project is going along nicely.