There is very sad situation regarding leakage current. The very common and "UL approved" power lead connector socket for all PC's and displays and many other appiances such as VCRs and other electrical equipment, containes RFI filtering comprising a capacitor from hot to ground/case and a capactitor from neutral to ground. What this means is, if you lose the ground, you get a capacitive potential divider that puts 120volts/2 on to the case of the appliance. It also means there is continuous ground current from such appliances and the total for a whole household or a lab-bench can be significant. I have measured 50mA, which is well into the potentially fatal region. I heard this happened with a piece of hospital equipment conencted to a paralyzed patient via a catheter, and they didn't notice he was in agony for days. When working as an electronic engineer in Europe, where the case rises to 240/2 volts if you lose the ground, I often got a shock when disconnecting a BNC cable linking two pieces of equipment, where the ground had come off one and not the other. I know of a case too where a chip foundry used a wafer tester that operated between 0 and -5v due to the polarity of the substrates they were testing. One time, somebody wanted to connect an HP logic analyzer in to do some manual tests, and the HP analyzer operated from 0 to +5v. So they took the ground off to float it. Upon connecting the two together, the leakage current blew up a $500,000 wafer tester.

We need this connector to be replaced by a safe one. It is amazing that this issue has been hidden for so long. The correct and safe RFI circuit is one which has the capacitors connected from hot to neutral and from neutral to ground. That way you have to lose both neutral and ground before the case becomes live; and there is no ground leakage current normally. We need much tougher standards on leakage curent. I estimate that, in the US, we are wasting 50 megawatts continuously to warm the earthworms. I would say a good standard would be a current level you can't feel, and that means less than 1mA.

There was a fatal case involving the same phenomenon at high power. A group of mineworkers were asked to connect power-factor correction capacitors to the 13200 volt lines feeding the winding gear. They connected them in Y-configuration from the three hots to ground. With good phase balance, the ground current should be zero, but one of the capacitors became disconnected, shoving a bunch of imbalance current into the ground rod which only had a resistance of 180 ohms. This raised the local ground system to about 300 volts, killing a worker using an electric drill nearby.

Moral of the story is: Never connect capacitors from hot to ground: Only from hot to neutral, hot to hot or neutral to ground.