I must correct a misconception. A Ground Fault Circuit Interrupter (GFCI) does not require any ground to exist. It is simply a fact that ground is the most common conductor for fault currents causing shock to follow.
Picture an isolation transformer with the two output terminals labeled phase P1 and P2. This is connected to a GFCI. The GFCI in turn powers a product. If you touch a primary circuit within the product and P1 or P2, the current which flows will trip the GFCI. No ground involved. I don't recommend you try this because you WILL get a shock. The GFCI's purpose is to keep this shock level to nonlethal levels while it trips. (You can use a resistor instead). The easiest way to picture the situation is to draw a bubble across the GFCI enclosing all its loads. Any currents leaving this bubble and returning somehow to the circuits powering the GFCI are detected by the GFI and if high enough will trip it. Note that outside this product bubble are the third wire ground, the metal product enclosures, most secondary circuits, etc. and penetrating this bubble are EMI capacitors and other leakage capacitance and resistance. These penetrating elements are the source of nuisance tripping. Bob Johnson Georg M. Dancau wrote: > > Rich Nute wrote: > > > Hi Georg: > > > > You pose an interesting question, "Is a GFCI allowed on > > an IT system?" > > > > I will answer a different question: "Will a GFCI work > > on an IT system?" > > > > I will answer yet another question: "Does a GFCI provide > > a protective function on an IT system?" > > > > A GFCI measures the difference current between the two > > supply conductors. Under normal load conditions, the > > difference current is zero. If the difference current > > exceeds the GFCI setting, usually 5 to 20 mA, then the > > GFCI disconnects the supply. > > > > The GFCI requires that one of the two supply conductors > > (i.e., the neutral) be grounded. Or nearly grounded. > > This provides a current path AROUND the GFCI such that > > a difference current is created and thereby causes the > > GFCI to operate. > > > > The IT system has no direct connection to ground, i.e., > > the neutral is not grounded. The "I" in IT means the > > neutral is grounded through an "I"mpedance. I understand > > that the value of the impedance is 1000 ohms or greater. > > > > This impedance has the effect of setting the maximum > > difference current in the system. The maximum difference > > current for a 230-volt system would be: > > > > 230 V > > I = ----------- = 230 mA > > 1000 ohms > > > > Clearly, if the impedance is 1000 ohms, then the GFCI > > will work. > > > > If the impedance is 100,000 ohms, then the maximum > > difference current is 2.3 mA, and the GFCI will not work. > > > > So, whether or not the GFCI will work depends on the value > > of the grounding impedance of the IT system. > > > > However, if the IT system is grounded through 100,000 ohms, > > then the maximum current that can flow through a body > > connected between one pole and ground is 2.3 mA. While > > unpleasant, it is not unsafe! > > > > So, if the value of the impedance of the IT system ground > > is high enough to not allow the GFCI to operate, then the > > IT system does not reguire a GFCI to be safe! > > > > Best regards from San Diego, > > Rich > > > > Hi Rich > > Thanks for your comment. > > You're right: > So, if the value of the impedance of the IT system ground > is high enough to not allow the GFCI to operate, then the > IT system does not require a GFCI to be safe! > > According to HD384.4.41 S2:1996 (I only got copies of 2 pages) RCD > (GFCI's) are > allowed as protection. > > They are right, under following circumstances: > - each consumer (or consumer group) is connected via GFCI > - there is one fault on one consumer. > > If now a person would touch an active wire (or you would have a second > fault) > BOTH GFCI's would trip. > > I.e. GFCI's could offer a protection in case of two faults. > > Best regards > > George > -- > ************************************************************** > * Dr. Georg M. Dancau * HAUNI MASCHINENBAU AG * > * [email protected] * Manager EMC Lab * > * TEL: +49 40 7250 2102 * Kampchaussee 8..32 * > * FAX: +49 40 7250 3801 * 21027 Hamburg, Germany * > ************************************************************** > * home: Tel: +49 4122 99451 * Hauptstr. 60a * > * Fax: +49 4122 99454 * 25492 Heist, Germany * > **************************************************************
