DC Ground Fault Detector Wherein Fault Is Sensed by Noting Imbalance of Magnetic Flux in a Magnetic Core

The present invention relates to dc ground-fault detectors.

The context in which the present invention is discussed is that of dc control power systems in power generating plants. These systems are composed of batteries which are connected to provide +125 volts dc and -125 volts dc and are maintained in a fully charged condition.

Typically a system to provide control power in such a plant includes hundreds of devices fed from a multiplicity of buses involving hundreds of miles of cable. Since the power supply to the system consists of two dc sources of opposite polarity relative to ground, the effect of a failure of insulation between the applied potential and ground with respect to any device leads to a flow of ground current. If a similar insulation failure between the opposite polarity and ground also occurs, a short circuit across the full power supply could eventually evolve. Since the dc power source supplies all of the control power of the plant, the integrity of the supply is critical. Devices exist which indicate failure of the entire insulation system. However, locating the fault often involves manual isolation of the system buses and feeds individually until the cable or device that has an insulation failure has been localized. A dc ground fault detector that is inexpensive and that can be placed on each cable or group of cables feeding a portion of the plant control power system will enable the fault location to be more quickly identified, will reduce the time of exposure of the system to a potential catastrophic second ground fault of the opposite polarity, and will reduce the human hazard of identifying fault location manually while the system is energized.

Accordingly, it is a principle object of the present invention to provide a dc fault detector, one which is economical to fabricate, reliable, and easy to install.

These and still further objects are addressed hereinafter.

The foregoing objects are achieved, generally, in a ground fault detector and the like that includes a high permeability toroidal core (or the like) having a square hysteresis loop. There are three windings wound about the toroid in a solenoidal configuration. Two of the three windings have an equal number of turns and, when energized, have an electric current flow therethrough such that the magnetomotive force within the toroidal core because of one of the two windings opposes (i.e., bucks) the magnetomotive force within the toroidal core because of the other of the two windings. The two windings, in an operating system, are connected in series with a load to be monitored, the load being connected serially between the two windings. A difference between the currents in the two windings shows the existence of a ground fault, said difference being the ground current at the fault. Sensing means is provided to note any such difference in electric current flow in the two windings. The sensing means includes a voltage source and series resistor connected across the third of the three windings; the polarity of the voltage applied by the voltage source to the third winding and series resistor is reversed every time the current in the third winding reaches a predetermined amplitude. Means is provided to determine the duty cycle of the voltage applied to the third winding and to relate that duty cycle to any fault current in the portion of the system supplied through the aforementioned two windings.

The invention is hereinafter discussed with reference to the accompanying drawing in which:

FIG. 1 is a diagrammatic representation of a dc gound fault detector embodying the present invention concepts;

FIG. 2 is a top view of a toroidal magnetic core showing three windings wound thereon;

FIG. 3 is a graph of voltage v. time of a voltage source and current v. time in winding W.sub.3 in FIG. 2;

FIG. 4A and 4B are graphs of flux linkage v. current in the same winding as represented by the graphs in FIG. 3; and

FIG. 5 shows diagrammatically a portion of a dc power control system that includes a plurality of dc ground fault detectors of the type shown diagrammatically in FIG. 1.

Turning now to FIG. 1, there is shown diagrammatically at 101 a ground fault detector for sensing ground faults in a dc system. The detector 101 includes a high permeability toroidal core 1 (see FIG. 2) having a square hysteresis loop (see FIGS. 4A and 4B hereof). There are three windings W.sub.1, W.sub.2 and W.sub.3 wound about the toroid in a solenoidal configuration to form a coil and core combination 15 which is a differential transformer. Two of the windings, W.sub.1 and W.sub.2, have an equal number of turns to receive a dc applied voltage between terminals A and D. A load (in FIG. 2) to be monitored for ground faults is connected between the terminals T.sub.3 and T.sub.4. Electric current flow in the detection circuitry is, then, from the terminal A, through the winding W.sub.1 (i.e., current i.sub.1 in FIG. 1), through the load L (i.e., current i.sub.L in FIG. 2) and thence the winding W.sub.2 (i.e., current i.sub.2 in FIG. 1) to the terminal D. In the actual device, the winding W.sub.1 is wound about the winding W.sub.2 (i.e., the winding W.sub.2 is within the winding W.sub.1) in a nested arrangement. The current flow in the winding W.sub. 1 is in a direction to create a circumferential magnetomotive force (i.e., mmf.sub.1) which, when i.sub.1 equals i.sub.2, is equal and opposite (i.e., bucking) to the magnetomotive force created by the winding W.sub.2 (i.e., mmf.sub.2). (The ampere turns in the winding W.sub.1 equal the ampere turns in the winding W.sub.2.) In the device 101, any differential between mmf.sub.1 and mmf.sub.2 is noted and is related to any leakage or ground fault current in the load L. In the situation in which no such leakage occurs i.sub.1 =i.sub.2 (and i.sub.1 =i.sub.2 =1.sub.L in FIG. 2). With reference to FIG. 1, the voltage applied to the terminals A and D of the coil and core combination 15 is that applied to terminals T.sub.1 and T.sub.2 of the detector 101. Most of the remaining circuitry serves to detect the relationship between the magnitudes of the currents i.sub.1 and i.sub.2 in the windings W.sub.1 and W.sub.2, respectively, as now explained.

The relationship between the currents i.sub.1 and i.sub.2 is sensed by noting the effect thereof on the third winding W.sub.3. The differential transformer 15 is particularly adapted to function in a dc current mode. The windings W.sub.1 and W.sub.2 carry dc load current. The core 1, as previously indicated, is chosen to have high permeability and a strong nonlinear characteristic. Thus, the magnetizing effect of any unbalance between i.sub.1 and i.sub.2 is exaggerated in the core 1. The winding W.sub.3 is a low-current tickler coil.

The tickler coil W.sub.3 is connected in series with an inductance 18 and a resistor 23. The series combination is driven between the output of an inverter 17 (which performs a buffer function) and a point 10 in FIG. 1, with a variable-duty-cycle voltage square wave of the type shown at 20 in FIG. 3. When the current in the tickler coil W.sub.3 reaches a pre-set value (either of the values designated +i and -i in FIG. 3) corresponding to mild saturation of the core 1, the polarity of the square wave 20 is reversed; reversal is effected by a comparator 2 which detects threshold values +i and -i by measuring the voltage across the resistor 23 and reversing the applied voltage of FIG. 3 whenever either threshold is reached. When the core is driven in the opposite direction and the current in the tickler coil W.sub.3 reaches the pre-set value of opposite polarity the square wave 20 is reversed back to the original polarity. If there is no net dc flux from the high-current windings W.sub.1 and W.sub.2 (i.e., the current in one winding is equal to the current in the other winding), the voltage square wave 20 driving the tickler coil W.sub.3 will be symmetric. If there is a net.dc flux from the high current windings W.sub.1 and W.sub.2, resulting in saturation of the core, it will take longer for the tickler current 21 in FIG. 3 to reach the pre-set value in one direction and less time for it to reach it in the other direction. Over a limited range, the duty cycle (d.sub.1 /d.sub.1 +d.sub.2, wherein d.sub.1 is the width of positive voltage pulses 20A.sub.2, 20A.sub.4 -in FIG. 3 and d.sub.2 is the width of negative pulses 20A.sub.1, 20A.sub.3 -) of the tickler voltage 20 will be proportional to the difference in dc current in the high current windings W.sub.1 and W.sub.2, as now discussed.

The unbalance just referred to is caused by the effect of a net flux produced in the core 1 due to unequal currents in the windings W.sub.1 and W.sub.2. This shifts the flux-linkage characteristic of the core as viewed from winding W.sub.3 from that having a symmetrical characteristic as shown in FIG. 4A to one having an unsymmetrical characteristic as shown in FIG. 4B. When the symmetrically excited core is energized by the square wave voltage source, the current waveform 21 is symmetrical. However, when the unsymmetrical characteristic is energized by the source, the current tends to have a negative average value. In the absence of the resistor 23, the actual current waveform would have no average value. However, the presence of the resistor 23 reduces the voltage across winding W.sub.3 to be the difference between the zero point at 10 and the voltage drop across the resistor 23. Since the current through the resistor 23 tends to be unsymmetrical for the characteristic of FIG. 4B, the voltage and therefore the flux linkage across winding W.sub.3 is unsymmetrical. As a result, the time for the flux to drive the current from a negative peak value 21A.sub.1, 21A.sub.3 -to a positive peak value 21A.sub.2, 21A.sub.4 -is longer than the time to drive the current from the positive peak value back to the negative peak value; an opposite polarity fault will shift the flux-linkage characteristic to the right rather than to the left as in FIG. 4B.

Thus, in the event of a ground fault, the time lapse between current peaks 21A.sub.1 and 21A.sub.3 -will either increase or decrease, depending on the polarity of the fault, with a respective decrease or increase between 21A.sub.2 and 21A.sub.3 -Hence, there is a change in the duty cycle d.sub.1 /d.sub.1 +d.sub.2 as well as the ratio d.sub.1 /d.sub.2 and either can be used to detect a fault. It is noted later, however, that large fault current results in a change (here a decrease) in the frequency of the current wave 21. Some discussion of the circuit elements to achieve the foregoing now follows.

The voltage across the resistor 23, as above indicated, is sensed by the comparator 2 which determines differences in electric potential between the circuit point 10 in FIG. 1 and the midpoint shown at 16 of a voltage divider formed by resistors 22 and 26. The voltage across the resistor 23 is proportional to the current i.sub.3 ; hence that voltage can be and is used to note the thresholds +i and -i in FIG. 3. A speed-up capacitor 30 eliminates high-frequency oscillations during polarity reversal of the voltage wave 20. The output of the inverter-buffer 17 is filtered by a two-pole low pass RC filter consisting of resistors 24 and 25 and capacitors 31 and 32, to determine the dc value of the variable-duty-cycle, square wave 20. Comparators 3 and 4 compare that dc value with reference set-point voltages at C and B from a voltage divider 13. The comparators 3 and 4 are connected in an OR configuration and are used to drive two inverters or buffers 41 and 43 which drive the LED 14. If both outputs of the comparators 3 and 4 are high, the LED 14 emits green light but, if either is low, the LED 14 emits red light. Resistor 28 is a pullup resistor for the outputs of the comparators 3, 4 and 5.

If the difference in the dc current (i.e., i.sub.1 -i.sub.2) is very large, the core 1 will be heavily saturated and appear as a short circuit to the tickler signal. As the core becomes more saturated, the freguency of the tickler coil voltage waveform 20 becomes higher. To limit this frequency to the range where all the circuit components operate normally, a small inductance (i.e., the inductance 18) is inserted in series with the tickler coil.

The frequency of the square wave 20 applied to the tickler coil W.sub.3 is measured by a phase locked loop 29 (in the analyzing circuit 102) which contains a voltage controlled oscillator (VCO) 34 and phase sensitive detector 33. The input to the VCO signal at 40 is proportional to the frequency of the square wave 20; that input is compared by the comparator 5 against the set-point B. The phase-sensitive detector (PSD) 33 whose input is the voltage wave 20 from the inverter buffer 17 provides a control input to the VCO 34. A feedback signal from the VCO 34 to the PSD 33 is provided along conductor 36. The comparator 5 is also in an OR configuration with the comparators 3 and 4 and its output must also be high for the LED 14 to emit green light. In this way, frequency (or period) as well as duty cycle, is employed to denote presence or absence of ground fault.

The circuitry in FIG. 1 is powered by the voltage divider 13 having input terminals T.sub.1 and T.sub.2 ; the terminal T.sub.1 received a plus dc voltage (+125 volts in an actual device) and the terminal T.sub.2 receives a minus dc voltage (-125 volts in the actual device). The circuit point labeled 12 is a zero point and it is connected through an operational amplifier 11 which serves as a buffer whose output at 10 provides the detector ground, that is, the output 10 is the voltage level used as a standard level or magnitude against which other voltage levels in the detecting circuitry to the right of 10 in FIG. 1 are compared. The designation G.sub.1 in FIG. 1 indicates common electrical connections in the detector 101. It is not believed that the resistors and capacitors that form the voltage divider 13 need be discussed further since the functions thereof are known. Outputs from the voltage divider 13 are at A,B,C,D and v.sub.1.sup.+. The connections between the outputs A,B,-and other elements in the circuits of FIG. 1 are indicated by the corresponding designation at the input to the particular element. The level of v.sub.1.sup.+ is maintained by a zener diode 9. The two back-to-back zener diodes 6 and 7 prevent noise spikes from damaging the input to the comparator 2.

The system shown in FIG. 5 includes N detectors, like the detector 101, to monitor N loads. A red light output by any one of the N detectors indicates the presence of fault current at the portion of the system monitored by that particular detector.

Modifications of the invention herein disclosed will occur to persons skilled in the art and all such modifications are deemed to be within the scope of the invention as defined by the appended claims.