Littelfuse Startco PGR-5701 Ground Fault Monitor With Variable Frequency Drive Test Report by Ross George, E.I.T. Technical Sales Merv Savostianik, P.Eng. Sales Engineering Manager Mike Vangool, P.Eng. Research and Development Engineer Littelfuse Startco 3714 Kinnear Place Saskatoon, SK S7P 0A6 (306) 373-5505 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Executive Summary Littelfuse Startco has conducted witness testing of the PGR-5701 Ground-Fault Monitor. The PGR-5701 was tested for use upstream (line side) of a variablefrequency drive (VFD) to verify its ability to detect a ground fault downstream (load side) of the VFD. The test system included of a 3-phase source connected to a delta-wye transformer. The transformer-secondary neutral was connected to a 200-ohm neutral-grounding resistor. The 480-V three-phase output of the transformer was connected to a VFD which was used to drive a small motor. The transformer-toVFD connection, VFD-to-motor connection, and neutral-grounding-resistor connection were monitored using PGR-5701 Ground-Fault Monitors and PGC3082 Current Transformers. The connection diagram is shown in Fig. 1. To configure the VFD for the testing, it was necessary to remove or disconnect various phase-to-ground capacitors because they were not rated for line-to-line voltage which is present during a bolted ground fault on a resistance-grounded system. The removal of these capacitors was not in the manual included with the drive; instructions were obtained from the manufacturer’s technical support group. This test shows that a PGR-5701 at each of the CT locations is capable of detecting a ground fault on the VFD-to-motor connection. Therefore, in VFD applications it is recommended that the CT and PGR-5701 be located upstream of the drive—this connection will detect a ground fault in the supply cable to the VFD, in the VFD, and downstream of the VFD. Additional testing was also performed to evaluate the performance of the PGR5701 with ground faults of various currents at other locations in the system. 2 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Table of Contents Executive Summary ............................................................................................................ 2 Table of Contents ................................................................................................................ 3 Introduction ......................................................................................................................... 4 Background ......................................................................................................................... 5 Test Circuit...................................................................................................................... 5 Ground-Fault Apparatus ................................................................................................. 7 Equipment ..................................................................................................................... 10 Controllable Variables .................................................................................................. 11 Data ............................................................................................................................... 12 Procedure .......................................................................................................................... 12 Measurements ............................................................................................................... 12 Initial Test ..................................................................................................................... 13 Variable-Frequency Testing.......................................................................................... 14 DC Bus Fault Testing.................................................................................................... 15 Upstream Fault Testing ................................................................................................. 17 Fault Current and Line-Side Leakage ........................................................................... 19 Analysis............................................................................................................................. 19 Initial Test Results ........................................................................................................ 19 Variable-Frequency Results .......................................................................................... 20 DC Bus Fault Results .................................................................................................... 22 Upstream Fault Results ................................................................................................. 22 Fault-Current and Line-Side-Leakage Results.............................................................. 23 Limitations .................................................................................................................... 23 Conclusion ........................................................................................................................ 24 3 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca This page intentionally blank. 3 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Introduction The microprocessor-based PGR-5701 uses an external zero-sequence current transformer (CT) and selectable DFT or peak-detection algorithms to detect groundfault current. This testing was completed to verify the ability of the PGR-5701 to detect a ground fault downstream of a VFD with the CT located upstream of the drive. The tests were performed at the Littelfuse Startco facility in Saskatoon, Canada. A test circuit was devised to allow the PGR-5701 to be tested at several locations in the circuit and under various operating conditions. Testing was done with a variableresistance ground-fault placed on the drive connection to the motor, at the VFD dc bus, and at the supply transformer-secondary terminals. The tests involved varying the drive output frequency such that a frequency response for the PGR-5701 could be observed for frequencies up to 60 Hz. Frequencies above 60 Hz were not tested, as the response of the PGR-5701 peak filter is the same from 60 to above 400 Hz. As frequencies this high are uncommon in industrial applications, testing was deemed unnecessary. 4 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Background Test Circuit The test circuit was a test-bench version of a typical industrial VFD installation, with the addition of ground-fault relays. The circuit was connected as shown in Fig. 1. Figure 1. Circuit Schematic The test circuit included a three-phase variable-voltage ac input, connected to a delta-wye 240:600-V, 3-kVA step-up transformer. Utilizing the variable-voltage input, the transformer was configured for 480-Vac line-to-line at the wye secondary. See Fig.2. 5 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Figure 2. Incoming power supply The transformer neutral was connected to a 200-ohm neutral-grounding resistor (NGR) in series with a power analyzer and a root-mean-square (RMS) ammeter. This circuit passed through a PGC-3082 zero-sequence current transformer which was connected to a PGR-5701 Ground-Fault Monitor (analog output A1). See Fig.3. Figure 3. Transformer, NGR, and RMS Ammeter 6 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca The transformer secondary was connected to the VFD. The output of the VFD was connected to 100 feet of 3C#10 Teck cable, which was connected to a ½ hp three-phase motor. The circuits from the transformer to the VFD and the VFD to the motor passed through PGC-3082 zero-sequence current transformers which were connected to PGR-5701 Ground-Fault monitors (analog outputs A2 and A3). See Fig. 4. Figure 4. VFD, Teck Cable, Variable Resistor and Motor Ground-Fault Apparatus A variable-resistance ground-fault apparatus was used to place a ground-fault on the system. The schematic of the apparatus circuit is shown in Fig.5. 7 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca The apparatus consisted of a variable resistance with a 0 to 2850 ohm range connected in series with a power analyzer and an RMS ammeter. The variable resistance can be comprised of the following components to achieve desired resistance values: (5) – 470 ohm resistors (1) – 500 ohm rheostat One end of the apparatus was connected through a push-button switch to one phase of the power circuit and the other was connected to ground to simulate a ground fault of various resistances. See Figs. 5 and 6. With a line-to-ground voltage of 277 V, this apparatus allowed a ground-fault-current range from 1.385 A using only the NGR to 91 mA with all resistors connected and the rheostat set to 500 ohms. Figure 5. Ground-Fault Apparatus Schematic 8 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Figure 6. Variable Ground-Fault Resistor Ground connections for ammeters and motor chassis bond were connected to a ground point on the VFD chassis. The VFD chassis was connected to the building ground, as shown in Figs. 7 and 8. Figure 7. Grounding Diagram 9 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Figure 8. Ground Connections at VFD Chassis Equipment The equipment used during the test is listed in Table 1. Variac Current Transformers Transformer NGR Ammeters Voltage Output Recorder Power Analyzer Motor Cable Variable Frequency Drive Superior Electric Variable Autotransformer A219111-008 PGC-3082, 5:0.05-A, 82 mm ID Hammond 240:600V 3kVA Transformer Dale RH-250 250 W 100 Ohm Resistor (x2) Fluke 87, Fluke 87 V HIOKI 8808 Memory HiCorder Yokogawa WT500 Power Analyzer Westinghouse FD78 ½ HP 3-Phase Motor 100 ft 3C#10 TECK 480 V 3-Phase 20 HP Table 1. Equipment List 10 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Figure 9. Connected Test Circuit Controllable Variables In order to simulate the power system and provide valid results, the test circuit was able to simulate a variety of conditions. There were several controllable parameters which are listed in Table 2. Variables Output Frequency of VFD Ground-Fault Resistance PGR-5701 Filter Selection PGR-5701 Trip Level Fault location Settings 0 – 320 Hz (0.01 Hz increments) 0 – 2850 Ohms Fixed or Variable Frequency 5 mA – 4.95 A (5 mA increments, 1 – 99%) Upstream or Downstream of VFD, VFD dc bus, and supply-transformer termination. Table 2. Controllable Test Parameters 11 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Data In order to analyze the operation of the system, data was collected from the RMS ammeters, power analyzer, and oscilloscope as well as the PGR-5701 GroundFault Monitor trip and analog outputs. This data will verify the ability of a PGR5701 to operate correctly in a VFD application. Procedure Measurements Measurements were recorded from the two RMS ammeters and the outputs of the three PGR-5701 monitors. The PGR-5701 analog output is a 0-5 V voltage output linearly representing the 0-100% rating of CT-primary current. When connected to a PGC-3082, the analog-output scaling is 1 mV per 1 mA of measured current. Non-filtered and filtered ground-fault (EL1) and NGR (EL2) currents and voltages were measured by the power analyzer. See Figs 1 and 5. The power analyzer has an internal selectable 500 Hz low-pass filter. Application of the filter was necessary as the unfiltered signals contained large amounts of noise, making waveforms difficult to visualize. This is seen in an example ground-fault with a 1A fault current at 10-Hz fundamental frequency. See Figs. 10 and 11. 12 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Figure 10. 10 Hz 1 A Fault, wide band Figure 11. 10 Hz 1 A Fault, 500 Hz Filter From top to bottom, the waveforms are; line-to-ground voltage, ground-fault current, NGR voltage, and NGR current. Initial Test An initial test with the VFD operating at 60 Hz was completed. This test was repeated using both PGR-5701 fixed frequency (DFT) and variable frequency (peak-detection) filters. A downstream ground fault was applied to the system (point GF1 in Fig.1.) and all three PGR-5701 Ground-Fault Monitors tripped, with very similar analog-output values as shown in Table 3. Frequency Filter Setting 60 DFT 60 PEAK PGR-5701 Analog Output (mV) NGR Current (mA) Fault Current (mA) Load-Side Fault Fluke 87 EL2 EL2 NF Fluke 87 EL1 EL1 NF A1 A2 A3 326 254 316 255 251 251 243 295 247 296 280 247 294 264 268 263 Table 3. Initial Test, Load-Side Fault, 60 Hz The unfiltered currents measured by the power analyzer, EL2 NF and EL1 NF, are similar to the values of the Fluke RMS meters, while the power analyzer filtered values, EL2 and EL1, are similar to the PGR-5701 analog output values. The waveforms in Figs. 12 and 13 were captured for the filtered and unfiltered fault currents from the power analyzer. 13 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca It can be seen from the waveform captures that the fault current contains many harmonic components. With the 500 Hz filter applied, a 180-Hz ripple current can be seen superimposed over the 60-Hz drive output. This 180-Hz ripple current was observed in all subsequent tests at various output frequencies. Figure 12. 60 Hz 300 mA Fault, wide-band Filter Figure 13. 60 Hz 300 mA Fault, 500 Hz Filter Variable-Frequency Testing Tests performed using various frequencies generated by the VFD were completed using both the DFT and peak-detection filters to demonstrate PGR-5701 performance with each filter when monitoring a system with frequencies within the range of 10 to 60 Hz. In these tests, all three PGR-5701 monitors were set to trip at 5%, corresponding to a ground-fault current of 250 mA. At the selected frequencies, the fault current was increased until all three PGR-5701 units tripped; the current values at the trip point were recorded. See Table 4. 14 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Table 4. Variable-Frequency Tests Using the 200 ohm NGR, it was not possible to generate a ground-fault current large enough to trip a PGR-5701 below 20 Hz with the DFT filter selected and the pick-up set to 5% (250 mA). Using a fault current of 1.0 A it was possible to trip the PGR-5701’s with the DFT filter selected, with settings of 4% (200 mA) and 1% (50 mA) for frequencies of 20 Hz and 10 Hz, respectively. DC Bus Fault Testing The ability to detect an internal dc-bus fault was tested. The voltage of the bus without a fault applied was first observed and recorded. See Fig. 14. With no fault applied, a 180 Hz ripple voltage can be seen on the bus. This waveform was also found on the negative DC bus. 15 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Figure 14. Positive DC Bus to Ground Voltage, No Fault With a 500 mA fault applied to the negative bus (point GF2 in Fig. 1.) the following waveforms were recorded. Figure 15. Negative DC Bus 500mA fault The ripple current is in the fault as well as through the NGR, with a sizeable DC offset. Fault data is shown in Table 5. 16 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Filter PGR-5701 Analog Output Setting NGR Current (mA) Fault Current (mA) (mV) AC DC AC DC A1 A2 +DC Bus Fault PEAK 50 273 50 273 70 70 Trips with 100 mA Setting +DC Bus Fault PEAK 91 500 91 500 108 108 Trips with 150 mA Setting -DC Bus Fault PEAK 90 -497 90 -497 107 107 Trips with 150 mA Setting TEST Table 5. DC Bus Fault Data A spectrum analysis of the negative DC bus fault yielded the output shown in Figs. 16 and 17. Figure 16. Negative DC Bus Harmonics Graph Figure 17. Negative DC Bus Harmonics Data The 180 Hz component of fault current is 13.7% of the total fault current. Upstream Fault Testing This test was performed with a ground fault placed at the input of the drive (point GF3 in Fig.1.). This was done to show that the PGR-5701 (A1) monitoring the input to the drive as well as the PGR-5701 (A2) on the NGR would trip and the PGR-5701 monitoring the drive-to-motor connection would not trip. The trip 17 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca levels of A1 and A2 were set to 5% (250 mA) and A3 was set to 1% (50 mA.) Currents were recorded at the point where both A1 and A2 had functioned at each frequency. Filter Frequency Setting 60 50 40 30 20 10 PEAK PEAK PEAK PEAK PEAK PEAK PGR-5701 #2 Measuring GF Current on PGR-5701 PGR-5701 #1 Measuring NGR Current Input of Drive #3 NGR Current Fault Current A1 NGR Current Fault Current A2 A3 No Fault Outputs, Drive ON. 92 92 74 249 236 238 258 246 241 81 251 236 239 259 245 238 80 252 237 239 260 245 237 73 251 237 240 259 245 238 80 252 237 241 258 243 240 77 253 237 237 262 247 244 76 Table 6. Fault Testing at VFD Input While A3 shows values of over 50 mV, it did not trip during any of the tests. The test was then repeated with the fault located upstream of PGR-5701 A2 (point GF4 in Fig.1). PGR-5701’s A2 and A3 were set to 1% (50 mA), while A1 was set at 5% (250 mA). The ground-fault current was gradually increased until A1 tripped at each frequency. Filter Frequency Setting 60 50 40 30 20 10 PEAK PEAK PEAK PEAK PEAK PEAK PGR-5701 #1 Measuring NGR Current NGR Current Fault Current A1 No Fault Outputs, Drive ON. 92 251 238 241 252 237 239 251 236 240 252 237 238 252 237 239 253 238 239 PGR-5701 PGR-5701 #2 #3 A2 A3 92 74 95 80 91 77 94 77 90 80 91 80 92 77 Table 7. Fault testing upstream of A2 Regardless of frequency, A1 would trip while A2 and A3 would not. 18 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Fault Current and Line-Side Leakage A comparison of the line-side leakage relative to fault current was completed in order to see what affect fault current has on leakage on the line side of the drive. This test was used to simulate a fault upstream of the line feeding the drive. In this test a ground fault was placed at the transformer secondary terminals (point GF4 in Fig. 1.), with varying current and the output of the PGR-5701 (A2) on the VFD input was recorded. See Table 8. System Fault Analog Output of PGR-5701 Current (mA) Min Max 0 71 90 100 70 93 200 68 81 300 71 97 400 70 96 500 90 112 600 79 113 700 84 116 800 83 124 900 87 106 1000 85 107 Table 8. Upstream Fault Current and PGR-5701 Output The PGR-5701 was set at 50mA and in all cases did not trip. Analysis Initial Test Results The initial test proved the capability of the PGR-5701 to detect a ground fault located at the load side of a VFD, with a zero-sequence CT located at any of three different positions in the circuit. 19 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca The PGR-5701’s monitoring the line side of the VFD, the load side of the VFD and the NGR showed nearly identical outputs. These outputs were very similar to the ones measured by the power analyzer with the 500 Hz low-pass filter selected. Testing with both DFT and peak filters yielded similar values with the drive output at 60 Hz. It can be seen during the peak-filter test that the current detected by the RMS ammeters is similar to the current measured by the power analyzer with no filter selected; both include all frequencies present in the current signal. This test has proven that zero-sequence currents on the load side of a VFD are not isolated from the line side of the VFD. The zero-sequence current is equally detectable by a PGC-3082 at the NGR, the VFD line side, and the load side. Variable-Frequency Results Testing of the PGR-5701 at various fault frequencies was completed using both the DFT and peak filter selection of the PGR-5701. The results show that the bandwidth of the DFT filter is too narrow to be used in VFD applications in which the VFD operates below 50 Hz. The peak filter is a better choice in this type of application—its wider filter bandwidth allows the PGR-5701 to detect frequencies below 50 Hz. Even with the peak detection filter, frequencies below 50 Hz are attenuated. This frequency response is shown in Fig. 18. 20 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Figure 18. PGR-5701 Frequency Response, Peak Detection Filter From Fig. 18 a recommended setting for the PGR-5701 can be calculated based on the lowest expected operational frequency. This setting can be found by multiplying the desired trip level by the normalized response value at the lowest expected operational frequency. Recommended setting = (desired trip level) x (response @ lowest frequency) For instance, if a 1,000-mA RMS trip level is desired and the VFD will be operating at 40 Hz, the recommended PGR-5701 setting is 1,000 mA x 0.8 = 800 mA, or 16%. 21 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca DC Bus Fault Results When a fault was placed on the VFD DC bus, the fault current had a predominant DC component with a smaller 180 Hz AC component as well as some harmonics of 180 Hz. The 180-Hz component is created by the VFD’s full-wave rectifier “front end”. A current transformer is unable to detect the DC component of the fault; however because of the sizeable 180-Hz component, it is possible for a PGR-5701 to detect this type of fault. The DC fault spectrum analysis in Fig. 17. shows the 180 Hz amounts to 10-15% of the fault. Therefore, if it is desired to have a PGR-5701 trip on a DC-bus fault it is recommended that the pickup level be set to 10% of the NGR let-through current. (This assumes that the measured-feeder charging current is less than 10% of the NGR let-through current, or the PGR-5701 could sympathetic trip when a ground fault occurs elsewhere on the system.) Upstream Fault Results When a fault is located upstream of the monitoring PGC-3082, the PGR-5701 does not trip. This shows that the PGR-5701 does not detect an upstream fault. The PGR-5701’s which were downstream of the fault were set to their lowest setting of 1%, corresponding to a 50 mA trip level. While it was noticed that the outputs of the PGR-5701 showed levels of up to 95 mV, the relays did not trip, and this is the desired response. 22 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Fault-Current and Line-Side-Leakage Results With a ground fault located upstream, a CT measures the charging current of the feeder downstream. This test was performed to determine what affect upstreamfault magnitude has on the measured charging current. Figure 19. Line-Side Leakage vs System Fault Current As can be seen in Fig. 19, an increase in upstream fault current has only a very slight effect on the level of charging current. The increase in charging current was not substantial enough to trip a PGR-5701. Limitations As these tests were conducted under lab settings, they were created to best simulate systems in the field. However, it is impossible to create a direct analogue to applications in the field. During these tests the system was operated with only one VFD fed by a dedicated input. The ½ hp motor was unloaded, and the NGR of 200 ohms limited the current of the fault to 1.38 A. These tests were performed over the course of two days at room temperature, in an indoor lab and therefore cannot simulate all possible conditions that may be experienced in a field application over a long period of time. 23 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca Conclusion Each PGR-5701, one each upstream and downstream of the VFD and one at the NGR, measured an equal amount of VFD-frequency zero-sequence current when a ground-fault was applied on the load side of the VFD. This proves that the VFD does not isolate zero-sequence current, and that a PGR-5701 placed at the line or load side of the drive will detect the same amount of zero-sequence current. A PGR-5701 placed upstream of a VFD is capable of monitoring the feeder, VFD, and the load downstream of the CT. When a PGR-5701 is used in a variable-frequency application, use the peak (variable-frequency) filter selection. This selection has a wider pass band and less attenuation at frequencies below 60 Hz. When used in these applications, a lower trip setting for the lower frequencies can be required due to the PGR-5701 frequency response. The lowest operating frequency expected should be considered and the normalized pick-up value from Fig. 18 should be used to calculate the appropriate trip level. A DC bus fault can be detected by a PGR-5701 monitoring upstream of the VFD. The DC component is not detected, but the 180 Hz ripple current can be. The 180 Hz component is roughly 10-15% of the total fault current. Therefore, it is necessary to have the PGR-5701 set at a level of roughly 10% of the NGR letthrough current in order to detect a DC bus fault. (The reduction in setting applies at this location only. Monitors upstream and on the NGR may require a higher setting as they will measure accumulated system harmonics.) When a fault is located upstream of a monitoring CT, the current is not detected by that PGR-5701. This shows that if a fault is located on the line side of, or 24 3714 Kinnear Place Saskatoon, SK S7P 0A6 Canada www.startco.ca inside of a VFD, with a PGR-5701 connected on the load side of the drive, the fault is not detected. The effects of fault-current magnitude with a fault at the transformer terminals were examined. This showed that when a fault is upstream of the CT, such as on another drive fed by the same feeder, that the PGR-5701’s will not trip. The change in charging current was minimal and the PGR-5701’s did not trip. 25