Vous êtes sur la page 1sur 6

11-5

DC-DC CONVERTER PLANTS AND THEIR ABILITY TO CLEAR DISTRIBUTION FUSES James A. Giancaterino Reliance ComdTec Lorain Products 1122 F Street Lorain, Ohio 44052

ABSTRACT Accidental faults in the distribution circuits of DC-DC converter plants can cause voltage dips below the acceptable operating range of the converter loads. Unlike battery plants, converter plants rely solely upon the power supplies to clear fault protection devices. Since the DC-DC converters are current limited, their ability to clear faults will be based upon the short burst of current from their output capacitors and the excess current available from the plant. As the technologies have taken the DC-DC converters to higher and higher operating frequencies, their output capacitance has decreased substantially. In the past, the larger capacitors would provide a higher momentary current when a fault was introduced. In many cases, as you will see, this is not enough. This paper will look at some of the considerations of converter plants with respect to fault clearing, as well as some practical approaches to resolving this problem. A Lorain 624L converter plant was tested for its fuse clearing capabilities with and without the additional capacitor bank. The results of this testing will be presented.

fault protection fuse, the DC bus voltage for all of the loads dips to near zero volts, causing an interruption of power to all of the loads. This in turn causes a momentary interruption in service to your customers. Since the DC-DC converter plants do not have the luxury of including a large battery on its output to instantaneously clear distribution fises, the fuse clearing must be done by the stored energy in the converter output capacitors and the maximum current limit rating of the plant. As you will see, this is not generally sufficient.
11. BACKGROUND

The evolution of the DC-DC converter, driven by advances in technology, has produced a much smaller, much lighter, and much more dense product. All of this is good; however, the reduction in physical size comes from the reduction in component size. Component size is determined primarily by heat and operating frequency. In particular, the output filtering circuit size is reduced substantially by the increase in frequency. Table 1 illustrates the relative capacitance found in three generations of converters.

I. INTRODUCTION
There is often a false sense of security following the design of a well thought out DC-DC converter plant. The converters have been sized for the maximum loads, redundancy has been included, and the metering and distribution has been tailored specifically for the application. Then unexpectedly, a fault occurs on one of the distribution circuits. In the process of clearing the
315

6,800 ufd

6 amp 20KHz Design

1,500 ufd

3 amp 1MHzDesign

150 ufd

TABLE 1. - Output Capacitance for a 24v input, 48v output Converter If these converters were configured to create a 30 amp non-redundant plant, the total output capacitance could be seen in Table 2.
0-78032034-4/94/ $4.00 1994 IEEE

11-5

Converter Type

Output Capacitance

Energy Stored

FIGURE 1 - HSB6H, 1 1/3A fuse, no Cap.

1 KHz 40,800ufd Design

47 J

KHz 7,500 ufd 20 Design

8.6 J

MHz 1,500 ufd 1 Design

1.7 J

TABLE 2. - Output Capacitance & Energy for Converter Plant As can be seen, there is a substantial decrease in available stored energy in the new, higher frequency designs. This stored energy is a key element in clearing fuses. One solution, which has been very successhl, is to add stored energy to the converter distribution bus. This is done by attaching an electrolytic capacitor assembly across the main DC distribution bus. This capacitance, connected directly across the line side of the fuses, can provide the very high momentary currents necessary for clearing distribution fuses while maintaning the bus voltage.111. 111. TESTS PERFORMED Two series of tests were conducted to evaluate the effectiveness of adding stored energy capacitors to the DC-DC converter plants. One series of tests involved looking at individual converters and their ability to clear fuses. This was done to better understand the coordination of capacitor size to converter size. The second series of tests involved a comprehensive study of a Lorain model 624L, 24 volt output, DC-DC converter plant in various configurations, considering the number of converters required to clear distribution hses with the additional capacitor assemblies.

1Ov/div

1Omsec/div

FIGURE. 1A - MZHSA3B, 1 1/3A fuse, no Cap

20v/div 1Omsec/div Four supplemental capacitor bank sizes were selected for the test, 10,000, 20,000, 30,000 and 40,000 microfarads. These capacitors were wired directly in parallel with the output of the converter. The load was wired to the capacitor, simulating the capacitors being installed near the distribution. IV. DATA

Two converters were selected for the study. A 20 KHz converter, Lorain HSA6B, and a pair of 1 MHz converters, Lorain MZHSA3B. Since it was not possible to obtain two converters with equal output ratings, two of the 3 amp MZHSA3B converters were used in parallel to equal the single 6 amp output of the HSA6B for the comparison.
A test fixture was created which included a fixed length of hook up wire, a large switch to simulate a short circuit, and a GMT type fuse holder. Each converter was subject to the exact same test set up.

Two fuse sizes, 1 1/3 and 3 amp, GMT fuses were used in the tests for comparison. The results of clearing 3 amp fuses will more dramatically demonstrate the problem. Figure 1 and 1A are fuse clearing waveforms using the 1 1/3 fuses without any supplemental capacitors. The difference in the waveforms is primarily due to the difference in the output capacitance of the two converters. Even with the 1500 microfarads of output capacitance the bus voltage of the HSA6B converter still dipped 25 volts momentarily. In comparison the M H S A 6 B converter dipped to zero volts and sustained it for 28 milliseconds.

3 16

11-5

FIGURE. 2 - HSA6B, 3 amp fuse, no Cap

FIGURE. 3 - HSA6B, 1 113A fuse, lOKufd

2vldiv 20msecldiv 20vIdiv msecldiv FIGURE 3A - MZHSA3B, 1 113A fuse, lOKufd FIGURE. 2A - MZHSA3B, 3 amp fuse, no Cap

2vldiv 20msecldiv 20vIdiv msecldiv Figures 2 and 2A demonstrate even further how the DC bus can be jeopardized using the 3 amp fuse. Figures 4 and 4A will look at how the 10,000 microfarad capacitor will aid in clearing the 3 amp fuse. Again, the larger fuse size, even with the help of the 10,000 microfarad capacitor, shows an significant voltage dip. The HSA6B dipped 5 volts, while the MZHSA3B dipped 7.5 volts. FIGURE 4 - HSA6B, 3A fuse, lOKufd

With the larger fuse, the HSB6A converter sustained near zero volts for 48 milliseconds. The MZHSA3B converters took 110 milliseconds to clear under the same conditions.

The next series of waveforms will show the results of adding the supplemental capacitors. 10,000 microfarads was added to the output of each converter and the tests were repeated.

The addition of the capacitor has reduced the voltage dip to just 3.2 volts with the HSA6B and 3.8 volts for the MZHSA3Bs.
317

5v/div 50msecldiv

11-5

FIGURE. 4A - MZHSA3B, 3A fuse, lOKufd

The converter plant was configured with up to 18 HSB30A , 30 amp converters, having their current limit set at 120%. Several configurations starting with 5 converters were tested in order to represent a variety of plant sizes. Three types of fuses were used for the study, NON30, BAN-30, and KTN-30. These are the typical fuses used in this type of plant. The fuse current ratings are the same, however, the clearing curves for the three fuses are quite different. The purpose of using the three fuse types was to determine which was best suited for the application.

5vIdiv

1Omsec/div

HSA6B

PAIR OF MZHSA3B

The stored energy capacitor banks were sized at 1.5 Farads each. Either one or two banks were used depending upon the overall capacity of the converter plant.

10,OOOufd 20,OOOufd 30,OOOufd 4O.OOOufd

2.2 1.3 1.2 I

3.2 2.0 1.6 I

2.4 1.6 1.2 I

5.0 3.5 2.0 I

TABLE 3 - Voltage dip wlwo Capacitors

Each test consisted of clearing a fault through the selected fuse. The variables were the number of converters, the resistive load on the plant and the number of capacitor banks connected. In this way it could be determined how many converters would be required to reliably clear the fuse at a given load condition. The NON type fuse was eliminated after the first round of tests. To maintain the 1.75 volt transient, with 120 amps of load and a single capacitor bank, required 14 converters (420 amps total). This can be compared to 10 converters (300 amps total) for the BAN type and 11 (330 amps total) for the KTN type. It would take 3 to 4 more converters in the same plant if the NON type fuses were to be used. A subsequent test found that by increasing the number of capacitor banks to 2, the number of converters required to clear the BAN fuse went down from 10 to 7 (2 10 amps total) and for the KTN fuse the number of converters went down from 11 to 8 (240 amps total). Due to space constraints, a third bank of capacitors was not added. Two banks was determined to be the most practical solution. Table 4 depicts a typical sheet of fuse blowing data. The objective for each sheet is to determine the minimum number of converters required to clear the fuse without allowing the bus voltage drop to exceed the 1.75 volt maximum. In the example shown in Table 2A, the fuse clearing data was taken with
318

Table 3, summarizes the remainder of the fuse clearing tests with these converters. Capacitors up to 40,000 microfarads were used.

The most significant improvements are found with the initial added capacitance. Small improvements can be appreciated with additional capacitors. It must be qualified that the converters were operating at no load prior to the short circuit, and the fuses were operated cold. Had there been other DC loads powered during the test, the fuse clearing times would have increased. Also, a fuse that has been operating under load long enough to temperature stabilize will exhibit a quicker clearing time.

The purpose of these tests was to determine the converter plant capacity necessary to clear fault protection fuses without substantially degrading the DC bus voltage. The limit of 23.75 VDC was selected as the minimum voltage that the distribution bus could tolerate. With an initial converter output set point of 25.5 VDC, this would allow a transient of only 1.75 VDC of undershoot.

11-5

Test Conditions: 120 amps load current, 2 capacitor banks (3 Farads total), BAN-30 fuse, converter current limit set at 120%

~~~~

TABLE 4 - Fuse clearing data showing Voltage Drop and Time to Clear (BAN-30)

seven different fault currents, from 130 amps to 500 amps. The fault currents were established by the sizing of the loop resistance to the fault. A fault closer to the distribution fuse, having less resistance, will demand a higher current. It is important to understand that the clearing characteristics of the fuse will vary with loop resistance. The worse case fuse clearing condition is not necessarily that with the highest fault current. Different fault currents were used to map the voltage drop and clearing time characteristics. Several data points were taken with the 5 converter configuration in order to establish a trend. Understanding of the fuse clearing trend will minimize the number of data points necessary to accurately predict the minimum number of converters required. The minimum number of converters is determined by finding the row of data, with the minimum number of converters, whos voltage drop does not exceed 1.75 volts. As can be seen from the data, 7 converters will be necessary to clear the fuse without dropping below the minimum bus voltage requirement. Table 5 summarizes the results for the 624L converter plant testing using the BAN-30 fuse and two capacitor banks.

Similarly Table 4 summarizes the results of the testing using the KTN-30 fuse and two capacitor banks.

120

3 60 420
440
16

TABLE 6 - KTN 30

V. ANALYSIS OF CAPACITOR SIZING Sizing the supplemental capacitor bank is not intuitively obvious. Several factors must be considered, such as the resistance of the distribution wiring, the voltage drop that can be tolerated, the clearing time of the fuses selected, the load on the plant and much more.
To precisely determine the capacitor size would require looking at such details as inductance of the loop, fuse clearing characteristics at the specific operating voltage of the plant and the negative resistance characteristics of some of the loads. Since this information is very difficult to obtain, certain assumptions must be made.

PlantLoad(amps) 120 180

1
I

Converters Required 7 8

420 440 TABLE 5 - BAN-30

15

In the following derivation we will assume that the loop inductance is small enough to ignore, the fuse clearing times in the data sheets will be close enough, and the negative resistance loads will not be significant for very small voltage changes during the fuse clearing (typically 1 to 2 volts).

3 19

11-5

The equation I, = C dv/dt will be the starting point for the capacitance derivation. In the equation I, is the capacitor current, C is the supplemental capacitance, dv is the voltage dip allowable, and dt is the fuse clearing time (taken from the characteristic curve for the fuse). It should be noted that the fuse clearing time is not the same as the total duration of the voltage dip. The fuse clearing time is over when the transient voltage begins to rise.

C = I (dtldv) C = ((350)-(216-1 20))(.02)/ 1.78 Since we already know the fault current from the table, we do not have to calculate it using (2V1-dv)/2R IR=6x36amps, 6 converters at 36 amps each I,=120 amps load By solving the equation we find C = 2.85 Farads. This can be compared to the actual capacitance of 3 Farads in the test. There will be a small error in the solution due to the assumptions previously made. To be safe, the capacitance should be rounded up to add margin. V. CONCLUSION

If we assume that the plant load current must be maintained during the transition, the plant current, not including the capacitor current, available to help clear the fuse will be (IR-I,) where IR is the rated current for the plant and I, is the load current. Now the current, IF, to clear the fault will be equal to the capacitor current, I,, plus the excess plant current, (IR- IL). So the capacitor current can be written as 1 , = IF - (IR-IL). The average fault current will be equal to the average bus voltage divided by the loop resistance. I, can be written as IF = (V,+V2)/2R, where V, is the initial voltage, V2 is the minimum voltage, and R is the loop resistance. Since V, - V2 is equal to the voltage change dv, the equation can be written as I F = (2Vi - dv)/2R. This now establishes our final equation C = (((2V,-dv)/2R)-(IR-IL))(dt)/dv. Let us compare this equation to the results found in the 624L converter plant testing. The example will come fiom Table 2A; 6 converters, 350 amp fault current, and 1.78 volt dip. The time listed in the table includes the converter response time. We require the clearing time for the BAN-30 fuse. From the Bussman data sheet, this fuse will take 0.02 seconds to clear at 350 amps.

The danger of a mass power interruption in DC-DC converter plant distribution can be avoided with the addition of properly sized energy storage capacitors and proper fuse selection. The capacitors can deliver the extremely high momentary currents necessary to clear fuses without allowing the bus voltage to dip below the minimum allowable load requirement limits. REFERENCES (1) Glen Olson, Lorain Products Fuse Clearing Test For 624L Converter Plant Dan McMenamin, Using Capacitors to Zap Fuses, TE&M Apr. 1990 P. 58 Bussman, Electronic and Small Dimension Fuses, 1986 P. 3-3-15

320

Vous aimerez peut-être aussi