Size and Cost Reduction of the Storage Capacitor in AC/DC Converters under Hold-up Time Requirements A. Lkaro, A. Barrado, J. Pleite, R. VAzquez, J. Vazquez, E. Olias. Universidad Carlos 111de Madrid Departamento de Tecnologia Electrhica Grupo de Sistemas Electrhicos de Potencia Avda. Universidad, 30; 2891 1, LeganBs, Madrid, SPAIN Tel.: +34-9 1-6249428; FAX:+ 34-9 1-6249430 E-mail:
[email protected]
Abstract -When a designer is looking for the best option to implement an AC/DC converter which complies with the customer requirements (size, cost, regulations compliance, etc.), the features and capabilities of each alternative topology must he checked. In AC/DC converters under hold-up time requirement, the size and cost of the storage capacitor is one of the decisive aspects, mainly in low power applications. Is a 400 V , storage capacitor voltage always the best option?, or some other voltage values can provide similar or even better results?. In this paper, the main factors that determine the size and cost of the storage capacitor will be revised. Besides, some design and selection criteria will be provided. I. INTRODUCTION
In any ACIDC converter, due to the fact that a low power level is obtained from the input when line voltage is near to 0 Volts, it is necessary to include an energy-storage element in order to feed the output during this interval. Moreover, in many computer-related applications, additional energy must be stored because the power supply have to maintain its output voltage after a dropout of the line voltage during a time known as hold-up time. The hold-up time is used to terminate orderly the operation of a computer andor to overcome the switching time of an off-line UPS. In most of cases, the energy-storage element is an aluminum electrolytic capacitor. When a designer is looking for the best option to implement an ACDC converter which complies with the customer requirements, the size, cost and the operation voltage of the storage capacitor have a high impact on the feasibility of a solution proposed to implement the AC/DC converter [ 1-31. Some aspects are the following: 1) In those applications where size and cost are critical, the good performance of the topology (good efficiency, regulations compliance, single switch and controller) may be masked by the cost and the size of the storage-capacitor. 2) The manufacturing complexity is translated into an increased cost. Taller components such as bulk capacitors result in extra mounting operation in the production line. 3) A high value of the storage-capacitor voltage could involve the use of high-voltage switches which present a lower conduction performance.
0-7803-7754-0/03/$17.00 02003 IEEE
In order to make an adequate selection of the topology and to obtain the optimization of any converter, the efforts should not be only focused on the optimization of a single parameter but all the significant aspects should be taken into account. However, this paper deals with the revision of the key factors which determine the size and cost of the storage capacitor in order to include its optimization in the selection and design process. The second aim of this paper is to answer the following question: “Is a 400 VDc storage capacitor voltage always the best option? Or are there some alternatives?”. 11. SUMMARY OF THE KEY FACTORS ON STORAGE CAPACITOR SEE & COST
The size of the storage capacitor depends on two main types of factors, those related to the manufacturingprocess of the electrolytic capacitors and those which depend on the topology and design selected as well as the hold-up time requirement. In Fig. 1 a scheme which relates all the aspect analyzed in this paper has been represented. Main Dependence
f
f J*h e w
U
c
Manufachuing
{
Outputpower Hold-upThne
* The needed capacitance value to store a
given energy and the capacitor size for a + given capacitance present opposite trends. Low number ofcommenial whes for capacitance,rated voltape and case sizes.
,_-
_____ ____ __: ___ -;
~xpioitation
1____ I
Cost & size relatiorshtp.
Fig. 1. Key factors on the storage-capacitor size & cost. 111. MANUFACTLJRMG ASPECTS OF ELECTROLYTIC CAPACITORS
Although more than 20 commercial series of electrolytic capacitor have been investigated among different manufacturers and different technologies (general purpose,
1959
low impedance, wide temperature range, axial and radial types...), in all of these series, the conclusions have been identical. Due to the space constraint, just a few of these series will be used to describe the more significant aspects.
A. Key manufacturing-dependent factors of electrolytic capacitor. These are the following:
1) It is well known that the capacitor size is increased both with the capacitance value and the rated voltage. However, this last two relationships present opposite trends when the same energy must be stored. These trends are shown in Fig.2.: Although more micro Farads must be used at low voltage to store the same quantity of energy (see Fig.2.a), low voltage capacitors presents a lower size (see Fig.2.b). The combination of these trends provides the final relationship between the volume or size of the capacitor and the voltage value in which the energy is stored, (see Fig.2.c).
Fig. 2. If the same energy have to be stored, the minimum required value of capacitanceand the size of the capacitor present opposite trends with the rated voltage.
2) The commercial rated voltage and capacitance values vary into discrete steps not in a continuous way. Thus, once the designer obtains the theoretical values for the required capacitance and rated voltage and goes to the commercial datasheets, the immediate upper values could be quite far. In this case, the exploitation of the capacitor is not good and its size is not the smallest. A practical recommendation that allows obtaining up to a 30% of size saving it is to attempt to get theoretical values as close as possible to the commercial values. 3) If the number of commercial values of capacitance and voltage rating are not so high, the number of capacitor cases provided by the manufacturers are still lower. Therefore, for the same case size, the capacitor with the higher capacitance value will present a higher exploitation and therefore, for the same stored energy, its volume will be lower. Fig.3 shows an excerpt of the datasheets of the electrolytic capacitors SREA series from SURGE Components [4]. The circles on Fig.3.a shown three examples of this previous aspect. Thus, for a 63 V rated voltage, the capacitor of 33 pF
from Surge Components[4].
and 47 pF are built using the same case. The use of 33 pF capacitors always involves an extra size. 4) Sometimes, all the capacitors corresponding to the same rated voltage present a better exploitation than the corresponding to another rated voltage. An example it is shown in Fig.3.b where the values inside the rectangle are showing that the 400 V capacitors uses the same case size that the equivalent ones for 450 V. Therefore, for this commercial series, the 400 V capacitors are poor exploited and its use involves an extra size. B. Cost-size relationship. Another interesting point is to know if the cost of an electrolytic capacitor is related to the capacitance value, the rated voltage andor the size. In Fig. 4 it has been presented a cost comparison for three different capacitor sizes. (Nichicon VX Series - General Purpose [5]. Cost Data have been obtained from Farnell Components[6]). As it can be seen in Fig. 4, even for the capacitors placed in
1960
V O U = 6.33 cm3
VOL3 = 10.43 cm3
Fig. 4. Cost comparison for Nichicon VX Series General Purpose [SI.
the extreme positions (marked with a rectangle: maximum capacitance - minimum rated voltage and minimum capacitance - maximum rated voltage), a quite similar cost is obtained if the same size is considered. However if the size is increased, (compare the cost for 0.8 cm3, 6.33 cm3 and 10.43 cm3) the cost increases as well. Therefore it can be concluded that the cost is not closely related with capacitance or with the rated voltage. On the contrary, the cost is closely related to the size of the capacitor. IV.HOLD-UP TIME & MINIMUMCAPACITANCE VALUE
The design factors presented in Fig.1 are linked by means of (1).
Where: CMNis the minimum capacitance value to provide the required energy during the hold-up time (HUT). The higher the required energy, the higher the capacitance value and the storage-capacitorsize. Po-FL is the output power at full load and r)Di& is the efficiency of the “discharge converter”. This converter is a part of total converter and it discharge the storage capacitor to feed the output after a line voltage dropout. This converter operates as a DCDC converter. VNOMis the nominal or design value for the storagecapacitor voltage under steady-state operation. VF is the final voltage value (or the minimum voltage value) reached on the storage capacitor when the discharge converter goes out of regulation by the saturation of its duty cycle.
Expression (2) has been applied to four different AC/DC converters and the obtained results have been presented in Fig.5. These results show how each one of the proposed solutions get a different nominal voltage and voltage swing on the storage capacitor. The results shown in Fig.5 have been obtained using the Panasonic M Series of electrolytic capacitors [7]. The first alternative analyzed is a two stage approach with Boost PFC front-end. The rest of converters studied are single-stage PFC ACDC converters which have been classified into three different groups. Note that these groups are not referred to a concrete topology, but each group obtains similar results concerning the storage-capacitor voltage. The storage-capacitor voltage results to be line and load dependant into the first group of single-stage converters, [8], [9]. The second group obtains a voltage on storage capacitor clamped to the peak value of line voltage, [IO]-[13]. Finally, the last family of converters to compare with present both, a low voltage range as well as a low voltage swing on its storage-capacitor. This solution has been proposed in [ 141. In Fig.5 it is shown, how the best results correspond to the two-stages approach due to the fixed storage capacitor voltage. Between the three single-stage solutions analyzed, the converter proposed in [I41 present a lower volume for the storage capacitor for the three different energies (PO x HUT) considered. The reduced voltage swing of VNOMin this solution reduces the size in spite of the low voltage reached, 20 + 31 V.
v. NOMINAL STORAGE-CAPACITOR VOLTAGE & VOLTAGE SWING In a two stages approach, the storage-capacitor voltage is regulated by means of the control loop of the PFC front-end, so this voltage is fixed and the size is lower than in case of a voltage swing on storage capacitor. However in single-stage converters this voltage is not regulated and it could varies with the line voltage and also with the load. Therefore (1) must be adapted in order to take into account this voltage swing, (2) is the rewritten equation. The voltage on storage capacitor at low line ( V N O M Bline) L~~ will determine the minimum value of the capacitance to comply with the required energy. On the other hand, the volume Dimeter I lcnglb (mm) voltage at high line will impose the rated voltage of the capacitor. Therefore, if the line variation is transferred to the Fig. 5. Storage capacitor size comparison in four ACDC converters.. storage capacitor voltage (and even increased if this voltage is load dependent), it will required both, high capacitance and The voltage swing is, perhaps, the main factor to high rated voltage, so the size of the capacitor will seriously determine the size of the storage capacitor. If the solution penalized. gets that the input-voltage range is not transfered to the storage-capacitor voltage, the size of this capacitor will be significantly reduced. As it can be seen in Fig.5 the voltage
1961
swing present a higher influence than the concrete value of VNOM. VI.
FINALSTORAGE-CAPACITOR VOLTAGE
The hold-up time supplied by the converter depends on both, the energy stored in the instant of the line fault (related to VNoM)and the energy which the discharge converter is able to extract from the storage capacitor (related to VF). The voltage VF is reached when the duty cycle takes its maximum value D-, therefore VFdepends on the regulation capability of the discharge converter. The main factors which have influence on the regulation capability of the discharge converter are the topology, the selected design and the conduction mode of the converter inductor, DCM (Discontinuous Conduction Mode) and CCM (Continuous Conduction Mode). These factors will be selected in order to improve the converter as a whole, not only to improve the hold-up time performance, however, it has been considered interesting to analyze how these aspects have an effect on the voltage VF and therefore on the size of the storage capacitor. In order to determine the influence of each previous factor, a sweep of different designs will be done over three basic topologies, which could appear as discharge converters. The different designs could be expressed in terms of the transformer turns ratio, n (l:n), for CCM operation or in terms of the dimensionless load parameter, K defined in (3), for DCM operation. (3) However in order to show in the same plot the results for both conduction modes, the variable selected to make the sweep is the nominal duty cycle, h O M Thus, . the parameters K and n must take those values which make the converter to comply with the conversion ratio, M defined in (4), for the desired DNOM. (4)
,=%!E VNOM
In (5) it has been defined a new parameter, 6. This adimensinal parameter allows to obtain the regulation capability of the “discharge-converter”, because it relates the voltages VF and V N ~Therefore, ~ . (1) can be rewritten in (6). (5)
‘&UN
=
2.Energy -
VNOM2 - v,’
2.Energy VNOM2.(l- 62)
Furthermore, the percentage of the extracted energy with respect to the total stored energy is given by (7). Extracted Energy (7) (%) = 100.(1 - a2) Total stored Energy
Since the voltage terms in (1) appear squared, if 6 5 0.3, in (7), it can be deduced that just a 10% of the total energy stored in the storage capacitor can not be extracted. In Table 1 are listed the value of 6 for each topology and conduction mode. In case of DCM the boundary duty cycle is also listed. This value means that, for a given M value, a > DBOUNDARY) does not higher nominal duty cycle (hOM guarantee the DCM operation along the complete hold-up time. In Fig.6 it has been plotted the parameter 6 as a function of the nominal duty cycle, DNOM, for two values of M. However, the same trends and conclusions are obtained for any other value of M. These plots allow comparing the regulation capability of the different topologies, designs, and conduction modes. TABLE1: EXPRESSIONSFOR THE DESIGN SWEEP CORRESPONDINGTO THRFE POSSIBLE TOPOLOGIESOF THE DISCHARGE CONVERTERAND ITS OPERATION MODE.
I
TOPO~OQY of the dischame converter
I
ROOST
I
FLYBACK
I
FORWARD
I
Since during the hold-up time process, the duty cycle of any converters is always getting higher, only three combinations of conduction modes are possible: - A converter designed to DCM operation that maintain the DCM operation along the complete hold-up time. - A converter designed to DCM operation that change to CCM operation before the hold-up time finished. - A converter designed to CCM operation, which never falls into DCM operation due to the increase of the duty cycle during the hold-up time. In Fig.6, the traces corresponding to DCM reach a saturation level equal to the value obtained for the same converter when it operates in CCM. Therefore, if the transition between DCM to CCM occurs during the hold-up time process, the final voltage on storage capacitor is fixed by means of the CCM conversion ratio for the maximum duty cycle, D-. Some considerations could be deduced from the plots of Fig.6: 1) The parameter 6 of non-isolated converters depends strongly of M, just in case of DCM operation and mainly in the Boost topology, the selection of and its corresponding load parameter K permits to reduce the value of 6. 2) When Flyback and Forward converter operates in CCM, its turns ratio, n, can be always selected to compensate the M value, therefore, the corresponding traces does not depend on M (see also Table I). In this operation mode, for any DNOM, the Flyback converter always produces better results than the Forward converter.
1962
3) An optimized design of CCM-Flyback converter correspondingto a 4.10r0.5 [15-16] presents a 6 = 0.1 1 and therefore the 98.7 % of the total stored energy can be extracted. Optimized Forward converters require h O above ~ 0.6 [15-161 which reduce this percentage to a 64 %. 4) The option DCM-Flyback is also interesting because with an usual design it can extract the most of the energy from the storage capacitor (6 I 0 . 3 ) , even in case of DCM to CCM transition during hold-up time operation.
A. Topology and conduction mode comparison.
The influence of the different topologies and conduction modes, can be observed in Fig.7. Plots in Fig.7 have been obtained for two nominal duty cycles (DNOM = 0.2 in Fig.7.a and DNOM = 0.4 in Fig.7.b) and the same output voltage conditions, Vo = 48 V. In these plots, the following aspects are shown: - The DCM Forward converter, always obtain the higher volume. - The DCM Flyback converter reaches the trace of total discharge (VF= 0) above a nominal voltage of 35 V. - The CCM Forward converter presents a volume similar to the obtained by the DCM Flyback converter when &OM = 0.2. When b o M = 0.4 the volume of the CCM Forward converter is higher than the corresponding to the DCM 35 V, Flyback converter, for some particular values of VNOM: 100 V and 250V. However, it should be notice that both values of hoM are too reduced in order to get a forward topology optimized. - The CCM Flyback converter always obtains the results corresponding to the total discharge. As conclusion, Flyback converter present a reduced storage capacitor size by means of its superior regulation capability. ..
DNOM Fig. 6. Parameter 6 as a function of nominal duty cycle for different topologies and conduction modes.
VII. SIZEvs. NOMINALSTORAGE-CAPACITOR VOLTAGE
At this point, the influence of all the factors listed in Fig.1 has been discussed. Therefore it is the moment to try to comply with the second goal of this paper. This was to answer the next question: It is a 400 V storage-capacitor voltage always the best option in order to reduce the size and cost of the storage capacitor?. In order to answer this question, some plots which represent the capacitor volume versus the nominal voltage (VNOM) will be developed. These plots will take into account the topology, the conduction mode and some possible designs. The plots shown in Fig.7 and Fig.8 have been calculated with the SREA Series of electrolytic capacitors [4]. Furthermore, to trace these plots, the minimum volume obtained among the different combinations of identical capacitors connected in parallel, has been considered.
8.3
10
16
25
50 63 1M 160 200 NominalVoltage (V)
35
250
350 400 450
Fig. 7. Size vs. nominal storage-capacitorvoltage plots. V0=48V,Hold-up time = 1Oms, PO= 100 W. a) DNOM = 0.2, b) DNOM = 0.4.
B. Nominal voltage comparison. When the influence of the topology, conduction mode and design can be neglected, (see the trace correspondingto DNOM = 0.45 in Fig.8.a, and the traces for DNOM = 0.05 and 0.2 in Fig.8.b), it can be seen how above a nominal voltage around 35 V the volume obtained are quite similar. At first sight, the lower size should correspond to the higher voltage, because the capacitance value has been
1963
reduced, however, this first idea has not taken into account the manufacturing aspects exposed in section 111. Above a nominal voltage around 35, there is not a value which drastically reduces the size of the storage capacitor.
Thus, VNoM= 100 V or 250 V or 350 V produced similar results in comparison with VN,, = 450. Furthermore, 400 V provide similar results to 50 or 63 V. The level of exploitation of the capacitor has a significant weight above these 35 V. As it can be seen in Fig. 3, the capacitors with a rated voltage of 400 V are poorly size-optimized in comparison with the 450 V ones. Because the manufacturer has used the same case for both type of capacitors. This fact is reflected as well in Fig. 7 and 8, where 400V involves a higher volume than 250,350 and 450 V.
-
-5
3) To improve the exploitation factor can also help to reduce size and cost of the storage-capacitor.Two main ways are open in order to increase the exploitation: The first one is to attempt to get theoretical values as close as possible to the commercial values. The second one is to detect which values are the poor sizeoptimized within a commercial series of electrolytic capacitors and consequentlyavoid these values. 4) Flyback topology is quite interesting to be used as discharge converter due to its high regulation capability which permit extract the most of the energy stored in the bulk capacitor.
40 35
REFERENCES B. Sharifipour, J.S. Huang, P. Liao, L. Huber, M.M. Jovanovic. “Manufacturing and Cost Analysis of Power-Factor-Correction Circuits”, IEEE APEC 1998, pp. 490494. J. Zhang, M.M. Jovanovic, F.C. Lee, “Comparison Between CCM Single-Stage and Two-Stage Boost PFC Converters”, IEEE APEC 1999, pp. 335-341.
30
-,
25
0 20
5, 2
A. Fernbdez, A. Ferreres, P. Villegas, J. Sebastian, L. Alvarez “Size Comparison Between a Half Bridge Converter with an AICS and a Two-Stage Boost Converter Operating in a Narrow Input Voltage Range”, LEEE PESC 2001.
15
i
10
0
5 0
63
10
16
25
35
50
63
100
180
203
250
350 4w
450
www.surgecomponents.com
NominalVoltageM
wwwnichicon-us.com 40
35
www.famell.com
33
www.panasonic.com
25
L. Huber and M.M. Jovanovic. “Single-stage, Single Switch, Isolated Power Supply Technique with Input Current Shaping and Fast Output-Voltage Regulation for Universal Input-Voltage-Range Applications”. LEEE APEC ’97. 4. 272-280.
20 15
. 10 5
63
10
16
25
35
50
6)
IW
160
xx)
250
350 4W
J. Sebastibn, M.M. Hemando, P. Villegas, J. Diu, A. Fontin, “Input Current Shaper Based on the Series Connection of a Voltage Source and a Loss Free Resistor”, IEEE Applied Power Electronics conference (APEC) 1998. Pp. 461467.
450
Nominal V o l e g e M
Fig. 8. Size vs. nominal storage-capacitor voltage plots. Vo=48V, Hold-up time = lOms, Po= 100 W. a) DCM Flyback converter, b) CCM Flyback Converter.
0. Garcia, J.A. Cobos, R. Prieto, P. Alou, J. Uceda. “Simple ACDC Converters to Meet IEC 1000-3-2”. IEEE APEC ’00. Pp. 487493.
Finally, all the aspects studied in this work will summarize in section VIII. VIII. CONCLUSIONS
A complete discussion about the key factors to determine the size and cost of the storage capacitor in AC/DC converters under hold-up time requirements has been presented in this paper. Some significant aspects could be the following: 1) The voltage swing on the storage-capacitor voltage is a decisive aspect to reduce the size and cost of this capacitor. The higher the voltage swing, the higher the size and cost. 2) For a given voltage swing and a given level of stored energy, different voltages on storage capacitor can produce similar storage capacitor size and cost. Therefore, lower voltages than 400V and 450V can be selected achieving analogous results.
1964
N. Vbzquez, C. Hemindez, R. Cano, J. Antonio, E. Rodriguez, J. Arau, “An efficient Single-Switch Voltage Regulator”, IEEE PESC ’00.Pp. 81 1-816. Q. Zhao, F.C. Lee, j. Qian, “Single-Switch Parallel Power Factor Correction ACDC Converters with Inherent Load Current Feedback”, IEEE APEC 2002.
W. piu, W. Wu, S. Luo, W. Gu, I. Bataresh, “A Bi-flyback Converter with Low Intermediate Bus Voltage and Tight Output Voltage Regulation for Universal Input Applications”, IEEE APEC 2002. A. Uzaro, A. Barrado, J. Pleite, R. Vizquez, E. Olias, “New Family of Single-Stage PFC Converters with Series Inductance Interval”. IEEE PESC 2002.
LL. H. Dixon, ‘Control Loop Cookbook”, Power Supply Design Seminar. SEM-1100, Topic 5. Unitrode 1996. A. Barrado, J. Peite, A. Uzaro, R. Vkquez, E. Olias, “Utilization of the Power Losses Map in the Design of DCDC Converters”. IEEE PESC 1998, pp. 1543-1547.
