Sie sind auf Seite 1von 19

A Better Uninterruptible Power Supply (UPS) By Jason Fontaine Jon Rasmussen ECE 345 Jon Benson (T.A.

) December 7, 1999 Project #16

Abstract
This report details the design, implementation, and testing of a circuit intended to function as an uninterruptible power supply (UPS) for a personal desktop PC computer. This circuit will offer several improvements of the typical UPS found in the public market. By reducing the number of AC and DC conversions, circuit efficiency will improve. The connection to the power supply will be through a diode rather than a mechanical switch. This change will offer greater reliability and increased switching speed. Finally, the backup power will directly connected to the DC bus of the computers power supply. This also increases efficiency and reliability. The circuit consists of three main sections: the rectifier and battery charger, the gate drive, and the flyback converter. The combination of these simple components offers the performance necessary for a supply backup power supply.

Table of Contents
i. ii. iii. Chapter 1. Title Abstract Table of Contents Page Introduction 1.1 Design......4 1.2 Performance.5 Rectifier and Battery 2.1 Design..6 2.2 Performance.6 Gate Drive 3.1 Design..7 3.2 Performance.7 Flyback Converter 4.1 Design..9 4.2 Performance...10 Project Results.11 Cost Analysis...12 Project Future...13 Post-Project Conclusions.14 Appendix A Circuits.15 Appendix B Data tables....17 Appendix C Equations and Calculations..18 References19

2.

3.

4.

5. 6. 7. 8.

1. Introduction
1.1 Design In order to better appreciate the improvements of a current uninterruptible power supply (UPS), one must understand how the UPS operates. Simply, the UPS takes the AC voltage from a wall receptacle and converts it to a DC voltage. This DC voltage is applied to charge the battery. In the event of needing backup power, the battery provides a DC output voltage. This voltage is converted to an AC voltage. This is done because the computers power supply is configured to receive AC voltage from a wall receptacle. One inside the computers power supply, the backup AC voltage is converted to DC voltage for distribution within the hardware of the computer. Conventional wisdom points out that converting from DC to AC, only to convert back to DC without any alterations, is unnecessary. By tapping into the DC bus of the power supply, there is no need to convert to AC and back to DC. This direct application reduces two AC/DC conversions with the total circuit.

The means by which the backup power is applied to the power system is another area of improvement. Current UPS design involves a mechanical switch the will alternate between the AC voltage from the wall and the backup converter AC voltage. The switch is set to move when it detects a lack of power coming from the wall. Our design connects the backup power through a diode to the power supply. The advantage of the diode is power will automatically flow through the diode as soon as a power loss occurs, and there is no need to detect a power loss. This fundamental property of the diode allows for the backup power to be constantly present, effectively waiting for a loss of

5 primary power. This behavior provides backup power faster than a mechanical switch and with more reliability.

1.2 Performance The specifications for the circuit are essentially predetermined. A sample power supply was opened, and from the DC bus the required input voltage was measured to be 326 V at 240 W. This high value is due to the presence of a doubler. The doubler allows for the computer to be operated in the United States at 120 V or in Europe at 220 V. Regarding battery sizing, a medium voltage battery was needed that could output high current. It was discovered that the type of battery commonly used in motorcycles suits our application perfectly. The battery outputs 12 V at high current levels. Furthermore, this outputs requires an AC transformer with a 10:1 turns ratio. This ratio is very commonplace and inexpensive. A complete table of the specifications can be found in Appendix B.

2. Rectifier and Battery Charger


2.1 Design The first step in providing backup power involves the backup source. As afore mentioned, a 12 V 20 A battery is the heart of the power. In order for this backup supply to be reliable, it must undergo constant charging will primary power is available. Initially, this constant charging presented the threat of draining the battery over time. However, the battery implemented can withstand constant charging at a value near the output value. Therefore, the 120 V from a wall receptacle is unsuitable for charging. After a 10:1 AC transformer bring the voltage down to 12 VRMS, it is fed into a modified full bridge rectifier. A basic full bridge rectifier consists of only four diodes. The bridge implemented uses eight diodes, placing two diodes in series where previously there was only one. The additional diodes increase the voltage drop, therefore reducing the DC average to a more suitable for charging the battery.

2.2 Performance The initial design used only the basic version of the full bridge rectifier. This circuit produced a DC average of 14.5 V from the 12 VRMS. This value was perceived to be too high to properly charge the battery over a long period of time. By doubling the number of diodes as mentioned in the design, the DC average was brought down to 13.5 V. Figure A.1 in Appendix A shows the final modified rectifier/charging circuit. Although the AC signal is not perfectly rectified, the requirements of the battery do not demand such a signal. Therefore, ripple is not a concerning issue, and the rough output signal is acceptable.

3. Gate Drive
3.1 Design The gate drive circuit is fairly straightforward. It was published in the lab manual for ECE 369 and recommended by Professor Phil T. Krein. The purpose of the gate drive circuit is to power the MOSFET of the flyback converter. Originally, a 555 timer was to be used for driving the transistor. However, complications from the timers input power and output voltage led to abandoning its use. Next, a Schmidt triggered NAND circuit was implemented to drive the MOSFET. This circuit also proved futile, as this driver only allows a 50% duty cycle. By only allowing one specific duty cycle, fine-tuning the circuit is not possible, and this would drastically limit our circuit. Finally, we consulted Professor Krein and his advice steered us to using his gate drive circuit. Additionally, the output of the 369 gate drive circuit is sent through a MOSFET driver circuit. This circuit consists of a push-pull amplifier made from two BJTs. The amplifier helps maintain the integrity of the signal while the MOSFET is loaded.

3.2 Performance The gate drive circuit allows for easy adjustments in the input frequency and duty cycle. This is the major advantage of the gate drive circuit. By adjusting the 25 k resistor, one can alter the input frequency to any value between 10 kHz and 500kHz, and adjustments to the 500 resistor results in changes in the duty cycle. Flyback converters operate optimally at a 50% duty cycle with a minimum switching frequency of 20 kHz. A 25 kHz switching frequency was originally used when testing the first two gate circuits. 25 kHz seemed a sufficient frequency, but after early problems with the gate circuit, the

8 switching frequency was increased to 50 kHz. This frequency is well above the minimum value, and proved adequate with the new gate drive circuit. The gate drive circuit produced a 50 kHz square wave output which effectively powered the MOSFET. Figure A.2 shows the gate drive circuit.

4. Flyback Converter
4.1 Design The flyback converter offers several advantages over other forms of DC/DC conversion. Common buck converters and boost converters use variations in the duty cycle to achieve voltage changes. However, in order to boost the 12 V output of the battery to the 326 V needed by the power supply, it requires a boost converter with a duty cycle of over 90%. Unfortunately, duty cycles of this degree are extremely difficult to maintain. Even small fluctuations in the duty cycle translate to large changes in the output voltage. Since a power supply require a very small ripple, using a boost converter is not practical. On the other hand, flyback converters achieve voltage boosting by means of a coupled inductor rather than duty cycle. By adjusting the turns ratio around an iron core, large voltage boosts can be attained. Furthermore, flyback converters offer an isolated output. Since the secondary side of the inductor is not directly connected to the primary, it does not have the same common ground. This is advantageous for power supply applications. The power supply requires a DC voltage difference independent of the common ground, rather than a certain voltage drop referred to ground. The flyback converter produces a plus to minus output voltage drop with no defined ground. Appendix D shows all the equations and calculations used in determining the parameters for the flyback converter, including the minimum volume for the inductor core.

10 4.2 Performance The flyback converter took the 12 V input and successfully boosted it to 326 volts. Unfortunately, we were only able to boost the voltage at 10% of the maximum load. We were unable to run the MOSFET with the high currents produced under full loading. Therefore, we focused on obtaining a successful voltage boost, and would later return to maintain the boost under full load. Unfortunately, time constraints did not allow us to progress this far. However, we were able to boost and maintain the desired voltage output. We would have preferred to measure the output under various loading conditions in order to understand the tendencies of the circuit. This would have been the first step in achieving the voltage boost at all power levels. We would have then been able to measure and plot the efficiency of the converter versus the various loading conditions.

11

5. Project Results
The design of the circuit proved to be very adequate for providing backup power. Under low loads, the circuit successfully boosted the 12 V to 326 V. Although the output load was calculated to be 443 Ohms, the circuit was tested using a load of 5 k-Ohms instead. Following the philosophy to learning to crawl before walking, we were attempting to achieve the voltage boost without having to account for the high currents that were creating problems for the MOSFET. Although the transistor had suitable ratings, burning out several chips prompted us to reduce to current and focus on the voltage boost. Additionally, we used a power supply in the lab during the testing of the flyback converter. This was done to prevent problems of draining the battery and maintain a reliable DC output. We were attempting to solve multiple problems, and did not want to be concerning with a battery that may or may not be fully charged. The battery will ultimately serve as the DC source, but we felt a power supply was more appropriate while troubleshooting the flyback circuit.

As mentioned in Chapter 4.2, we were unable to complete all the tests we had planned. A successful output was only achieved very late in the project timeframe, and a single output does not assist mapping the efficiency over various outputs.

12

6. Cost Analysis
The total project cost is the cost of labor plus the cost of the parts. The total time for the project was 144 h. The total labor cost was calculated the equation below.

Labor

total

( 25$ / h ) ( 25 . ) ( labor

actual

The total labor cost was $9,000. The grand total of the project was $9069. The estimated cost was $8780. The cost of the battery was not initially factored in the estimate. Additionally, MOSFETs with high power ratings cost drastically more than normal MOSFETs. This extra cost also brought the end total above our original estimates. Furthermore, a larger budget for research should have been allocated. Numerous transistors, resistors and other chips were over-saturated and rendered useless. These replacement costs were unanticipated. In general, the labor costs tend to make up the highest portion of a total project budget and we were pleased to have been very accurate in our estimate.

13

7. Project Future
Due to time constraints, several issues of the project are unresolved. These issues can be addressed in the near future. First and foremost the output must be tested at full load conditions. The voltage boost is useless unless full power can be reached at full load. In order for this to be accomplished, the high MOSFET temperatures must be reduced. Installing a cooling fan and larger heat sink should drastically lower the temperatures.

The inductor currently used is substantially too large. An oversized inductor does not negatively affect the circuit performance, however, it represents the largest amount of space. By reducing the inductor as much as possible, the overall circuit can be combined into a smaller package. Once that package is created, the interfaces can be added. An input plug for the wall receptacle will be added, as well as an input plug for the power supply.

One final aspect to be considered for the project will be providing backup power for the display. It will be determined if the best solution involves the UPS or the power supply providing power to the monitor in the event of a loss of primary power.

14

8. Post-Project Conclusions
Overall, this project was well understood. Professor Krein did a quality job of teaching this material over the course of the semester. The theory behind a flyback converter is neither complicated nor abstract. Overall, the UPS circuit is fairly simple. Fortunately, the gate drive circuit had already been designed, tested, and proven by Professor Krein. Unfortunately, a simple ideal circuit does not imply a simple real circuit. There were numerous real-life implications within the circuit we had not anticipated. All of the wiring in the circuit added inductance we had not accounted for. We reduced the inductor connection wires to remove as much excess inductance as possible and had to design the circuit to withstand high current and power ratings. Lastly, we learned the importance of using capacitors to improve output signals.

Finally, we learned that design simulation could have saved us many a headache. Unfortunately, the simulation software was incomplete and not applicable to portions of our project. This resulted in most of our testing and engineering to be completed by the trial and error method. Consequently, this method contributed to our damaged devices, and the difficulty in troubleshooting.

In summary, our project provided a great experience in combining the implementation of theory and with real application effects. We achieved small successes, and feel the necessary tests and modifications can be performed in order to produce a circuit that meets the original design intentions.

15

APPENDIX A. Circuits

Figure A.1 Rectifer and Charger

Figure A.2 Gate Drive

16

APPENDIX A. Circuits

Figure A.3 Flyback Converter

17

Appendix B. Data Tables


Figure B.1 Project Specifications AC Input Voltage120 VDC Input Voltage12 VDC Input Current20 AInput Frequency50 kHzDC Output Voltage326 VDC Output Current0.74 AOutput Power240 WOutput Ripple 5%Converter Efficiency 85%

Appendix C. Equations and Calculations


C.1 C.2 POUT = PIN IOUT = (POUT)/(VOUT) 240 W = 240 W (240 W)/(326 V) = 0.74 A

18 C.3 C.4 C.5 C.6 C.7 C.8 C.9 C.10 C.11 C.12 C.13 C.14 C.15 C.16 IIN = (PIN)/(VIN) RLOAD = (VOUT)/(IOUT) (240 W)/(12 V) = 20 A (326 V)/(0.74 A) = 440

D = D1 = D2 = 0.5 (with duty cycle of 50%) IT1 = (2*IIN)/D IT2 = (2*IOUT)/D VT1(BLOCK) = 2*VIN VT2(BLOCK) = 2*VOUT T = D/f V = (VOUT)*(VRIPPLE) I = (IIN)*(IRIPPLE) L = (VIN)/ [(I)/( T)] C = (IOUT)/ [(V)/( T)] W = 0.5*LI2 (40 A)/(0.5) = 80 A (1.48)/(0.5) = 2.96 A 2*12 V = 24 V 2*326 V = 652 V (0.5)/(50 kHz) = 10 S (326 V)*(0.02) = 0.65 V (20 A)*(0.02) = 0.4 A (12 V)/[(0.4 A)/(10 S)] = 300 H (0.74 A)/[(0.65 V)/(10 S)] = 11.4 F (0.5)(12 H)(402) = 9.6 mJ

W = BSAT2 (VolCORE)/(2R0) 9.6 mJ = (12)(VolCORE)/(2*25*4x10-7) VolCORE = 6.03 x10-7 m2

References
P. T. Krein, Elements of Power Electronics. New York: Oxford, 1998

19

Das könnte Ihnen auch gefallen