You are on page 1of 6

96125

LOW TEMPERATURE OPERATION OF A BOOST CONVERTER

Barry S. Moss, R. R. Boudreaux and R. M. Nelms f Electrical tngineering Department o 200 Broun Hall Auburn University AL 36849-5201 Phone: (334) 844-1830 FAX: (334) 844-1809

ABSTRACT The development of satellite power systems capable of operating at low temperatures on the order of 77K would reduce the heating system required on deep space vehicles. The power s u p plies in the satellite power system must be capable of operating at these temperatures. This paper presents the results of a study into the operation of a boost converter at temperatures close to 77K. The boost converter is designed to supply an output voltage and power of 42 V and 50 W from a 28 V input source. The entire system, except the 28 V source, is placed in the environmental chamber. This is important because the system does not require any manual adjustments to maintain a constant output voltage with a high efficiency. The constant 42 V output of this converter is a benefit of the application of a CMOS microcontroller in the feedback path. The switch duty cycle is adjusted by the microcontroller to maintain a constant output voltage. The efficiency of the system varied less than 1% over the temperature range of 22C to -184OC and was approximately 94.2% when the temperature was -184C. INTRODUCTION Due to recent advances in low temperature microelectronics, power converters that can operate efficiently in a low temperature environment have become a design goal for engineers working on deep space applications. The low temperatures require the use of MOS-type devices, instead of commonly used Bipolar devices which do not work well in extremely low temperatures (Ray, 1995a). The ability of a satellite power system to begin operation from a cold start is very important. If the system can cold start, then it may remain dormant until it is needed. This obviously includes converters, such as the one discussed here, which is to be used as a battery charger. The reduction of heating systems for the power

supplies on the satellite would be instrumental in conserving the total energy used by the satellite. One would actually expect certain circuitry to operate more efficiently in low temperature rather than at room temperature due to the better thermal, electrical, and electronic properties of materials at low temperatures (Ray, 1995b). The converter, shown in Figure 1, would use the 28V bus in the satellite to supply a 50 W load operating at 42 V.All of the elements in the dashed box were: placed in an environmental chamber and operated at low tempcrauures. Other research has been done on converters at low temperatures. Oftentimes, however, the output voltage is controlled by manual adjustments of components located outside of the test chamber. For instance, in one case an external 10 K potentiometer is adjusted to regulate the output voltage (Ray, 1995b). In another case the input and output filter capacitors as well as the control circuitry were kept at room temperature (Ray, 1995~). The inclusion of the feedback loop with the microcontroller allows regulation of the constant output voltage without manual intervention. This is very useful for low temperature applications which have the individual circuit components change values or change actual performance due to the low temperature, affecting the output voltage and overall efficiency. With a changing temperature, values for resistors and possibly capacitors are expected to change, while at the same time most MOStype semiconductor devices are expected to improve in overall performance (Ray, 1995a). However, inductances are insensitive to temperature variations (Ray, 1996). This paper discusses the design and performance of a boost converter. The efficiency and output voltage are monitored as the temperature is varied from 22C to -184C. The next section discusses the design of the power stage of the boost converter and the design of the control circuitry of the converter. The experimental test setup and results will then
be discussed.

599

0-7803-3547-3-7116 $4.00 0 1996 IEEE

BOOST CONVERTER

Power Stane The design of the inductor in Figure 1 was based on keeping the converter in the continuous mode to keep the ripple current through the inductor small. The switching frequency output by the microcontroller is 31.3 kHz. Using fs=31.3kHz, and assuming D=lB and an efficiency of loo%, equation (1) was used to find L (Rashid, 1993).

IL avg at boundary

kD(l-D)/(2Lfs)

(1)

where IL avg is the average value of the inductor current at the boundary of being a continuous or discontinuous current. Using an I L avg value of 1.183 A, equation (1) yields b 8 4 pH for operation in the continuous current mode. A target value for the inductor was 100 pH. The inductor was designed using Magnetics, Inc. design software (Magnetics, 1993). Using MPP core #408T-58867-A2 the design software recommended 35 turns. The inductance was measured using an LCR meter and found to be 98 pH. Metallized polypropylene film capacitors were used for the input and output filter capacitors. The output capacitor was designed around a ripple voltage of 0.84 V (2% of Vo) and an inductor ripple current of 3 A (to remain in the continuous mode). Equation (2) was used to calculate the output filter capacitor (Rashid, 1993).

serial interface. The 87C752's PWM output has a fixed resolution of %bits (256 discrete values from 0 to 100%). For an oscillator frequency of 16 MHZ, the maximum PWM frequency is 31.373 MHZ. Since the P W M output is solely under s o h a r e control and directly controls the MOSFET switch, a high degree of flexibility for controlling the switch can be achieved. The internal &bit A/D converter has five multiplexed input channels allowing five external voltages to be measured. The A/Dconverter has a sampling time of 6 p and a conversion time of 30 p for each channel (16 MHz oscillator). The conversion time and the time necessary to calculate the duty cycle limits the frequency at which the duty cycle can be changed. Because the microcontroller operates from a 5 V supply, a voltage divider is placed across the output of the converter to divide down the output voltage to a level compatible with the microcontroller. The voltage divider is designed to supply 2.5 V to the A/D pin when the converter output voltage is 42 V. EXPERIMENTALSETUP AND RESULTS The boost converter was placed in an environmental chamber and the temperature was lowered from 22OC to -184C in steps of 20OC. The -184OC temperature is the lowest possible temperature which can be reached by the environmental chamber. At each temperature step both the input voltage and current and the output voltage and current were measured. The efficiency was calculated using the ratio of the output power to the input power. The input power was calculated from measurements of the input current and voltage. The output current and voltage were measured to calculate the output power. The circuit was allowed to remain at each temperature for approximately 4 to 5 minutes before data was recorded. Two different circuits were actually tested in the environmental chamber. One test used linear regulators to power the microcontroller and the E2110 (see Figure 2). Figure 3 contains a plot of the efficiency as a function of temperature with the linear regulators. The efficiency at -184OC was 94.2% while at 25OC the efficiency was 95%. Even though the efficiency initially dropped, it varied less than one percent over the range of temperatures. Figure 4 contains a plot of the output voltage versus temperature. Notice the output voltage also remains constant at 41.4 V. The +1SV and +5V supplies were produced using standard linear voltage regulatw manufactured by National Semiconductor with part numbers LM78L1 SAC and LM78LOSAC, respectively. The regulators worked satisfactorily as the temperature was reduced to 184C. f f for about 5 minutes and then a cold The converter was turned o start was attempted by simply turning on the 28 V power supply. It is important that a stable +5V supply be present before the microcontroller attempts to start to prevent latch up. For the linear regulators, a cold start was possible only afler allowing the regulators sufficient time to dissipate power internally to heat up enough to provide stable power supply voltages. With the linear regulators, the reset pin for the microcontroller was held high manually until the linear regulators heated up and produced the required voltages. In a second test zener diodes are utilized to produce the +15V ) . The zener diodes are and +5V power supplies (see Figure 5

Equation (2) requires that 0 1 4 pF. The value of the output capacitor used was 20 pF. The input filter capacitor, determined experimentally, was 30 @. The W O 4 4 is a power MOSFET manufactured by International Rectifier. It has a voltage breakdown rating of 60 V and a current rating of 30 A. The on-state resistance is 28 mR. The diode, an MUR5015 manufactured by Motorola, has a voltage rating of 150 V and a current rating of 50 A.

Control Circuitry The IR2110, also manufactured by International Rectifier, is a CMOS driver chip which supplies the drive signals to the W 0 4 4 . The IR2110, with CMOS outputs, has a 2 A peak output current capability and only a 25 ns switching time (International Rectifier). Regulation of the converter output voltage is achieved using a CMOS microcontroller. This microcontroller is a Philips 87C752 8-bit microcontroller, which is a derivative of the 80C51 architecture. It has a maximum oscillator frequency of 16 MHz and draws less than 60 mW of power under normal operation at an oscillator frequency of 12 MHZ. The oscillator, Seen in Figure 1, is a CTS 16.000 MHz CMOS clock oscillator. The major features of this microcontroller are 2 kilobytes of internal program memory, 64 bytes of internal random access memory, five 8-bit A/D converter channels, one 8-bit P W M output, 21 I/O lines, and a bi-directional

1N4689 models manufactured by Motorola. The series resistance of 200 2 ! was used to keep the current to the microcontroller and the IR2110 to a minimum value while still supplying the required current the zener diodes need to produce their voltages. Figure 6 is a plot of the efficiency versus temperature with the zener diodes. The efficiency of the system with the zener diodes is slightly lower than with the linear regulators. Figure 7 is a plot of the output voltage versus temperature. The output voltages are 41.5 V and 43.1 V at -184OC and 25OC, respectively. The output voltage using the zeners is obviously not as stable as the voltage using the linear regulators. Using the zeners the output voltage varies by 2.2 V over the temperature range. These difficulties may be explained by the equivalent model of the zener diode. The voltage across the zener is partially dependent on an internal resistance. As the temperature is lowered this resistance changes. The +5V to the microcontroller represents the logic 'high' for the system. Some values may be misinterpreted if the logic level drops too far. Successful cold start, though, can be achieved with the zener diodes. The converter was turned off for about 5 minutes and then a cold start was attempted. Note the switch in series with the source is used to simulate a cold start by simply 'switching' the supply and converter together and was only used with the zener diode test setup. A capacitor was placed across the reset pin of the microcontroller. This capacitor prevented the reset pin from reaching a logic level 1 until stable voltage references were obtained.

lating the power supply voltages of the microcontroller. The inclusion of the feedback loop in the environmental chamber permits regulation of the ouqpul voltage as the components vary with temperature without human intervention. This is particularly important because human intervention is not possible for deep space missions. Overall, both circuits were very efficient at all temperatures from 22OC to -184OC. For the linear regulators, the efficiency was 94.2% at -184C and 95% at 22C. For the zener diodes, the efficiency was; 92.2% at -184OC and 93.5% at 22OC. The output voltages of both systems were acceptable at all temperatures.

Acknowledament This research was supported by the Center for Space Power and Advanced Electronics with funds from NASA grant NAGW-1192CCDS-AL., Auburn University, and the Centers' industrial partners.
REFERENCES IR2110 High Voltage Bridlge Driver, preliminary data sheet No. PD-6.011, International Reclifier, El Segundo, CA Magnetics, Inc., 1993, "Designing Powder Core Inductors with MPP, High Flux, and Kwl Mu Cores," Butler, PA Rashid, M. H., 1993, WM Electronics: Circuits. Devices, and ADDlications, second edition, Prentice-Hall, Inc., New Jersey, pp. 317-330. Ray, B., Gerber, S. S . ,Patterson, R. L., and Myers, I. T., 1995a, "Power Control Electronics Ifor Cryogenic Instrumentation," Proceedings o f the 2nd International Conference on Advances in Instrumentation and Control, Toronto, Ont, Can 1995 Apr 25-27. Ray, B., Gerber, S.S.,Patterson, R. L., and Myers, I. T., 1995b, "Liquid Nitrogen Temperature Operation of a Switching Power Converter," Sym. Low Temp. Electronics and High Temp. Superconductivi&, Vol 95-9, The Electrochemical Society, May 1995. Ray, B., Patterson, R. I,., 1995c, "Wide Temperature Operation f the 1995 of a PWM DC-DC Converter," Conference Record o IEEE Indusby Applications Socie& 30th USAnnual Meeting, Vol. 2, pp. 971-975. Ray, B., Gerber, S. S . ,Fatterson, R. L., and Dickman, J. E., 1996, "Low Temperature Performance of a Boost Converter with MPP and HTS Inductors," Proceedings, Eleventh Annual Applied Power Electronics Conference, Vol. 2., pp. 883-888.

CONCLUSIONS DC-DC converters capable of operating at low temperatures would have applications in areas such as deep space probes. This paper has presented test results for a boost converter designed for operation at temperatures down to -184OC. An 8-bit CMOS microcontroller was employed in the feedback loop to maintain a constant output voltage as component values varied with temperature. Two different designs for +5V and +15V power supplies for the microcontroller and other control chips were investigated: linear regulators and zener diodes. In comparison with the linear regulators, the use of zener diodes to produce +5V and +15V power supplies result in a lower efficiency and less stable output voltage. However, the zener diodes cold start better than the linear regulators. So the cost of having a better cold start is also having a lower efficiency and a less stable output voltage. This issue will be investigated further to pursue an even more efficient manner of regu-

601

98 pH

M U R 5015
I4

28Vdc

I -

!D

T
p 6 l
Oscillator

pgy ADC -Microcontroller Phjlips 87C752


I

FIGURE 1. EXPERIMENTAL BOOST CONVERTER

2av

FIGURE 2. LmEAR REGULATOR TEST SETUP

95.2 95 94.8 v 94.6 c 94.4 a , ' 0 94.2 94 W 93.8 93.6 -/

.=

-184 -160 -140 -120 -100 80 -60 -40 -20

25

Temperature ("C)
FIGURE 3. EFFICIENCY OF SYSTEM WITH LINEAR REGULATORS

602

45

E 43
Ql

2 -

41 39

-184 -160 -140 -120 -100

-80

-60

-40

-20

25

T e m p e ra ture ( " C )
FIGURE 4. OUTPUT VOLTAGE OF SYSTEM WITH LINEAR REGULATORS

i i 2
28V

To IR2110
e 5;

L
loo

3OuF

htz

1st

ktz -% .

To Microcontroller k
A

5v

FIGURE 5. ZENER DIODE TEST SETUP

95

85

-184

-160

-140

-120

-80

-40

25

Temperature ( " C )
-

FIGURE 6. EFFICIENCY VERSUS TEMPERATURE WITH ZENER DIODES

603

m
I I C ,

45 43

3 41
.cr
IC,

2
3

39
37 -1$4 -160 -140 -120
-80
-40

25

rature ("C)
FIGURE 7. OUTPUT VOLTAGE VERSUS TEMPERATURE WITH ZENER DIODES

604

You might also like