|
Buck converter designs that use a catch diode suffer from low conversion efficiency especially at low output voltages (1 to 3.3 volts) because the power dissipated in the diode, which has a typical forward voltage drop of 0.35-0.5 volt, is a large percentage of the total power loss. Synchronous buck converters, on the other hand, replace the diode with a MOSFET. The voltage drop in the MOSFET is proportional to its turn-on resistance and current, and the typical voltage drop is about 0.1-0.3 volt.
Consequently, MOSFETs reduce the power loss significantly. But the input-voltage range in a low-voltage output application, such as automotive for example, can be high—12 to 24 volts, with 40-volt spikes not uncommon. In these cases, two-stage conversion is usually applied to maximize the converter's efficiency. Here's a single-chip, single-stage design, however, that's superior.
The conversion issue
If you have a high input, low-output voltage converter, chances are it uses two-stage conversion. The first stage converts a high input voltage to an intermediate voltage, and the second stage converts the intermediate voltage to the required low output voltage. The PWM operates at a small duty ratio. If you have a 24-volt input and 1.2-volt output, for example, you need a duty ratio of 0.05. But it's not optimized for efficiency or performance. And in the common buck converter, a very low duty ratio often isn't easily achieved. In addition, designs using devices supporting an output voltage of lower than 1.2 volts can't usually deal with an input voltage of more than 10 to 15 volts. Automotive systems are often called on to handle up to 40 volts, as previously mentioned, and in such converters you can expect the minimum output voltage to be greater than 1.2 volts. Thus two-stage conversion is useful for a high-input, low-output voltage application.
Poor efficiency in two-stage conversion
Ironically, however, efficiency is one of the major concerns for two-stage conversion. Each stage may exhibit high efficiency, but the overall efficiency will often be low (each stage's efficiency is multiplied together). Figure 1 shows the efficiency curves for the individual stages of a buck converter that converts a 12- or 24-volt rail to 5 volts. The figure also shows the efficiency curve of a buck design for converting a 5-volt input to a 1.2-volt output. Both buck converters operate at 550 kHz and show about 80 percent efficiency at what amounts to about half load. However, the overall efficiency of these two-buck designs in a two-stage conversion is just 60 to 70 percent, as shown in Fig. 2.
Figure 1: Efficiency, buck converters, individual stage
Besides efficiency, the two-stage conversion requires more components and board space as compared with a "single-stage" conversion. The number of ICs, inductors, and bulk capacitors used are essentially doubled. Moreover, a two-inductor design involves careful synchronization to minimize EMI. The design and debugging process takes longer.
Figure 2: Overall efficiency, two-stage conversion
Improving efficiency
Synchronous buck converters with a wide-input range and low feedback voltages can provide much greater efficiency and smaller size and cost than two-stage conversion. Consider implementing a chip such as National's LM3103, for example. The input voltage of the LM3103 can be up to 42 volts, and the output voltage can drop down to 0.6 volt. Further, you'll benefit from IC (versus discrete) designs having imbedded power MOSFETs, as does the LM3103. It also employs a constant on-time control scheme, and so no compensation circuit is required. You can optimize this class of converters simply by adjusting a few components.
A LM3103-based 1.2-volt converter is shown in Fig. 3. Capacitors CIN and COUT are bulk capacitors. CIN3 and COUT3 are used for filtering out high frequency noise. Capacitors CSS and CBST are for the soft-start and bootstrap functions. CVCC is used for the internal regulator, and CFB is for feeding back the output ripple.
(Click on Image to Enlarge)
Figure 3: LM3103, 1.2-volt buck converter design
The components for setting the output or circuit parameters include RFB1 and RFB2, for the output voltage; RON, for the operating frequency; and L, which determines the inductor's current ripple.
Output voltage
The output voltage, VOUT, is determined by:
VOUT = 0.6 (RFB1 + RFB2)/RFB2
where 0.6 is the chip's internal reference voltage. With VOUT = 1.2 and selecting RFB2 = 10 kilohms, we calculate RFB1 = 10 kilohms.
Operating frequency
Resistor RON is used to determine the on-time of the converter, which is directly related to the operating frequency fSW (programmable up to 1 MHz in the LM3103). Once we determine fSW, we can find RON from:
RON = VOUT / {8.3(10-11) (fSW)}
In this case, fSW is 550 kHz, so RON is 26.3 kilohms.
Current ripple
The ripple is related to the input and output voltages, and the operating frequency as well. The required inductance is given by:
L = VOUT (VIN - VOUT)/(0.3)(fSW)(VIN)
where 0.3 amps is the LM3103's required inductor current ripple. Let VIN = 12. Then L is 6.55 microhenries.
Figure 4: Efficiency, LM3103 design
The LM3103's efficiency curves are shown in Fig. 4. As we can see by comparing Fig. 4 with Fig. 2, the efficiency of the LM3103's single-stage design exceeds the overall efficiency of a two-stage conversion by 5 to 10 percent. The resultant gains in efficiency, component count, and solution size are summarized in Table 1.
(Click on Image to Enlarge)
Table 1: Comparing the LM3103 versus dual-stage conversion
About the authors
L. K. Wong is a senior product application engineer, and T. K. Man is a product application manager, at National Semiconductor's Hong Kong Power Management Design Center.
|
