For the PDF version of this article, click here.
The switchmode power supply (SMPS) accounts for more than 10% of the total system weight in a typical portable computer. Power supplies of today's smaller and lightweight systems represent a considerable part of a variety of system dimensions and weight. To keep up with the trend to smaller systems, manufacturers of SMPSs have defined roadmaps to increase power density and efficiency while decreasing size and weight.
Reducing total power loss, resulting in a decreased necessity for cooling devices (heatsinks or fans), is one typical approach to miniaturization. Another approach includes reducing the size of the passive components by increasing switching frequency and/or reducing the size of EMI filters by decreasing noise generation.
A third option, which has become available with the appearance of silicon carbide (SiC) Schottky diodes, is to decrease total power loss and reduce component size by increasing switching frequency.
The way in which an SMPS is switched has a considerable effect on power loss and frequency. One approach uses soft-switching techniques to turn power on and off in a semiconductor device without an abrupt change. This constrains the switching of power devices to time intervals when the voltage across the device, or the current through it, is nearly zero, which reduces the device switching losses and allows higher switching frequencies. However, soft-switching techniques have some major drawbacks. For example, they require additional passive components in the commutation circuit with potential waiting time for recharging.
The traditional hard-switching technique is much simpler to implement than soft-switching, but switching losses in the semiconductors limit how much the operating frequency can be increased with hard switching. Obviously, the power semiconductors in a hard-switching supply must achieve a significant reduction in power losses to allow higher frequencies. Such unipolar devices as MOSFETs and Schottky diodes are replacing bipolar devices. The benefit of unipolarity is the absence of stored carriers and, therefore, theoretically instantaneous switching transients that are limited only by small parasitic capacitances. This allows the minimization of switching losses by reducing the “on-to-off” and “off-to-on” transition times of the power MOSFET switches, reducing those times in which high instantaneous currents and voltages occur simultaneously.
Compared to standard “ultra-fast” power diodes, silicon Schottky devices offer improved performance because of their lower forward drops and reduced forward and reverse recovery characteristics. However, their low breakdown voltages have limited their use to low-voltage applications. The recent emergence of silicon carbide (SiC) Schottky diodes has dramatically changed the situation. Silicon carbide has a high blocking voltage capability (up to 3500 V), a bandgap three times higher and a breakdown field 10 times higher than silicon, and a thermal conductivity that is comparable to that of copper. The result is a high-performance power diode with low switching losses that can readily be used in high-voltage applications.
The combination of power MOSFETs with SiC Schottky diodes sets a new efficiency benchmark in hard-switching circuits, making much higher switching frequencies feasible.
You can understand the influence of switching frequency on power loss in SMPS semiconductors and passive components by looking at a continuous conduction mode (CCM) boost converter configured as a power-factor corrector (PFC). The power loss contribution and physical dimensions of each major component in the boost converter can be analyzed as they relate to switching frequency. Then, you can see that decreasing power loss and physical size leads to no degradation in performance or increase in performance is possible with no increase in power loss or size — all through the use of SiC Schottky diodes.
A boost converter can have as few as four basic components — a power semiconductor MOSFET switch (M), a diode (D), an inductor, or “boost choke,” (L) and a bulk storage capacitor (C) — but also typically includes an EMI filter and rectifier (Fig. 1).
The essential control mechanism of the circuit involves turning the power semiconductor switch on and off. When the switch is “on,” the current through the inductor increases and the energy stored in the inductor builds up. When the switch is “off,” current through the inductor continues to flow via the diode, bulk capacitor and load back to the source as the inductor discharges its energy. The inductor acts like a pump, receiving energy when the switch is closed and transferring it to the bulk capacitor and the load when the switch is open. Because the time constant is very much larger than the “on” period of the switch, the output voltage remains relatively constant.
Power factor is the ratio of total active power to total apparent power in an ac circuit, in which voltage and current are rms values. Any mismatch results in harmonics and phase displacement. It is a measure of the effect the supply and the load have on the ac line. A constant voltage/current ratio will have a power factor of 1.0, and the input will appear resistive. When the ratio varies, the input contains phase displacement, harmonic distortion, or both, and power factor correction must be implemented. Power factor correction is a method of increasing the power factor to as close to 1.0 as possible. A power-factor corrector accomplishes this by precisely controlling its input current to match the waveform of the input voltage.
Boost converters can be driven in discontinuous conduction mode (DCM) or continuous conduction mode (CCM). Discontinuous mode is a soft-switching technique that uses a zero current switching (ZCS). That is, the switch, M, is turned “on” after the current through the diode, D, gets to zero. In other words, the inductor current must fall to zero before the start of the next cycle. The amount of “dead time” during which the current stays at zero defines how strongly the supply operates in DCM. This has drawbacks:
- Oversized circuit components because of the high peak currents.
- System instability at light load.
- Complex EMI filtering system requirements
In CCM, the switch is turned “on” before the current through the diode reaches zero. The CCM solution does not have the disadvantages of the DCM. Circuit components are not oversized, the system remains stable at light load and EMI filter requirements are less rigid. But the choice of switching frequency is limited by power losses caused by the diode's reverse recovery.
Boost Converter Analysis
The inductance value of the boost choke is usually chosen according to a reasonable value of the input current ripple at a given input voltage, which is described by:
This inductance value depends on the switching frequency:
An analysis of CCM PFC chokes has been done for a choke with the following characteristics:
- Core shape: Toroid
- Core material: Magnetics Kool µ
- Winding style: Single-layer turns
- Optimized to minimum core volume
- Total power loss of the choke is limited by a constant value of:
The table, on page 42, shows the results of this analysis.
As would be expected, the inductance value (L) was found to decrease with frequency (Fig. 2). Permeability (µREL) also decreased with higher frequencies, which is equivalent to larger air gaps in the core.
The different elements contributing to power loss are shown in Fig. 3. At higher frequencies, the minimum total power loss was at PLOSS CORE = PLOSS CU.
The frequency dependence of the core volume (VCORE) can be seen in Fig. 4. Because of the constant relation PLOSS TOTAL/ASURFACE CORE, the frequency dependency of total power loss is similar to that of the core. Therefore, at high frequencies, the size of the boost choke can be significantly reduced. This also has a strong impact on the inductor cost.
Selection of the bulk capacitance value is based on the “hold up” time requirement of a typical SMPS. Today's bulk capacitors are based on aluminum electrolytic technology. Their total impedance, which depends on frequency, can be represented by an equivalent circuit consisting of a series resistance, inductance and capacitance (Fig. 5).
Due to ripple and leakage current, the voltage drop on the equivalent series resistance (ESR) primarily causes the power losses in the bulk capacitor. The leakage current depends on capacitor type, temperature and operating voltage and is not influenced by the switching frequency.
However, as shown in Fig. 5, the ESR is frequency-dependent. That's why it is necessary to distinguish between the low power-main line frequency and the higher boost converter MOSFET switching frequency. The ESR power losses should be calculated as a superposition of effective ESR value and rms current at the power main frequency and effective ESR value and RMS current at the switching frequency.
As shown in Fig. 6, the power loss in the bulk capacitor increases slightly with switching frequency. The capacitance value and, correspondingly, the volume of the capacitor can remain the same for different frequencies due to the “hold up” time requirement and 50 Hz/60 Hz load.
Although it seems a higher switching frequency would require a larger EMI filter, experience reveals this not to be true. For example, 140 kHz and 500 kHz boost converter boards can have filters of the same size, if the 500 kHz board uses a jittering technique to achieve EMI compliance.
One contributor to switching losses is the reverse recovery of diodes during turn-on of the MOSFET. As Fig. 7 illustrates, ultra-fast bipolar silicon diodes show a dramatic increase in power loss as frequency goes up. Even a configuration using two 300V silicon diodes in series, while better than a single diode, has a power loss increase.
Silicon carbide (SiC) Schottky diodes don't show the reverse recovery behavior of bipolar diodes. For example, there's no need to remove excess carriers from the N-region, as there is for traditional silicon PN diodes. Instead, a displacement current for charging the metal-semiconductor-junction capacitance of the diode can be observed.
As shown in Fig. 7, the power loss of SiC Schottky diodes is the least frequency-dependent, due to the low switching power losses. SiC Schottky diodes therefore enable designers to go to higher switching frequencies.
An SiC Schottky diode operating at greater than 300 kHz can replace an ultra-fast silicon diode at 70 kHz, or two serial 300V silicon diodes at 120 kHz, without an increase in semiconductor power loss. The heatsink can remain the same. Alternatively, going from 70 kHz with an ultra-fast silicon diode to 140 kHz with an SiC Schottky diode allows a reduction in the size of the heatsink and the PFC choke.
Performance evaluations have been made on a hard switching 400W CCM PFC boost converter with the following characteristics:
- Wide input voltage range: 85V to 265V
- Output power: 400W
- Output voltage: 385V
- Power factor meets EN61000-3-2
- Conducted emission meets EN55011-B
- CoolMOS™ Switch: SPP20N60C3
- SiC Schottky Diode: SDP04S60
The measured efficiency of the 400W CCM PFC boost converter with SiC Schottky at 85Vac reaches 93% at 140 kHz, and goes down to 91.5% at 500 kHz. This corresponds to total power losses in the system of 28W at 140 kHz and 34W at 500 kHz. The power loss contribution of each identified component is given in Fig. 8. (These figures are based on a combination of measured and calculated values.)
The evaluation demonstrates that in a CCM PFC boost converter, increasing the switching frequency leads to the following:
- Power losses in the EMI filter, line rectifier bridge and shunt resistor are constant, so the size of these components can remain the same.
- The PFC choke dissipates less power due to lower winding losses. Its volume can be dramatically reduced.
- The bulk capacitor introduces insignificantly greater power loss. Its capacitance and volume remain the same due to the “hold up” requirement.
- Switching losses in the power semiconductors increase significantly, so the heatsinks should be increased.
Fig. 9 demonstrates total power losses in the CCM PFC boost converter using different diodes at different switching frequencies. The following list details some of the test results.
- Replacing a conventional ultra-fast Si diode with an SiC Schottky diode at 140 kHz (point 1) reduces the total power loss by 8.7W at 400W output power (about 2% more efficiency).
- Replacing a conventional ultra-fast Si diode at 70 kHz with an SiC Schottky diode at 140 kHz (point 2) reduces the total power loss by 4W at 400W output power (about 1% more efficiency), and reduces the PFC choke volume by 33%.
- Replacing a conventional ultra-fast Si diode at 70 kHz with an SiC Schottky diode at 350 kHz (point 3) keeps the same power loss and heatsink size, but reduces the PFC choke volume by 65%.
- A comparison with a “double diode” configuration using two 300V Si diodes in serial produces similar results.
These test results indicate that introducing an SiC Schottky diode gives the designer a new degree of freedom during the optimization of the converter.
For more information on this article, CIRCLE 332 on Reader Service Card