Developing an Efficient Power Supply for Extremely Wide Input Voltage RangesMay 18, 2020 by Florian Mueller
This article highlights Texas Instrument method for designing an efficient ultra-wide input power supply and gives some practical tips for achieving a well-optimized design.
Imagine the possibilities if you could design a power supply that meets the input and output requirements of all applications. But designers must optimize a power supply for a specific operating range for several reasons, one being that controllers have internal limits. Techniques such as zero voltage switching, variable switching frequency, or synchronous rectification reduce the power loss of the parts of the power stage, but these techniques reduce the achievable input and output voltage ranges.
Many applications require a wide input voltage range which results in very short or long duty cycles and can lead to poor performance and high losses. This article will present a method for designing an efficient ultra-wide input power supply and gives some practical tips for achieving a well-optimized design. Suppose you need to design a 75-W flyback with an input voltage range between 20 V and 375 V.
For a power level up to 100 W, a flyback topology is a good choice, mostly because it´s the most cost-effective isolated topology. Gone are the days when controllers switched with a constant frequency; modern controllers now modulate the switching frequency to achieve high efficiency. Typically, the switching frequency changes depending on the input and output conditions.
As the designer, you must deal with limits such as the minimum on-time, the maximum duty cycle, and the minimum and maximum switching frequency. These limits make it difficult for the controller to operate over a large input voltage range. For an extremely wide input voltage range like 20 V to 375 V, you must take a different approach, like the two-stage power supply shown in Figure 1.
Figure 1: Two-stage power supply
The first stage consists of a pre-boost and is only active if the input voltage is below 130 V. The boost generates a boost output voltage, V boost, of about 130 V. Therefore, even with a minimum input of 20 V, the amplification factor is below 7, to ensure proper operation.
If the input is higher than the V boost, the control loop stops the operation automatically and the boost controller becomes inactive. The galvanic connection between the input and output of the boost ensures that the input voltage connects directly to the second stage.
The second stage consists of a modern flyback controller. The most efficient flyback uses an active clamp technology, which recovers the leakage energy and ensures soft-switching or even zero voltage switching. Efficiency as high as 94% is possible if using a secondary-side rectifier.
Note that the overall efficiency is the product of the first stage (pre-boost) and the second stage (flyback). When the input voltage exceeds 130 V, however, the pre-boost becomes disabled and the efficiency depends only on the second flyback stage. As a result, efficiency much greater than 90% is possible over a wide input voltage range.
A Reference Design Example
The “High efficiency, ultra-wide input (20 VDC to 375 VDC) isolated power supply reference design”1 from Texas Instruments (TI) has an input voltage range from 20 V to 375 V, an output voltage of 24 V and a maximum output current of 3.5 A. Figure 2 shows the efficiency vs. input voltage curve.
Figure 2: Efficiency vs. input voltage
The efficiency is over 90% for an input voltage range between 25 V and 375 V, with a peak efficiency of 94%. How is this possible? This reference design follows the same approach as Figure 1. Basically, the design consists of three parts: a pre-boost stage, an active clamp flyback (ACF) stage, and a startup circuit. The pre-boost stage uses the TI UCC28C42 current-mode controller and the ACF stage uses the TI UCC28780 flyback controller.
Tips for Designing a Startup Circuit
At the beginning of schematic design, you should think about the startup circuit, as it presents a challenge for a wide input voltage range. The pre-boost and active clamp VDD capacitors must be charged to enable startup. It is known that a resistive startup method leads to an increase in losses, especially in high input voltage applications. Standby mode is a very common state for power supplies. Therefore, an active startup circuit is often required to reduce standby losses. Such a circuit may include a “normally on” device such as a depletion-mode metal-oxide semiconductor field-effect transistor (MOSFET). Figure 3 shows a simplified startup circuit.
Figure 3: Simplified schematic of an active startup
The depletion MOSFET Q1 charges the VDD capacitor when the controller is not yet operating. Once the VDD voltage has reached its Undervoltage lockout level, the controller starts to work. The auxiliary winding supplies the controller via the diode D2 and Q1 can switch off (via the auxiliary transformer winding D1 and Q2). The schematic of the isolated power supply reference design shows this depletion-mode MOSFET startup circuit in more detail. The auxiliary winding of the ACF flyback transformer is used for multiple functions: to turn off the depletion-mode MOSFET and to supply the boost and ACF controller.
Tips for Designing a Pre-Boost Circuit
The pre-boost is designed to operate in continuous conduction mode.
Figure 4: Simplified schematic of a boost
During turn-OFF of the diode, a high reverse recovery current in a silicon diode would produce high losses. It is recommended to use a fast switching MOSFET and a silicon carbide (SiC) Schottky diode, as this drastically reduces losses, mainly because a SiC diode has almost no reverse recovery current. Incidentally, you could use a bypass diode to avoid high surge currents through the SiC diode, Dboost. As mentioned earlier, the output voltage is regulated to 130 V.
The feedback loop stops the operation of the boost controller when the input voltage is greater than 130 V. Nevertheless, all components must be designed for the maximum input voltage of 375 V (plus some margin) and must withstand the maximum currents. A freeware tool from Texas Instruments called the Power Stage Designer™ tool2 can display the voltages and currents of all common topologies. This makes it very easy to select which devices can withstand the maximum peak and root-mean-square currents and voltages.
Tips for Designing an ACF
The second stage consists of an ACF. A normal discontinuous conduction- mode flyback with a passive clamp dissipates the leakage energy of the transformer on a passive snubber circuit, while an ACF recycles the leakage energy and achieves zero voltage switching over a wide operating range. Figure 5 shows the simplified diagram.
The ACF operates in transition mode and modulates the primary peak current and switching frequency. Q_HS helps recycle and store the leakage energy in a snubber capacitor. In addition, the ACF uses the magnetizing current of the transformer to discharge the switch-node capacitance, Csw, and to bring the switch-node voltage down to 0 V before Q_LS turns on. This enables zero voltage switching and eliminates the switching losses.
In order for everything to work well, you must pay special attention to the transformer. Among other things, the primary inductance and turns ratio will determine the operating modes over the entire load range. Therefore, it is recommended to follow the data-sheet rules and carefully specify the transformer’s minimum on-time, switching frequency range and maximum primary peak current.
Figure 5: Simplified schematic of an ACF
The Power Stage Designer tool2 makes transformer specifications a lot easier.
Finally, a special winding technique is recommended because the windings must be well coupled. For example, split the primary and sandwich the secondary and bias layers in between. To further improve efficiency, consider replacing the output diode with a synchronous rectifier. The UCC28780 ACF works with a drain-to source-sensing (VDS) synchronous rectifier such as TI’s UCC24612. The VDS sensing principle uses the voltage drop of the MOSFET RDS(on) and the body diode to turn the synchronous rectifier MOSFET on and off. The synchronous rectifier can be placed on the positive or negative side of the output winding. When the synchronous rectifier is in the positive path, the common-mode electromagnetic interference noise is lower, but the controller cannot simply be powered by the output voltage. In this case, you must use an additional winding or a resistor-capacitor diode circuit to supply the synchronous rectifier controller.
The Texas Instruments isolated power supply reference design shows a good way to deal with a very wide input voltage range. With an approach like the two-stage power supply, an efficiency of over 90% and very good performance can be achieved. The database of Texas Instruments power reference designs contains solutions for many applications. Consider consulting it early in your development. You can often find a design with similar specifications, which will serve as a good starting point and speed up the design process.
About the Authors
Florian Mueller was born in Rosenheim, Germany, in 1976. He received his degree in electrical engineering from the University of Haag. After working for several years as a freelancer in the field of electrical engineering, he joined TI in 2011 and is working in the European Power Design Services Group, based in Freising, Germany. His design activity includes isolated and non-isolated DC/DC and AC/DC converters for all application segments.
This article originally appeared in the Bodo’s Power Systems magazine.