Current-Fed Power Processing—Ride Through Robustness
This article was written by Magna-Power Electronics and originally appeared in the March 2019 issue of the IEEE Power Electronics Magazine.
Selecting the best power convert-er topology for a specific application can be somewhat over-whelming considering the advances made in the past 20 years. Many of these advances were accomplished with better semiconductor switches providing faster switching rates, lower on-state voltages, lower power loss, lower power driver requirements, and so on. Selecting the best power convert-er topology must balance the application’s demands with priorities of cost, size/packaging, efficiency, and robustness. Magna-Power Electronics (Magna-Power), a programmable dc power product manufacturer in Flemington, New Jersey, United States, prioritized robustness in its topology selection. The topology choice was driven by the customers’ need for high reliability in demanding industrial environments.
With Magna-Power products often being used in experimental, prototype validation applications, customer satisfaction required consistent power converter performance well beyond typical operating conditions. Having shipped tens of thousands of power supplies, the vast majority of field failures results from customer abuse, such as:
- abnormal input voltage such as lightning, power line transients, and power line harmonics
- output abuse, such as back-fed volt-ages and excessive ac currents
- susceptibility of electromagnetic interference in abnormal environments
- poor packaging of power supplies in equipment racks resulting in restrictive air flow and excessive heating.
Typical methods to improve reliability are the implementation of n + 1 redundancy and lowering the mean time between failure. These techniques can only partially improve long-term reliability if the cause of failure is a result of external conditions. For example, a power supply placed in an industrial environment, which is subjected to high incoming voltage transients, will be more reliable at utilizing input rectifiers with a higher blocking voltage rating. In this case, improving reliability with higher-voltage devices can be more beneficial than having redundant power supplies with lower component ratings, which will all fail from the same external environmental condition. Surviving an abnormal external influence, even if operation is temporally aborted, is far better than the product failing and relying on a spare.
One of Magna-Power’s key objectives has been to develop power circuits capable of riding through potentially damaging conditions. Strong ride-through capability, or fault resistance, lowers field returns, reduces cost-ly repairs, and, most importantly, preserves customer satisfaction. Fault protection cannot be achieved under all conditions, but knowing the weak points in a design can give a better understanding to possible improvements.
Most power supply designs in the 1-kW and up power range use a volt-age-fed topology or a similar derivative. As illustrated in Figure 1, the input stage is a dc source feeding a capacitor, bridge inverter, a trans-former for ohmic isolation, an output rectifier, and inductor-capacitor low-pass filter. The weak point of the design is the bridge inverter. If one of the devices should fail or erroneously turn on, the resulting effects can be quite dramatic: failure of other bridge inverter devices, flames, and damage to surrounding circuitry. If the bridge inverter circuit does not switch with a near-perfect volt-second balance, the transformer core will saturate, possibly resulting in similar fail-ure conditions.
Figure 1. A voltage-fed converter
There have been considerable advances in the fault detection of voltage-fed converters. However, protection schemes must still operate on the order of microseconds. Such failures are thermal and device dependent. Protection advances include detection of on-state conduction with antisaturation-sensing circuitry and current-mode controllers that limit peak current through transformers to prevent core saturation. Any external influence causing a wrong switching state can result in a potential failure with little time to take corrective action.
Figure 2 shows the electrical dual of the voltage-fed converter, i.e., a current-fed converter. The current-fed topology has been around for a long time, but it is rarely commercially deployed because of the need and added cost for creating a current source input plus the demand for a low-leakage transformer . The differences between volt-age-fed and current-fed converters are very subtle. The voltage-fed bridge devices should never short the input dc bus, as opposed to the current-fed bridge devices, which should never open the input dc bus. Second, output inductor L1 in a voltage-fed converter is on the output side of the converter, but it is on the input side of the current-fed converter.
Figure 2. A current-fed converter
As illustrated in Figure 3, the requirement for a current source input can be achieved with a buck converter or chopper operated as a voltage-to-current converter. While this is indeed a negative point of the topology, this added stage can be utilized for protection between the two converter stages.
The inverter stage, which has the primary function of ohmic isolation and voltage transformation, should operate at a near 50% duty cycle. This stage can be operated at a lower switching frequency with virtually no degradation in dynamic performance. All of the converter control is governed by the buck converter stage. A low-leakage impedance for transformer T1 ensures high converter efficiency and low output voltage ripple. During com-mutation of the two inverter poles, the dc bus across the inverter shorts, and current is blocked from flowing through the output stage.
Figure 3. A current-fed converter with a buck converter
With the current-fed converter, the time to protect any semiconductor device is dependent on the switching period of the buck converter and the design of inductor L1; both of these parameters govern the time needed to prevent core saturation of inductor L1. For a buck converter operating at 20 kHz, a typical 10-μs period can easily be deployed to protect controlled semiconductor devices within the normal operating limits. This protection can be applied to the inverter or buck semiconductor switches in the event of erroneous switching states, trans-former shorts, or output diode shorts. During a fault condition, the dc bus across the bridge inverter collapses, protecting the remaining devices from catastrophic failure. With the dc bus cur-rent being limited to a current level set and maintained by the buck converter, core saturation of trans-former T1 is virtually impossible.
With sufficient time, fault protection circuitry can be devised to protect the inverter stage with the buck converter stage and the buck converter stage with the inverter stage. Also, failure of a controlled semiconductor switch is current limited by the buck converter, which prevents catastrophic failures. Because of the extended time for protection, special anti-saturation circuitry is not required.
One additional attribute of current-fed topologies is scalability to higher-power, physically larger systems. Adding a little extra inductance in series with inductor L1 has virtually no effect on system performance. Stages can be easily paralleled without major concern for lead inductances.
Despite the advantages of a current-fed design, voltage-fed designs are far more prevalent in industry. Voltage-fed topologies do require one fewer power conversion stage and have fewer magnetic components. Both of these requirements have an impact on cost and efficiency. Manufacturing a single-board power supply can be more cost-effective when power levels are lower—in the 1-kW and lower range. Many manufacturers gang multiple assemblies of the same design to achieve higher power levels, but such designs can have a negative effect on system reliability by additional parts count.
The Challenges of the Current-Fed Converter
Manufacturing current-fed converters have their challenges in a competitive market. There is an extra power conversion stage taking up physical volume along with the materials cost associated with it. Conventional control circuits cannot be used be- cause of the duality of the design, and these circuits have to be designed from fewer integrated devices. Transformers designed for extremely low leakage inductance require unique core geometries. The demands of tighter packaging and specialized magnetic circuits have required Magna-Power Electronics to vertically integrate its manufacturing operations to minimize outsourcing and to optimize its designs.
Today, Magna-Power Electronics manufactures all its assemblies in house. This decision has helped keep the company competitive in a global market, allowing it to realize the cur-rent-fed topology plus improving quality and delivery times. Manufacturing operations include sheet metal fabrication, powder coating, robotic heat sink and fastener assembly, automated surface mount and through-hole printed circuit board assembly, magnetics winding and core fabrication, computer numerical control machining, wire harness fabrication, final assembly, and testing.
 I. Abraham, Pressman: Switching Power Supply Design, 2nd ed. New York: McGraw-Hill, 1998.
Article from magna-power.com