Analysis of the advantages and disadvantages of common three-phase PFC structures, one article get√

Before delving into the technical details and characteristics of the Vienna rectifier, it is necessary to understand its history, but more importantly, to come to a common understanding of what is being discussed. The Vienna rectifier is a pulse-width modulated rectifier invented by Johann W. Kolar in 1993. Before Kolar invented it, people used single phase per phase (with or without neutral) and load sharing to balance phase currents. Today, the term "Vienna" usually refers primarily to a three-phase AC/DC converter, but sometimes also refers to a DC/AC or inverter. For example, the neutral point clamped (NPC) and T−NPC three-level topologies are sometimes called "Vienna" even when operating as an inverter. When discussing so-called "Vienna" converters, it is recommended to identify which "Vienna" it is.

Regarding the characteristics of the "Vienna" rectifier, it is a three-phase connected boost PFC as shown in Figure 7. Single-phase boost PFC consists of inductor, switching device and rectifier diode. In a three-level architecture, there is a "boost" rectifier diode (D _xBy ) for each half-wave or bus voltage (excluding the common ground in between). Then, there is a bidirectional switch, consisting of a full-wave diode rectifier bridge (DxPy _and DxZy ₎ with a unidirectional switch (Qx ₎ within it. We get the following schematic.

Figure 7. Vienna PFC schematic

Switch Q _x is rated for 600V or 650V. All diodes can also be rated for 600V. This will help reduce losses since devices rated for 1200V are not required. On the other hand, diode losses are important. There are always two high-frequency diodes in series in the current path. With these diodes, there is always a trade-off between voltage drop and reverse recovery.

For PWM it's very simple as there is only one switch per phase. Modulation is applied directly to the switch after the inverse Clark and Park transformations. However, depending on the direction of the input sine wave, the current path changes. Depending on the input voltage sign and/or current direction/flow, the diode rectifier bridge and "boost" diode "automatically" participate in the current path. This is well illustrated in Figure 8.

Figure 8. Vienna boost PFC current path

(for storing and releasing energy modes) and phase voltage

As mentioned before, since the current flows from one phase or two phases to the remaining two phases or one phase respectively, only one branch (or one-phase schematic diagram) is drawn in the above figure. Depending on the sector of operation, two modes for each phase (U, V or W) can be derived using the above scheme (the phase voltage first stores energy in the boost inductor and then releases the energy to the output capacitor).

The main advantage of this topology is the use of one switch per phase. Even though the schematic looks more complex due to the number of diodes involved, it makes control much easier. The cost of this topology is also very low because the number of switches is very small. This topology is unidirectional.

A major disadvantage of this topology is the high number of diodes. There are always two diodes in the current path, which affects efficiency. All drives are floating and require a specific floating power supply.

The choice of switch can be based on the power level, using superjunction MOSFET or IGBT. For higher frequency operation and/or smaller size, SiC MOSFETs may also be used. For diodes, silicon STEALTH™ 2 or SiC diodes are recommended.

Unlike the original "Vienna" introduced in the "Vienna Rectifier (Three-Switch Boost)" section, the T-shaped neutral point clamp (T−NPC) implements bidirectional switching in a different way. Instead of using a rectifier bridge to convert a one-way switch into a two-way switch, T-NPC uses a back-to-back switch configuration, as shown in Figure 9. It is also possible to conduct from the body diode when the switch is not conducting and the current is flowing in the "reverse" direction compared to the normal switching current of this switch. This is the case with bipolar devices like IGBTs. With unipolar devices such as MOSFETs, the switch can be turned on to reduce conduction losses if necessary.

Figure 9. T-NPC boost PFC schematic diagram

Switch Q _xy is rated for 600V or 650V. Diode D _xBy is rated for 1200V. The component count is much lower than the original Vienna PFC. Conduction losses are much lower because only one diode is in series with the current loop at a time. However, since the "boost" diode is a 1200V device, the switching losses are slightly greater than the 600V diode. Since there are far fewer diodes, it is difficult to predict which topology will have the best efficiency. In fact, this T−NPC topology has better efficiency due to the smaller number of diodes. Figure 10 highlights the current path for one of the phases.

Figure 10. T−NPC boost PFC current path

(for storing and releasing energy modes) and phase voltage

The same feedback method can be used here with Clark and Park direct and inverse transforms to obtain PWM signals.

Since the two back-to-back switches share the same emitter or source pin node, the driver can drive the two back-to-back switches using the same PWM signal directly outside the control loop. Otherwise, depending on the sine wave sign (positive or negative), the corresponding switch needs to be driven. In this case, there are 6 switches to drive. This makes the PWM decoding scheme to drive the correct switches slightly more complex.

In both cases, the drive needs to be floating like the original Vienna.

One advantage of this topology is that there are far fewer active components. For the original Vienna, there are 6 active elements per phase. If we consider the body diode as part of the switch, there are only 4 active elements per phase in the T−NPC. Another advantage is lower conduction losses, making this topology more suitable for higher powers.

The main disadvantage of T−NPC is the need for 1200V diodes. This may offset the efficiency gains from lower conduction losses and may impact overall cost.

The T−NPC structure is also used as an inverter. In this case, the "boost" diode is replaced by a switch, as shown in Figure 11. Compared with PFC, the output direction is opposite. In this way, the T-NPC topology in which all switching devices can work in both directions can achieve bidirectional power transmission, and the transmission direction is defined by the control loop.

Figure 11. Bidirectional T−NPC boost PFC schematic diagram

The implementation of the bidirectional switch has changed again. The NPC topology uses two switches, one for each (positive or negative) sine wave half-cycle. The diode bridge is now a hybrid bridge, combining a diode and a switch, as shown in Figure 12. The two front-end diodes act as a kind of "gearbox" to switch positive or negative phases. Then, the diode connected to the output and the switching tube connected to ground are used as a boost switching unit. This is obvious since all topologies described here (Vienna, T-NPC and NPC) operate in boost mode.

Figure 12. NPC boost PFC schematic diagram

Switch Qxy is rated for 600V or 650V. All diodes (DxBy _and DxPy ₎ can also be rated for 600V or 650V. This will help reduce losses since devices rated for 1200V are not required. On the other hand, there are always two components {i.e. 1 diode and (1 diode or 1 switch)} in series with the current path. This NPC topology has higher conduction losses than T-NPC.

The same feedback method can be used here with Clark and Park direct and inverse transforms to obtain PWM signals.

The three switches here are floating and require floating gate drivers. The other 3 switches are grounded, they don't need floating drivers. This can be seen as an advantage, but this advantage can be affected by two reasons. First, depending on the power level, a Kelvin pin to the switch node may be required to drive the switch and improve efficiency. Secondly, in order to avoid current harmonics, the positive and negative sine waves are required to operate in symmetrical phases. This means that floating and grounded gate drive signals should have the same delay. Therefore, for this reason, float switches and ground switches often use the same drive schematic.

Depending on the sine wave polarity (positive or negative), the corresponding switch needs to be driven. This makes the PWM decoding scheme to drive the correct switches slightly more complex than the three-switch Vienna. The current path for this topology is shown in Figure 13.

Figure 13. Vienna boost PFC current path

(for storing and releasing energy modes) and phase voltage

Due to the absence of 1200V diodes, this topology has clear advantages in terms of losses and fewer components compared to the original Vienna. Driver pairing and latency matching are critical and can be seen as a drawback.

In this structure, replacing diodes with switches also makes the topology bidirectional, as shown in Figure 14. This structure is called A-NPC (Active Neutral Point Clamp).

Figure 14. Bidirectional NPC boost PFC schematic,

Also known as A-NPC boost PFC

6-switch, 6-Pack, or three-phase half-bridge inverter is widely used to drive motors, especially BLDC motors. When a motor is braked, energy is pulled from the motor's rotation and stored in the bus capacitor. The inverter works in reverse mode to provide power to the motor shaft. It is the same power flow as PFC. Power flows from the three-phase supply to the DC bus. In this circuit-breaker operating mode, the motor inductor acts as a "boost" inductor. The difference between this motor braking mode and the PFC mode lies in the control strategy given by the control loop. Therefore, the 6-switch PFC is the same as the motor inverter schematic in reverse mode (where the load is the source and vice versa). As shown in Figure 15, it is the simplest topology. All switches (Q _xy ) are 1200V devices. At any time, there is only one switch per phase in the power flow. This is an efficiency advantage that makes up for the shortcomings of a device rated for 1200V. It is also a two-level topology. So, modulation is straightforward. Today, some devices rated at 900V are also available in this topology. Those 900 V devices perform better than the 1200V devices. This helps reduce the disadvantages of switching devices above 650V.

Figure 15. Bidirectional three-phase half-bridge two-level boost PFC

Since we have 3 half-bridges connected to ground, it is much easier to build the driver using a half-bridge driver, and techniques like bootstrapping can be used to create a floating power supply. This simplifies the schematic using a well-known and widely used technique (in motor control applications). For better understanding, Figure 16 shows the return and forward paths. Since there are no intermediate points (since it is a two-level topology), the current path is not very obvious in this case.

Figure 16. Mode for storing and releasing (boosting) energy

Three-Phase Half-Bridge Boost PFC Current Path with Phase Voltage

Power modules are available for motor drives as well as PFC applications for ultra-high power applications. This topology is completely bidirectional in nature. As mentioned at the beginning of this article, the main disadvantages are mainly related to the objective advantages and disadvantages of the two-level topology.

Rather than using a dedicated three-phase topology with complex control (usually requiring a digital controller), a simpler alternative is to use three single-phase PFCs with a neutral connection, as shown in Figure 17. In this configuration, the neutral is essential if the system is unbalanced, even if three single-phase PFCs are connected to the load sharing control to distribute power equally among the three phases.

Figure 17. Three-phase PFC using 3 single-phase PFCs in parallel

Since single-phase PFC is so popular, it seems easier to use it this way. Some argue that the advantage of three independent converters is in the event of a failure: even if one fails, two are still available. That's true if the fault doesn't disrupt the grid. For example, if a short circuit fault occurs in the input stage, and this short circuit is somehow transmitted to the grid before the fuse blows. If it disrupts the grid and the neutral point changes during this fault, the full phase-to-phase voltage can be applied to the remaining PFC. To avoid failure, the remaining PFC will have to maintain this transient voltage, which increases PFC losses, size, and cost.

The advantage of this structure is that it is much simpler to design since single-phase PFC is widely used. However, the need to use a neutral wire makes the distribution network more expensive and suboptimal. Additionally, single-phase PFC cannot handle power above a few kilowatts. To handle higher power, parallel connection is required.

Table 1 summarizes the advantages and disadvantages of each topology with respect to the design criteria discussed previously.

Table 1. Summary of advantages and disadvantages of the common topologies discussed in this article

Three-phase PFC systems are complex, with many possible designs to meet the same electrical requirements, and a wide range of considerations and trade-offs. Finding the best solution for each application is not easy and requires system expertise at both system and component levels.

Onsemi can assist your three-phase power supply development and create more value. Our application notes, evaluation boards, simulation models, and expert applications team can help you gain insights into three-phase PFC. We support developers in choosing the right topology based on application requirements and finding the best components for each case.