Super-Junction MOSFETs vs. Traditional Trench Power MOSFETs in High-Efficiency AC-DC Switched-Mode Power Supplies
Contents
- Key Takeaways
- Why the MOSFET Structure Matters in an AC-DC Supply
- SJ MOSFET vs. Trench MOSFET: Practical Comparison
- Where Super-Junction MOSFETs Help in PFC and Server Power Supplies
- Does an SJ MOSFET Make a 96% Efficient Server PSU Possible?
- What Happens When Gate Charge and Miller Effects Are Ignored?
- Question 1: Why Are SJ MOSFETs Common in High-Efficiency Server Power Supplies?
- Question 2: Is the Lowest (R_{DS(on)}) Always the Best Choice?
- Final Selection: Start With the Switching Position
Key Takeaways
Super-junction MOSFETs are widely used in high-voltage switching stages because their charge-balanced drift region reduces the resistance normally associated with high blocking voltage.
Trench MOSFETs are a common choice for low- and medium-voltage power conversion, where reducing channel and cell resistance is especially important.
Lower (R_{DS(on)}) does not guarantee lower total losses. Gate charge, output charge, switching frequency, and the operating topology all matter.
A MOSFET does not make a power supply 80 PLUS Titanium by itself. The PFC stage, DC-DC converter, magnetic components, control strategy, and thermal design all contribute to system efficiency.
Why the MOSFET Structure Matters in an AC-DC Supply
A server power supply may have a 400 V DC bus after its power factor correction stage. The MOSFET switching this bus must block a high voltage when it is off, but it also needs to conduct current with as little loss as possible.
These two requirements pull in different directions.
In a conventional high-voltage silicon MOSFET, the drift region needs to be relatively thick and lightly doped to withstand the blocking voltage. That region contributes substantially to the device's on-resistance. Simply making the cell structure denser does not solve the problem as effectively as it does in a lower-voltage transistor.
This is where the super-junction MOSFET becomes useful.
An SJ device uses alternating P-type and N-type columns in its drift region. Their charge compensation allows a more heavily doped current path while maintaining high voltage-blocking capability. The result is a much better trade-off between breakdown voltage and specific on-resistance.
A traditional trench MOSFET takes a different approach. Its gate sits inside a vertical trench, increasing cell density and helping reduce on-state resistance. This structure is particularly useful in lower-voltage devices, such as MOSFETs used in DC-DC converters and synchronous rectification.
One distinction is worth keeping in mind:trenchdescribes the gate geometry, whilesuper-junctiondescribes the drift-region structure. They are not strictly opposite classifications, so the exact construction still needs to be checked in the manufacturer's documentation.
SJ MOSFET vs. Trench MOSFET: Practical Comparison
|
Design factor |
Super-junction MOSFET |
Traditional trench MOSFET |
|
Main structural feature |
Charge-balanced P/N columns in the drift region |
Vertical trench gate |
|
Common voltage range |
Typically 500–900 V silicon devices |
Commonly low- and medium-voltage devices, often below 250 V |
|
Main advantage |
Better breakdown-voltage and on-resistance trade-off at high voltage |
High cell density and low on-resistance in lower-voltage designs |
|
Typical SMPS position |
High-voltage boost PFC and primary-side switching |
Low-voltage DC-DC and synchronous rectification |
|
Important parameters |
(R_{DS(on)}), (Q_G), (Q_{GD}), (Q_{OSS}), (E_{OSS}) |
(R_{DS(on)}), (Q_G), (Q_{GD}), body-diode behavior |
|
Main design concern |
Output-capacitance behavior, switching transients, gate-drive optimization |
Conduction loss, gate drive, thermal limits, and reverse-recovery behavior |
These voltage ranges describe common product positioning, not absolute limits. The datasheet of the exact part number should decide whether a device is suitable for a given bus voltage.
Where Super-Junction MOSFETs Help in PFC and Server Power Supplies
In a conventional boost PFC stage, the MOSFET repeatedly switches against a relatively high DC bus. At this point, both conduction and switching losses matter.
The conduction component can be approximated by:
[P_{\text{cond}} \approx I_{\text{RMS}}^2 \times R_{DS(on)}]
The actual calculation must account for the switching duty cycle and current waveform. It should also use an appropriate hot (R_{DS(on)}), rather than relying only on the value measured at 25°C.
Switching loss is a separate issue. A MOSFET with low on-resistance may still dissipate considerable power during turn-on and turn-off. Output-capacitance energy and gate-drive requirements can become important, especially when switching frequency increases.
For this reason, engineers should not assume that every SJ MOSFET has a lower (Q_{OSS}) than every conventional MOSFET. SJ devices have a distinctive, nonlinear output-capacitance profile. Manufacturers have continued to optimize this behavior, but (Q_{OSS}) and (E_{OSS}) must be compared under the relevant voltage conditions.
A useful selection check is to compare:
(R_{DS(on)}) at the expected junction temperature.
Total gate charge (Q_G) and gate-drain charge (Q_{GD}).
Output charge (Q_{OSS}) and stored output energy (E_{OSS}).
Turn-on and turn-off energy at the intended drain current and voltage.
Switching frequency and actual PFC operating mode.
A newer SJ generation may improve several of these parameters without necessarily having the lowest value for every individual metric.
Does an SJ MOSFET Make a 96% Efficient Server PSU Possible?
It can help, but the device is only one part of the result.
For example, the 80 PLUS Titanium requirements include efficiency targets at defined input-voltage and load conditions. In the 230 V internal redundant category, the 50% load target is 96%. Other input and PSU categories have different thresholds.
A server PSU also contains a PFC stage and one or more DC-DC conversion stages. Losses in the bridge or totem-pole PFC circuit, switching devices, transformers, inductors, rectifiers, control electronics, and cooling system all affect the final measurement.
An SJ MOSFET can reduce losses in an appropriate high-voltage switching position, making the thermal and efficiency targets easier to reach. It cannot guarantee that the complete power supply will pass a particular efficiency certification.
What Happens When Gate Charge and Miller Effects Are Ignored?
Fast switching is useful, but it creates another problem: parasitic turn-on.
When the drain voltage changes rapidly, current flows through the MOSFET's gate-drain capacitance:
[I_{\text{Miller}} = C_{GD} \times \frac{dV_{DS}}{dt}]
This current can raise the gate voltage of a device that is supposed to remain off. In a half-bridge or other synchronous switching circuit, unintended turn-on may cause cross-conduction and excessive device heating.
The relevant capacitance parameters include:
(C_{RSS}), which represents the reverse transfer capacitance and is closely related to (C_{GD}).
(C_{ISS}), which is the input capacitance, approximately (C_{GS}+C_{GD}) under the datasheet's measurement conditions.
(Q_{GD}), which helps describe the gate charge associated with the Miller region.
The (C_{RSS}/C_{ISS}) ratio can provide a useful initial comparison, but it is not enough to predict false turn-on by itself. The actual result depends on drain-voltage slew rate, gate impedance, threshold voltage, source inductance, and driver behavior.
A few design details deserve attention. Select the turn-off gate resistance based on measured switching behavior rather than choosing it only to achieve the fastest edge. Keep the gate loop short, minimize common-source inductance, and use a Kelvin-source connection when the selected package provides one. A Miller clamp or negative turn-off bias may also be appropriate.
The goal is controlled switching, not simply the highest possible (dv/dt).
Question 1: Why Are SJ MOSFETs Common in High-Efficiency Server Power Supplies?
Their charge-compensated drift region provides a much better high-voltage on-resistance trade-off than a conventional silicon drift region. This makes them attractive for 600 V and 650 V switching applications, including boost PFC stages.
Their contribution to overall efficiency still depends on gate charge, output-capacitance energy, switching conditions, and the rest of the power-conversion circuit.
Question 2: Is the Lowest (R_{DS(on)}) Always the Best Choice?
No. A device with lower (R_{DS(on)}) may have higher gate charge or less favorable output-capacitance behavior. At high switching frequencies, these differences can offset the conduction-loss advantage.
Compare total loss at the intended operating point, including light-load and full-load conditions. For server power supplies, light-load performance can be particularly important because efficiency is evaluated at multiple load levels.
Final Selection: Start With the Switching Position
For a high-voltage PFC stage, a properly selected SJ MOSFET is often the more practical silicon option. Its drift-region design addresses the resistance penalty associated with high blocking voltage.
For a low-voltage DC-DC output stage, a trench MOSFET may offer the better combination of low resistance, package size, and cost.
Before approving a device or a second source, compare its complete part number, voltage rating, hot (R_{DS(on)}), gate-charge curves, (Q_{OSS}), (E_{OSS}), thermal resistance, package pinout, and switching-loss data. Also confirm that the replacement has been evaluated in the actual gate-drive circuit.