An uninterruptible power supply must do two things exceptionally well: protect the load and minimize the energy consumed while providing that protection.
In a double-conversion UPS, incoming AC power is converted to DC and then reconstructed as regulated AC. This architecture isolates critical equipment from many utility disturbances, but every conversion stage produces some loss. In a large data center or industrial facility, even a small improvement in UPS efficiency can reduce electricity use, cooling demand and operating cost.
Silicon carbide devices and advanced insulated-gate bipolar transistors are helping UPS manufacturers reduce those losses without giving up the power quality and availability required by critical loads.
Where does UPS power loss come from?
A conventional online UPS contains several power-handling components:
- Input rectifier or converter
- DC link
- Battery interface
- Output inverter
- Static bypass
- Magnetic components
- Filters
- Cooling fans
- Control and auxiliary power supplies
Power is lost in semiconductor conduction, switching transitions, inductors, transformers, capacitors, conductors and cooling systems.
The semiconductor devices in the rectifier and inverter are particularly important because they must carry the UPS load continuously while switching thousands of times per second.
Total semiconductor loss can be divided into two broad categories.
Conduction loss
Conduction loss occurs while a semiconductor is carrying current. It depends on the device’s on-state voltage or resistance, load current, duty cycle and junction temperature.
A device with lower on-state voltage reduces the power converted to heat during each conduction interval.
Switching loss
Switching loss occurs while a device changes between its off and on states. Voltage and current overlap briefly during each transition, producing an energy loss.
Total switching loss increases with:
- Switching frequency
- Bus voltage
- Load current
- Turn-on and turn-off energy
- Diode reverse recovery
- Parasitic inductance
- Gate-drive behavior
Reducing either form of loss can improve UPS efficiency, but semiconductor design involves tradeoffs. A device optimized for very low conduction loss may not provide the lowest switching loss, and vice versa. The best device depends on the UPS topology, voltage, switching frequency and load profile.
Why UPS efficiency matters
UPS losses become heat inside the electrical room. That creates two operating costs:
- The electricity lost in the UPS
- The additional electricity needed to remove the heat
Consider a UPS delivering a constant 1 MW load. At 96% efficiency, its approximate input is 1.042 MW and its internal loss is about 41.7 kW. At 97.5% efficiency, the input is approximately 1.026 MW and the loss falls to about 25.6 kW.
That 1.5-percentage-point improvement reduces UPS loss by roughly 16 kW at this operating point—before accounting for the associated reduction in cooling energy.
Actual savings depend on the load profile, operating mode, redundancy configuration, electricity rate and cooling-system efficiency. This is why an efficiency curve is more useful than a single peak-efficiency number.
The role of IGBTs in modern UPS systems
The insulated-gate bipolar transistor, or IGBT, combines a voltage-controlled gate with strong high-voltage and high-current capability. IGBTs have been widely adopted in medium- and high-capacity UPS converters and inverters.
Fuji Electric’s UPS history illustrates the effect of this transition. Compared with earlier thyristor-based systems, its adoption of IGBTs for both the converter and inverter increased conversion efficiency and substantially reduced equipment size.
IGBT technology has continued to advance. Improvements include:
- Thinner semiconductor structures
- Optimized trench-gate designs
- Improved field-stop layers
- Lower on-state voltage
- Reduced switching energy
- Improved diode performance
- Higher allowable junction temperature
- Lower-inductance packaging
- Reverse-conducting and reverse-blocking functions
These developments allow UPS designers to reduce losses without simply increasing semiconductor die area.
How advanced IGBTs improve efficiency
Better conduction and switching tradeoffs
IGBT designers generally seek to reduce on-state voltage without creating an unacceptable increase in switching loss. Newer chip structures provide greater control over the carrier distribution inside the device, improving this tradeoff.
Fuji Electric’s 7th-generation X-Series IGBT technology uses thinner and optimized chip structures to reduce inverter loss. In one published comparison involving 1,200 V, 75 A devices operating at 8 kHz, the X-Series reduced inverter loss by 10% relative to the preceding V-Series under the stated test conditions.
This is a device and test-condition comparison, not a guarantee that every complete UPS will gain 10% efficiency. System performance also depends on topology, controls, magnetics, cooling and the operating point.
Improved freewheeling diode performance
The switching device does not operate alone. In many converter topologies, current commutates between the IGBT and a freewheeling diode.
Diode forward-voltage and reverse-recovery behavior affect:
- Conduction loss
- Turn-on loss
- Voltage overshoot
- Electromagnetic interference
- Device stress
An improved diode can therefore increase the benefit obtained from an advanced IGBT.
Reverse-conducting IGBTs
A reverse-conducting IGBT integrates IGBT and freewheeling-diode functions into one chip. This can increase power density and simplify chip placement inside the module.
Fuji Electric has developed X-Series RC-IGBT modules that combine 7th-generation IGBT and diode functionality, supporting higher power density and improved performance in applicable industrial power converters.
Reverse-blocking IGBTs
Most conventional IGBTs are designed primarily to block voltage in one direction. A reverse-blocking IGBT, or RB-IGBT, is designed to block voltage in both directions.
In suitable three-level converter circuits, an RB-IGBT can reduce the number of series-connected semiconductor elements in a current path. Fewer conducting junctions can mean lower conduction loss.
Fuji Electric has applied proprietary RB-IGBT technology in advanced T-type neutral-point-clamped conversion circuits used in high-efficiency UPS systems.
What makes silicon carbide different?
Conventional power semiconductors are made from silicon. Silicon carbide, or SiC, is a wide-bandgap semiconductor material with physical properties that support:
- Higher electric-field strength
- Higher-voltage operation with thinner active regions
- Faster switching
- Lower switching energy
- Lower reverse-recovery loss
- Higher-temperature capability
- Higher switching frequency
The two SiC devices most relevant to UPS designs are SiC Schottky barrier diodes and SiC MOSFETs.
SiC Schottky barrier diodes
A SiC Schottky barrier diode, or SiC-SBD, has very little reverse-recovery charge compared with a conventional silicon bipolar diode.
When a silicon diode changes from conducting to blocking, stored charge must be removed. The resulting reverse-recovery current increases loss in both the diode and the switching transistor. It can also contribute to current spikes, voltage overshoot and electromagnetic noise.
Replacing the diode with a suitable SiC-SBD can:
- Reduce diode recovery loss
- Reduce IGBT turn-on loss
- Lower current overshoot
- Reduce thermal stress
- Support higher switching frequency
- Simplify management of switching transients
A combination of silicon IGBTs and SiC diodes is often described as a hybrid SiC solution. It captures an important portion of SiC’s efficiency benefit while retaining the established high-current capability and cost structure of IGBTs.
Fuji Electric has developed 1,200 V SiC-SBD products for systems including data-center and telecommunications power supplies, targeting lower forward loss and improved surge-current capability.
SiC MOSFETs
A SiC MOSFET replaces the silicon IGBT as the actively controlled switching device. It can switch more rapidly and with lower switching loss, particularly at light and medium loads where an IGBT’s characteristic voltage drop can represent a significant penalty.
Potential UPS benefits include:
- Lower turn-on and turn-off loss
- Reduced part-load loss
- Higher switching frequency
- Smaller magnetic components
- Smaller filters
- Reduced heat-sink requirements
- Higher power density
SiC MOSFETs also conduct reverse current through their channel when controlled appropriately. Their behavior differs from both silicon MOSFETs and IGBTs, so the gate drive, dead time and protection system must be designed specifically for the device.
Fuji Electric notes that faster switching can reduce power loss and that higher switching frequencies can contribute to smaller power-conversion equipment.
Why faster switching can make a UPS smaller
Higher switching frequency moves the converter’s switching ripple to a higher frequency. This can reduce the inductance and capacitance needed to filter it, allowing smaller reactors and filter components.
A smaller magnetic component can provide several system-level benefits:
- Reduced copper loss
- Reduced core material
- Lower weight
- Smaller enclosure
- Shorter power paths
- Greater power density
However, frequency cannot be increased without limit. Higher frequency increases the number of switching events per second and may increase losses in semiconductors, magnetic cores, capacitors and conductors.
SiC expands the useful design range by reducing the energy lost in each switching transition. The optimum switching frequency remains a system-level decision.
Three-level topologies reduce semiconductor stress
A traditional two-level inverter switches each phase output between the positive and negative DC rails. A three-level topology adds an intermediate voltage state.
Depending on the circuit, this can:
- Reduce the voltage switched by an individual device
- Lower switching loss
- Reduce output-voltage steps
- Reduce filter requirements
- Improve waveform quality
- Support a higher DC bus with available device ratings
The semiconductor benefit depends on the specific topology and current path. A poorly optimized three-level circuit can add conduction devices and control complexity. Advanced devices such as RB-IGBTs can help reduce that penalty.
Fuji Electric’s advanced T-type neutral-point-clamped architecture combines RB-IGBT and SiC technology to reduce conversion loss in applicable UPS systems. Its 7400WX-T3U achieved a published efficiency of 97.4% and at least 96% efficiency at a 25% load factor under the product’s stated conditions.
Part-load efficiency is essential
UPS systems rarely operate at their nameplate rating continuously.
Common reasons include:
- N+1 or 2N redundancy
- Future capacity reserved for growth
- Variable IT or production loads
- Conservative initial sizing
- Modular systems operating with spare capacity
A system designed for redundancy may place each UPS module at 25%, 40% or 50% load during normal operation. Peak efficiency at 100% load may therefore say little about annual energy consumption.
SiC and advanced IGBT technologies can improve part-load efficiency by reducing switching, conduction and auxiliary losses. System controls can add another layer of improvement by operating the appropriate number of modules near a more efficient load point.
Fuji Electric’s 7400WX-T3U includes an optimum-load operating mode that adjusts the number or loading of operating UPS units to improve overall system efficiency.
When evaluating a UPS, request efficiency at:
- 10% load, if available
- 25% load
- 50% load
- 75% load
- 100% load
Apply those values to the facility’s expected annual load distribution.
Reduced loss lowers cooling requirements
Every kilowatt not lost in the UPS is a kilowatt that does not have to be removed as heat.
Lower semiconductor loss can reduce:
- Heat-sink size
- Cooling airflow
- Fan power
- Electrical-room cooling demand
- Local hot spots
- Component temperature
- Acoustic noise
Lower temperature can also support reliability because many power-electronic component aging mechanisms accelerate as temperature increases.
This benefit is not automatic. Higher-temperature-capable SiC devices should not be used as a reason to expose capacitors, fans, control boards and batteries to excessive heat. The entire UPS contains components with different thermal limits.
The better design objective is often to use the semiconductor’s lower loss and temperature capability to create thermal margin—not simply to operate every component hotter.
Higher power density can reduce facility footprint
When lower semiconductor loss is combined with smaller magnetics and reduced cooling requirements, a UPS can deliver more power from a smaller footprint.
This is valuable in:
- Data centers
- Semiconductor plants
- Hospitals
- Financial facilities
- Telecommunications sites
- Industrial control rooms
Floor space carries a significant opportunity cost in these facilities. A more compact UPS can leave additional space for production equipment, IT racks, batteries or future capacity.
Fuji Electric has demonstrated a SiC-based research power unit with 81 kW/L power density and 99.5% efficiency at a 37 kW rated output under the prototype’s test conditions. [6] Although this is not a complete UPS specification, it illustrates how low-loss SiC devices can support compact power-conversion hardware.
Efficiency mode and semiconductor efficiency are different
Many UPS products offer an economy, high-efficiency or bypass-based operating mode. Under acceptable utility conditions, such a mode supplies the load through a bypass path rather than continuously processing all power through the rectifier and inverter.
This can produce efficiency higher than normal double-conversion operation. For example, Fuji Electric’s UPS7500WX lists 98.5% efficiency in its HE mode.
It is important to distinguish:
- Efficiency gained from lower-loss semiconductors in the active conversion path
- Efficiency gained by bypassing some or all of that conversion path
The operating modes provide different levels of power conditioning and may have different transfer behavior. The appropriate mode depends on:
- Load sensitivity
- Utility power quality
- Transfer-time tolerance
- Harmonic and power-factor requirements
- Facility risk policy
- Applicable standards
Semiconductor improvements increase the efficiency of online operation, helping reduce the energy penalty when the application requires continuous power conversion.
Efficiency improvements can support reliability
Lower internal loss may reduce thermal stress on:
- Power semiconductor junctions
- DC-link capacitors
- Magnetic components
- Busbars
- Cooling fans
- Nearby control electronics
Reduced fan demand may also extend fan life or allow redundant cooling arrangements to operate with greater margin.
However, efficiency and reliability must be validated together. Faster switching can create:
- Higher (dv/dt)
- Higher (di/dt)
- Voltage overshoot
- Common-mode current
- Electromagnetic interference
- Insulation stress
- More demanding gate-drive requirements
Successful SiC application depends on low-inductance packaging, compact busbars, suitable gate drivers, effective protection and careful EMC design. A fast semiconductor installed in a layout developed for slower devices may not deliver the expected reliability or efficiency.
Packaging is part of the efficiency equation
A semiconductor chip cannot perform independently of its package.
Package features affect:
- Electrical resistance
- Parasitic inductance
- Switching overshoot
- Heat transfer
- Current sharing
- Maximum junction temperature
- Power-cycling capability
Lower-inductance packages allow faster switching with less voltage overshoot. Improved insulating substrates and thermal interfaces move heat away from the semiconductor more effectively.
Fuji Electric’s 7th-generation X-Series power modules combine optimized chips with lower-inductance packaging and improved heat dissipation.
This coordinated chip-and-package approach is essential because reducing device switching loss is of limited value if parasitic inductance forces the designer to slow the device substantially.
Choosing between SiC and advanced IGBT solutions
SiC is not automatically the most economical option for every UPS.
Advanced IGBTs may remain attractive when:
- Power levels are high
- Switching frequency is moderate
- Conduction loss dominates
- Cost sensitivity is significant
- Proven silicon platforms meet the efficiency target
- Existing manufacturing and service infrastructure is important
Hybrid SiC solutions may be effective when:
- Diode reverse recovery is a major loss mechanism
- Improved switching performance is needed
- A full SiC design is not required
- The designer wants to balance performance and cost
All-SiC solutions may be attractive when:
- Very low switching loss is important
- Part-load efficiency is a priority
- High switching frequency enables smaller magnetics
- Power density is valuable
- Cooling capacity or footprint is constrained
The comparison should be made at the system level. Semiconductor cost should be weighed against savings in magnetics, heat sinks, fans, enclosure volume, cooling infrastructure and lifetime electricity use.
How to compare high-efficiency UPS systems
When evaluating a UPS that uses SiC or advanced IGBT technology, ask for:
- Efficiency in normal online mode
- Efficiency at 25%, 50%, 75% and 100% load
- Efficiency in optional economy or high-efficiency modes
- Whether auxiliary and fan power are included
- Test voltage, power factor and environmental conditions
- Heat rejection at representative loads
- Input power factor and harmonic performance
- Output waveform performance
- Overload and short-circuit capability
- Semiconductor topology
- Redundancy and module-loading strategy
- Transfer behavior between operating modes
- Cooling-system redundancy
- Expected capacitor and fan service life
- Applicable efficiency-test standards
Avoid comparing values measured under different modes or test conditions. A bypass-mode efficiency from one product should not be compared directly with the online double-conversion efficiency of another.
Calculate annual loss—not just nominal efficiency
The most useful metric is the annual energy lost under the expected operating profile.
For each load interval:
[
P_{\text{loss}} = P_{\text{out}}\left(\frac{1}{\eta}-1\right)
]
Annual loss energy can then be estimated by multiplying the loss at each operating point by the number of hours spent there.
Include:
- Normal load variation
- Redundancy configuration
- Planned future growth
- Battery charging
- Economy-mode operating hours
- Cooling energy
- Module staging
- Maintenance bypass periods
This lifecycle analysis may show that a product with a small peak-efficiency advantage offers less annual benefit than one with a flatter part-load curve.
Power semiconductors and UPS design are evolving together
The efficiency of a UPS is not determined by one device alone. It results from the interaction of semiconductor technology, circuit topology, gate control, packaging, magnetics, cooling and system operation.
Advanced IGBTs continue to improve the balance between conduction loss, switching loss, power density and cost. SiC-SBDs reduce reverse-recovery loss, while SiC MOSFETs enable faster switching and smaller passive components. Three-level circuits and intelligent module controls turn those device improvements into system-level savings.
Fuji Electric develops both UPS systems and the power semiconductors used in power-conversion equipment. Its portfolio includes advanced IGBT, hybrid SiC and SiC technologies intended for high-efficiency converters and uninterruptible power supplies.
By evaluating performance across the actual load profile—and separating online conversion efficiency from bypass-mode efficiency—facility designers can choose a UPS that protects critical loads while reducing energy use, cooling demand and total operating cost.