# Transformers and power electronics: Science and limits

Why transformers are nearly perfect, and drives stumble at light load

Source page: https://www.materialinsights.org/energy/e_power_electronics/science/

What this tab covers What physically happens in the conversion, step by step, the science behind it, and the hard limits that physics sets on how well it can ever work.

Transformers change AC voltage using magnetic induction in an iron core; power electronics use fast semiconductor switches to turn AC into DC (rectifiers), DC into AC (inverters) and one voltage into another (converters and chargers). Most units are 92-99% efficient at normal load, and transformers are typically around 99%.

#### The conversion, step by step
**Transformer steps voltage up or down**
Electrical (AC, one voltage) → Electrical (AC, another voltage) · Typically above 99% (ABB); a standard 2,500 kVA unit is about 98.95% at full load and 99.3% at half load (calculated)^[4](https://library.e.abb.com/public/929bc703922d49069e6920d5ce6ab9c1/53-57%202m522_EN_72dpi.pdf)**Rectifier**
Electrical (AC) → Electrical (DC) · High, but no single published figure found; included within drive and charger figures^[5](https://www.energy.gov/sites/prod/files/2014/04/f15/motor_tip_sheet11.pdf)**Inverter (for example a motor drive)**
Electrical (DC) → Electrical (AC at chosen frequency and voltage) · Whole drive (rectifier plus inverter) 95-97% at full load for 5-200 hp units^[5](https://www.energy.gov/sites/prod/files/2014/04/f15/motor_tip_sheet11.pdf)**DC fast charger (rectifier plus DC-DC converter)**
Electrical (grid AC) → Electrical (DC at battery voltage) · 92.3% AC to DC measured on a 50 kW unit (INL, 2016)^[15](https://avt.inl.gov/sites/default/files/pdf/presentations/INL_DCFastChargerInfrastructure.pdf)**Losses**
Electrical → Heat in iron cores, copper windings and semiconductor chips · Removed by oil, air or liquid cooling^[4](https://library.e.abb.com/public/929bc703922d49069e6920d5ce6ab9c1/53-57%202m522_EN_72dpi.pdf)

#### The science

A transformer has two coils wound on a shared iron core. Alternating current in the first coil makes a magnetic field in the core that keeps reversing; that changing field induces a voltage in the second coil (Faraday's law). The ratio of turns sets the ratio of voltages, while power in roughly equals power out, so stepping voltage up steps current down. Because only a changing field induces voltage, a transformer cannot work on DC. Power electronics fill that gap: transistors (IGBTs and MOSFETs made of silicon or silicon carbide) switch on and off thousands of times a second, chopping a voltage into pulses whose average can be shaped into DC, a sine wave of any frequency, or a different DC level, then smoothed by capacitors and inductors. Losses come from the resistance of copper (rising with the square of current), magnetising the iron core every cycle (present even with no load), and the brief moments when a switch is half-on.

#### Limits set by science

| Limit | Value | Why it exists | Best today |
|---|---|---|---|
| **No thermodynamic ceiling** | Not applicable^[4](https://library.e.abb.com/public/929bc703922d49069e6920d5ce6ab9c1/53-57%202m522_EN_72dpi.pdf) | Both devices convert electricity to electricity, so there is no Carnot-type limit. Efficiency is set by resistance, magnetic and switching losses, which better materials can keep reducing. | Transformers are typically above 99% efficient. |
| **No-load (core) loss** | 2,782 W in a standard 2,500 kVA unit, falling to 1,747 W with high-permeability steel^[4](https://library.e.abb.com/public/929bc703922d49069e6920d5ce6ab9c1/53-57%202m522_EN_72dpi.pdf)^[2](https://www.energy.gov/node/4839605) | The core must be magnetised every cycle while energised, whatever the load. At low average loading (such as a solar farm running at 20-30% of peak on average) this fixed loss becomes a large share of the total. | Better grain-oriented steel cut no-load loss by about 37% in this example (calculated); amorphous metal cores go further. |
| **Load (copper) loss and part-load efficiency** | Full-load efficiency of the standard 2,500 kVA unit about 98.95%, rising to about 99.31% at half load (calculated, unity power factor)^[4](https://library.e.abb.com/public/929bc703922d49069e6920d5ce6ab9c1/53-57%202m522_EN_72dpi.pdf) | Copper loss scales with current squared, so at half load it is a quarter of its full-load value; efficiency peaks where copper loss equals core loss. | Low-loss designs trade more copper and steel (higher price) for lower losses. |
| **Light-load penalty in drives** | Drive efficiency 35-61% at 1.6% load versus 95-97% at full load^[5](https://www.energy.gov/sites/prod/files/2014/04/f15/motor_tip_sheet11.pdf) | Fixed losses in control circuits, cooling and switching dominate when little power flows. | Above about 25% load, drives stay at 88-95% or better. |
| **Materials and supply** | Power transformer prices up about 75% since 2019; large units can take up to four years to obtain^[1](https://www.iea.org/reports/building-the-future-transmission-grid/executive-summary) | Grain-oriented electrical steel, copper and specialist manufacturing capacity are bottlenecks rather than physics. | Not applicable |

#### Sources

1. [Building the Future Transmission Grid: executive summary](https://www.iea.org/reports/building-the-future-transmission-grid/executive-summary) International Energy Agency, 2025
2. [DOE Finalizes Energy Efficiency Standards for Distribution Transformers (press release, 4 April 2024)](https://www.energy.gov/node/4839605) US Department of Energy, 2024
3. [Transforming revenue (ABB Review 2/2015)](https://library.e.abb.com/public/929bc703922d49069e6920d5ce6ab9c1/53-57%202m522_EN_72dpi.pdf) ABB, 2015
4. [Adjustable Speed Drive Part-Load Efficiency, Motor Systems Tip Sheet #11](https://www.energy.gov/sites/prod/files/2014/04/f15/motor_tip_sheet11.pdf) US Department of Energy, 2012
5. [DC fast charger infrastructure testing (ABB Terra 53 efficiency)](https://avt.inl.gov/sites/default/files/pdf/presentations/INL_DCFastChargerInfrastructure.pdf) Idaho National Laboratory, 2016

## Other sections

- [Summary](https://www.materialinsights.org/energy/e_power_electronics/index.md)
- [Overview](https://www.materialinsights.org/energy/e_power_electronics/overview/index.md)
- [Correlation](https://www.materialinsights.org/energy/e_power_electronics/correlation/index.md)
- [Value chain](https://www.materialinsights.org/energy/e_power_electronics/valuechain/index.md)
- [Unit economics](https://www.materialinsights.org/energy/e_power_electronics/economics/index.md)
- [Engineering methods](https://www.materialinsights.org/energy/e_power_electronics/methods/index.md)
- [Equipment and suppliers](https://www.materialinsights.org/energy/e_power_electronics/equipment/index.md)
- [Relationships](https://www.materialinsights.org/energy/e_power_electronics/links/index.md)