How Custom Transformers Improve Power Efficiency

Where Transformer Losses Come From

A transformer transfers electrical energy from one circuit to another through magnetic coupling. In an ideal transformer, 100% of the input power reaches the output. In a real transformer, some energy is lost as heat inside the component. These losses fall into two major categories: core losses and copper losses. Understanding each type is the first step toward minimizing them through custom design.

Core Losses

Core losses occur in the magnetic core material itself, driven by the alternating magnetic flux that the transformer produces during normal operation. Core losses have two primary components.

Hysteresis Loss

Every time the magnetic flux reverses direction in the core, the magnetic domains must realign. This realignment requires energy, which is dissipated as heat. The energy lost per cycle is proportional to the area enclosed by the B-H hysteresis loop of the core material. Materials with narrow hysteresis loops (low coercivity) have lower hysteresis losses.

Hysteresis loss scales linearly with frequency: double the operating frequency and the hysteresis loss doubles, because the domains reverse direction twice as often per second. The loss also increases with the peak flux density, following a power law: P_h is proportional to f × B_maxⁿ, where n is typically between 1.6 and 2.5 depending on the material.

Eddy Current Loss

The changing magnetic flux induces circulating currents (eddy currents) within the conductive core material. These currents flow in closed loops within the core cross-section and dissipate energy through I²R heating. Eddy current loss scales with the square of both frequency and flux density: P_e is proportional to f² × B_max² × t², where t is the thickness of the core lamination or tape.

This is why transformer cores are built from thin laminations or wound from thin tape rather than solid blocks. Reducing the lamination thickness by half reduces eddy current loss by a factor of four. Ferrite cores, being electrical insulators, have negligible eddy current loss, which is one of their key advantages at high frequencies.

Copper Losses

Copper losses (also called winding losses or I²R losses) are the heat generated by current flowing through the resistance of the transformer windings. These losses apply to both the primary and secondary windings.

DC Resistance Loss

The fundamental copper loss is P_cu = I_rms² × R_dc, where I_rms is the RMS current in the winding and R_dc is the DC resistance. This loss is present at all frequencies. Reducing DCR by using heavier gauge wire directly reduces this loss component.

AC Resistance Effects

At higher frequencies, two effects increase the effective resistance of the winding beyond the DC value.

Skin effect causes current to concentrate near the surface of the conductor. The effective conducting cross-section shrinks as frequency increases, raising the AC resistance. At 100 kHz, the skin depth in copper is approximately 0.21 mm. Wire diameters larger than twice the skin depth carry current inefficiently at that frequency.

Proximity effect occurs when adjacent turns in a multi-layer winding interact magnetically. The magnetic field from each turn induces eddy currents in neighboring turns, further increasing the effective resistance. Proximity effect losses increase dramatically with the number of winding layers, scaling roughly as the square of the layer count.

For high-frequency transformers, techniques such as Litz wire (multiple individually insulated strands woven together), foil windings, and interleaved primary/secondary layers are used to mitigate AC resistance effects.

How Custom Design Reduces Losses

Core Material Optimization

Every core material has a frequency range where it performs best. A custom transformer is designed with the core material matched to the actual operating frequency and flux density of your application.

Core Material	Optimal Frequency	Bsat	Core Loss Characteristic
Grain-Oriented Silicon-Iron	50 Hz to 10 kHz	1.6 to 2.0 T	Lowest loss at line frequency
Non-Oriented Silicon-Iron	50 Hz to 1 kHz	1.5 to 1.8 T	Good for rotating field applications
Amorphous Metal	50 Hz to 100 kHz	1.5 T	70 to 80% lower core loss than silicon-iron
Nanocrystalline	1 kHz to 500 kHz	1.2 T	Excellent across wide frequency range
MnZn Ferrite	20 kHz to 2 MHz	0.35 to 0.50 T	Lowest loss above 100 kHz

A 60 Hz power transformer uses grain-oriented silicon-iron because the high saturation flux density allows maximum power transfer in a given core volume. A 200 kHz switching transformer uses MnZn ferrite because the core loss at that frequency would be prohibitive in silicon-iron. A custom design ensures this matching is precise, because even within a material family, different grades have different loss characteristics.

Wire Gauge Optimization

The wire gauge determines the winding resistance and therefore the copper loss. Thicker wire has lower resistance but fills the winding window faster, which may require a larger core. The optimal wire gauge balances copper loss against core size and cost.

In a custom design, the wire gauge is selected based on the actual RMS current in each winding, the operating frequency (which determines whether AC resistance effects are significant), and the available winding area. This frequently leads to different wire gauges on the primary and secondary windings, each optimized for its own current level.

Turns Ratio Design

The turns ratio (N_p/N_s) determines the voltage transformation ratio. But the absolute number of turns also affects performance. More turns increase inductance (reducing magnetizing current) but also increase winding resistance and leakage inductance. Fewer turns reduce copper losses but increase the flux density in the core, potentially pushing it closer to saturation.

The optimal number of turns depends on the core material, core size, operating frequency, and the acceptable levels of magnetizing current and core loss. Custom design finds this balance for your specific operating point rather than using a general-purpose compromise.

Winding Geometry

How the windings are arranged on the core affects both copper loss and leakage inductance. Interleaving primary and secondary layers (P-S-P-S instead of P-P-S-S) reduces leakage inductance and proximity effect losses. Careful control of winding width and layer count keeps the AC resistance increase manageable.

In toroidal transformers, the even distribution of windings around the full core circumference provides inherently good coupling and low leakage. In bobbin-wound transformers, sectional bobbins can separate windings for safety isolation while maintaining acceptable coupling.

Thermal Management

The total loss (core + copper) heats the transformer. The temperature rise depends on the total loss, the thermal resistance of the component, and the ambient temperature. For every 10°C increase in operating temperature, the expected life of the wire insulation is roughly halved (following the Arrhenius relationship).

• Toroidal geometry provides the best surface-to-volume ratio for natural convection cooling
• Potting or encapsulation improves thermal conductivity from windings to the surface
• Larger core sizes run cooler because the loss density (watts per cubic centimeter) decreases
• Higher-temperature wire insulation classes (Class F: 155°C, Class H: 180°C) provide thermal margin

Real-World Efficiency Impact

The financial impact of transformer efficiency depends on the operating profile. A transformer in a data center UPS runs 24/7 at moderate to high load. A transformer in a battery charger runs intermittently. The annual energy cost of transformer losses is: Cost = P_loss × Hours × Electricity Rate.

Scenario	Rating	Efficiency	Loss	Annual Cost (at $0.15/kWh)
Standard 1 kW transformer	1,000 W	95%	50 W	$65.70
Optimized custom 1 kW	1,000 W	98%	20 W	$26.28
Standard 5 kW transformer	5,000 W	95%	250 W	$328.50
Optimized custom 5 kW	5,000 W	98%	100 W	$131.40

At the 5 kW level, the optimized transformer saves roughly $200 per year per unit in electricity costs. For an installation with 100 units, the annual savings reach $20,000. Over a 10-year equipment life, the total savings cover the initial premium for custom-designed transformers many times over.

When Custom Transformers Make Sense

• High-volume production where the per-unit efficiency improvement compounds across thousands of units
• Continuous-duty applications (servers, telecom, industrial) where the transformer runs 8,000+ hours per year
• Space-constrained designs where optimized performance from a smaller transformer eliminates the need for additional cooling
• Regulatory requirements (DoE efficiency standards, EU Ecodesign Directive) that mandate minimum efficiency levels
• High-reliability applications where reduced thermal stress extends component life

The Design Process

1. Define the operating conditions: Input voltage, output voltage and current, operating frequency, duty cycle, ambient temperature
2. Select the core material: Match to frequency and loss requirements
3. Size the core: Calculate the required flux density at the operating frequency and ensure the core can handle it with margin below saturation
4. Calculate turns: Determine the primary and secondary turns for the desired voltage ratio and acceptable flux density
5. Select wire gauge: Size each winding for its RMS current, accounting for skin and proximity effects at the operating frequency
6. Verify thermal performance: Calculate total losses and confirm the temperature rise stays within the insulation class limits
7. Prototype and test: Build samples, measure actual efficiency, and iterate if needed