Toroidal Transformer Design Calculator: Turns, Losses and Inrush Current

Measure the dimensions with calipers; this is not a workaround. Bare toroidal cores often arrive without a useful part number, and even the catalogs consulted here publish diameters and weight but not the height, which can vary between batches. Three measurements are enough to characterize the core.

The magnetic path length is one of the few quantities that can be obtained exactly from toroidal geometry: it is the mean circumference, without the approximations required for an E-lamination stack. Weight also follows from the geometry, and it is worth comparing the calculated value with a scale reading. If they disagree, the stacking factor of your core differs from that of the selected steel grade.

For the mean turn length I provide two values: McLyman’s formula, which is used in the calculations, and the geometric perimeter of the bare core cross-section. The first is larger because it accounts for winding build-up; the second is the theoretical lower limit. Their difference indicates how sensitive winding resistance — and therefore copper loss — is to the way the transformer is wound.

The calculation is the same as for a laminated transformer — turns per volt from Faraday’s law — but the result differs because the design flux density is higher. For the same core cross-section, operating at 1.65 T instead of 1.2 T requires roughly one quarter fewer turns, which means less copper, lower resistance, and lower copper loss.

The secondary allowance is lower than for a conventional laminated transformer because toroidal regulation is better: manufacturers quote values from about 15% for 50 VA sizes down to roughly 6–8% above 150 VA. Read the calculated regulation in Card 4 and use it here to refine the design.

This is the card that makes toroidal design different from the others. With a conventional laminated core, the copper either fits the window or it does not, and the criterion is a percentage. Here the limiting quantity is a physical dimension: the hole shrinks with every layer, and once it becomes too small the winding shuttle — or your fingers, when winding by hand — can no longer pass through.

The last value is the one to watch. The rule states that the final hole diameter must remain at least half the original diameter. It appears independently in two publications by the same author, a design handbook and a NASA technical report. It also corresponds closely to a 40% overall fill factor: if one condition is met, the other will generally be met as well.

The actual area occupied by a winding is more than twice the bare copper area because of enamel, voids between round wires, and winding imperfections. This explains why a toroidal core always appears much fuller than the raw square millimeters of copper would suggest.

A toroidal transformer can achieve better efficiency than a laminated transformer of equal power for two geometric reasons, not because of any special effect: the mean turn length is shorter, so less copper is required for the same number of turns, and fewer turns are needed because the core operates at higher flux density. Core losses, however, increase because operation is closer to the knee of the magnetization curve.

Regulation is dominated by resistive voltage drop, and this is the value that determines the secondary allowance in Card 2. The practical method is iterative: enter an allowance, read the regulation, adjust the allowance, and repeat until the full-load voltage is the value you need.

As with the other two calculators, temperature rise is not calculated here because it depends strongly on mounting. A toroid resting on a metal plate cools differently from one suspended on rubber pads. One manufacturer uses 2.5 W per kilogram of core and a 50 °C rise above ambient as design limits; these are useful reference values, not a universal formula.

At switch-on, the transformer initially behaves like an inductor with an unloaded secondary. If the residual core flux has the unfavorable polarity relative to the incoming half-cycle, the two add and can drive the core well beyond the knee. Once saturated, the core provides very little inductive opposition and the current is limited mainly by the primary winding resistance.

That value is calculated because it is determinable and provides a definite physical upper bound. The actual peak depends on the exact switching instant and on residual flux, which is inherited from the previous cycle. None of the manufacturers consulted publishes a predictive formula for it, and quoted guidance ranges from fifteen to one hundred times nominal current. Such a wide range is not useful as a design equation.

For this reason, the calculated bound is shown alongside published measurements from four commercial transformers ranging from 5 to 1000 VA. They provide a realistic order of magnitude for sizing an inrush limiter, and comparison with the calculated limit shows how conservative that upper bound is.

The limiter resistance is sized from the mains peak voltage divided by the maximum current you want to allow. However, if a large capacitor bank is connected downstream, the energy the limiter must absorb may be dominated by capacitor charging rather than the transformer, and that must be checked separately.

The table uses a series of bare cores for which a manufacturer publishes diameters and power ratings. For each one, it searches for the core height required to deliver the requested power while satisfying both the fill-factor limit and the remaining-hole rule. The search uses bisection with the same calculation engine as the previous cards, so the result is consistent with the values obtained when those dimensions are entered in Card 1.

The catalog’s published VA rating is a valuable cross-check. If your calculation gives a power capability very different from the rating associated with that core size, one or more starting parameters — flux density, current density, or fill factor — are probably inappropriate.

The calculated height is the magnetic height required by the model. A commercial core has a fixed height, so the practical method is to select a suitable row, find the closest real core, and enter its actual measured dimensions in Card 1.

Cores: dimensions from a catalog of bare grain-oriented steel cores, reported in the original published units — inches — with metric conversions added alongside. The height is not published in that catalog, which is why the calculator treats it as a measured input rather than a menu item. The published cross-section also does not match the product derived from the other dimensions, so I did not use it to infer height: a missing value is preferable to an unreliable deduction.

The hole rule comes from McLyman’s design handbook and a NASA technical report by the same author, which state it independently: half of the inner diameter must remain clear for the winding shuttle, so 75% of the original hole area is available for the winding. The resulting overall fill factor is approximately 0.4.

Flux density: 1.65 T according to the technical guide of one toroidal transformer manufacturer, which describes it as “about 40% higher” than for a conventional laminated transformer; a second manufacturer specifies its core losses at 1.7 T. Both identify the absence of magnetic joints as the technical reason.

Losses: calculated from the two published data points for each grain-oriented steel grade, with the exponent derived from those points and shown in the global-parameters note. As a cross-check, one manufacturer specifies approximately 1.1 W/kg at 1.7 T for its cores.

Inrush: the four peaks in the table are published measurements from commercial transformers. The limiter criterion — peak voltage divided by maximum allowed current — comes from the same source. None of the sources consulted publishes a predictive equation for the actual inrush peak, and this page does not invent one.