How to Choose a Litz Wire Grade: Temperature, Insulation, Cross-Section Shape

Choosing a litz wire grade is three independent engineering decisions: the strand thermal class (105, 155 or 200 °C, limited by the enamel chemistry and the Arrhenius equation), the type of outer insulation (PTFE, polyimide, Arselon, PET, extruded polyethylene or polyurethane — each rated for its own environment and mechanical load) and the cross-section shape (round for standard coils or rectangular for maximum winding-window fill). It is easy to confuse these axes — and you end up with a wire that cannot be tinned, or a class 200 whose service life is not limited by temperature at all.

Choosing a grade is not about finding the «best» litz wire by general feel, but three independent engineering decisions. The strand class is the enamel chemistry and the Arrhenius equation, which decides how many years the wire will last. The outer insulation is a specific failure in a specific environment: chemistry, vibration, radiation, moisture. The cross-section shape is the geometry of the winding window and copper economics. It is easy to confuse these axes: it seems you have «merely» raised the thermal class from 155 to 200, but in fact you got a wire that cannot be tinned, and overpaid a quarter of the cost for a service life that is still limited by something else. In this article we break all three axes down into physics with formulas, tie them to the 52 grades of TU DSGT.685660.019, and turn a chaotic «help me pick» into five sequential steps from application to order line.

1. Three selection axes — and why they cannot be mixed

When a request arrives — «we need litz wire, 2.5 mm² cross-section, 40 kHz operating frequency, induction heater» — that short phrase hides six separate decisions:

1. Strand diameter and strand count — dictated by frequency and current. These get a separate detailed treatment in the articles «Skin effect and proximity effect» and in our online calculator, where everything is computed per Dowell and Sullivan.
2. Stranding construction — how many strands in the primary bundle, how many steps, which multipliers. This is the topic of the article «What is continuously transposed litz wire» with its rule of five and the categorical ban on sixes and sevens at the transposing steps.
3. Strand thermal class — 105, 155 or 200 °C. This is the enamel-wire chemistry, and no sheath will pull it out from the outside.
4. Outer insulation — from bare enamel to a combined PTFE-plus-Arselon. It is dictated by the environment.
5. Cross-section shape — round or rectangular. Dictated by the core window and copper economics.
6. Termination and mechanics — tinning, lugs, impregnation.

The first two axes are about electromagnetics; we have already covered them. Axes three through five are about materials and geometry, and it is exactly these that are encoded in the TU grade. Take a long designation from the registry — Литц 200 кФоА2 (ПЭКФоА2-200-М) 500×0,224. It decodes as follows:

• 200 — thermal class; the strand is rated for 200 °C (axis 3).
• кФоА2 — combined serving: PTFE F-4EO plus a double-layer Arselon (axis 4).
• No Пр prefix — the cross-section is round (axis 5).
• 500×0,224 — 500 strands of 0.224 mm (axes 1 and 2).

The last three axes — temperature, insulation, shape — are geometrically unrelated to one another. A decision on one does not pull the others along, and none of them backs up its neighbor. A thick polyimide sheath will not save a class F strand if the winding runs at 170 °C for days on end: the polyimide outside stays cool and happy, while the enamel in the center of the bundle burns out in a season, because heat from the middle simply does not reach the outside — it escapes not through the sheath but through neighboring turns and air.

The conclusion is simple: each axis is an independent engineering decision. Let's take them one at a time.

2. Axis 1. Temperature — what 155 °C actually means

2.1 Where the numbers on the label come from

The insulation thermal class is not the temperature at which the enamel melts. It is a point on the thermal-aging curve where the insulation life equals a standardized number of hours. Per ASTM D 2307 and IEC 60216 for enamelled wires, the norm is 20,000 hours to failure. In real terms — about 2.3 years of operation without a shutdown. The physics of this aging is described by the Arrhenius equation, behind which sits the oxidation chemistry of the polymer film:

• L(T) — insulation life at absolute temperature T, hours
• L₀ — the pre-exponential factor, depending on the enamel composition, hours
• E_a — activation energy of the thermal-oxidative degradation reaction, eV (for modern enamels — 0.9…1.3 eV per ASTM, IEC 60216 and the aerospace studies of Ji et al., 2023)
• k — Boltzmann constant, 8.617·10⁻⁵ eV/K
• T — absolute insulation temperature, K (T [K] = t [°C] + 273.15)

From this comes the famous engineering «10-degree rule». For typical values E_a ≈ 1.0 eV and an operating temperature around 150 °C, a rise of 10 °C cuts the service life exactly in half. The trick is not empirical but a direct substitution: the aging acceleration factor is computed as

Go lower down the scale and the effect gets stronger. At 125 °C (T₁ = 398 K) the same ten degrees already give a 2.3× aging acceleration. The reason is that the numerator holds the difference of reciprocal temperatures, and that changes faster at the lower part of the range.

The second standard an engineer cannot do without is IEC 60085 (in Russia it lives under the number GOST 8865-2023, «Electrical insulation systems. Thermal evaluation and classification»). It designates classes not by the failure temperature but by the temperature at which that 20,000-hour life is met:

Class (IEC 60085)	Old class	°C	What it means physically
105	A	105	Polyurethane and polyvinyl formal enamels; 20,000 h to a doubling of tan δ
130	B	130	Modified polyesters
155	F	155	Polyesterimide with a modifier — the base class for HF windings
180	H	180	Polyamide-imide (PAI); aerospace assemblies
200	—	200	PAI with a primer coat; EV traction motors
220	C	220	Composite PAI + PEEK; special aviation
240	C	240	Polyimide (ML, «Kapton»-type films and enamels)

Our TU (clause 1.2.3) uses three classes: A (105 °C) — the single grade Литц фПЭ with an extruded polyethylene sheath; F (155 °C) — the main array of 36 grades on self-solderable enamel wire; class 200 — another 14 grades on a strand with PAI enamel.

2.2 Where the difference between «155» and «200» hides

To the eye the two strands are indistinguishable. A thin copper wire in light-bronze enamel — both of them. Chemically, though, these are two different polymers, and from that difference follows a practice that often decides the fate of a project.

For us, class F is made on self-solderable enamel wire. Its main virtue is the termination method. The end of the litz wire is dipped into a solder bath at 380–400 °C, the enamel thermally decomposes and evaporates, and the fresh copper is immediately wetted by the solder. A minute — and the multi-strand bundle is neatly tinned with a common layer. For multi-section transformer windings with hundreds of short leads, this is not convenience but a question of production economics.

Class 200 is made on a polyamide-imide (PAI) enamel. It holds temperature but does not self-solder. Termination requires either mechanical stripping (and anyone who has tried to strip enamel off a 0.071 mm strand understands why this is a bad idea), or chemical removal with a solvent, or a laser. All of this is an order of magnitude slower and more expensive.

From this comes a simple rule that rarely makes it into reference books: if your product has only two ends, and the calculated winding temperature is above 135 °C — take class 200. If there are hundreds of ends but the temperature margin is sufficient — stay on class F. The service life can lose you not because the wire burns out, but because the assembly does not come together.

Alongside the strand temperature, the grade also specifies the sheath temperature. Here it is important not to confuse two fundamentally different modes:

• Continuous operating temperature is set by the strand — 155 or 200 °C. Above that the enamel ages irreversibly, and whatever was on the outside will not help.
• Short-term overheating temperature is set by the sheath — PTFE holds up to 230 °C, polyimide up to 330, Arselon up to 400. This is not an operating mode but insurance: a five-second spike at equipment start-up, an emergency transient, an accidental protection trip. The sheath must survive this peak without a turn-to-turn short while the strand cools down.

Practical takeaway. When calculating the thermal path of a winding, the continuous temperature is set by the strand, and the sheath provides a margin for peaks — no more. A coil rated for «continuous 400 °C» built from class F litz wire in a polypropylene braid is nonsense: the enamel in the strand will burn out in hours, while the braid outside stays intact and pretty.

2.3 Temperature margin: why «maxing out» is not always better

Standard industrial practice is not to buy the maximum class but to build in a 15–25 °C margin between the calculated hot spot and the class temperature. The same Arrhenius, but working in reverse: drop the operating point 20 °C below the rating, and the service life jumps from 20,000 hours to 80,000–100,000. Ten to twelve years untouched. For an EV OBC with a calculated winding temperature of +135 °C, the correct choice is exactly class F with its 155 °C and a twenty-degree margin, not class 200 «just in case». The margin pays off with ten years of trouble-free operation, not an extra quarter of the cost.

Temperature 150 °C

105 °C: — h 155 °C: — h 200 °C: — h

A family of lines in ln L vs 1/T coordinates for the three thermal classes. The horizontal line at 20,000 h is the rated norm of IEC 60216 / ASTM D 2307. Move the slider: the life values for each class recompute in real time by the Arrhenius equation with E_a = 1.0 eV. Note the rule «minus 20 °C → life ×4–5» and «plus 10 °C → life ÷2».

3. Axis 2. Insulation — six materials, six failure physics

If the class is «how many years to failure», then the outer insulation is «what exactly the wire will fail from first». Our TU describes six base materials and dozens of their combinations. Each material is optimal for its own failure mechanism.

3.1 PET: economics, mechanics, adhesion

Polyethylene-terephthalate film, 25–50 µm thick, is the workhorse among servings. Its dielectric strength is high (E_d ≈ 150 kV/mm), its mechanical strength excellent, its chemical resistance moderate. It withstands gasoline, oils, most solvents.

PET is often underrated on temperature in the reference books — let's set the numbers straight. Biaxially-oriented electrical-grade PET (Mylar, Melinex, Hostaphan and their Russian equivalents — films 6020/6021) has a glass-transition temperature T_g ≈ 80 °C, a melting temperature T_m ≈ 250 °C, and UL certification for class B (130 °C) continuous as a single insulation. Heat-stabilized grades (Mylar S, Melinex 238) qualify for class F (155 °C). In composite systems — for example, the pairing «class F enamel wire + PET film» — PET operates as part of a class F system, i.e. 155 °C continuous. And this is formalized: per the NEMA MW1000 and IEC 60085 documents, the thermal class of the final construction is determined not by an individual material but by the assembled system. In our grades Литц ПЭТ (ПЭПЭТ-155-М) this is recorded directly by the suffix 155 in the GOST designation.

Peak temperature — up to 200 °C without melting: between T_g and T_m the material plasticizes and loses mechanical strength, but keeps its insulating properties for short seconds-to-minutes. For emergency overheating at start-up or a protection trip, this is usually enough.

A separate and often underestimated virtue of PET is its good adhesion to impregnation compounds. Epoxies, polyurethane varnishes, thermally-conductive compounds of the Silagum type wet PET, flow into the seams between winding turns, and after curing hold the coil as a monolithic block. On this parameter PET is well ahead of PTFE, to which, as we will see below, compounds do not stick at all. If the process calls for the coil to be vacuum-impregnated and subsequently cured after winding (and this is the vast majority of production transformers and inductors), PET turns out to be not merely economical but directly optimal.

Its weak spot is hydrolysis. In contact with water above 80 °C or with alkalis, PET breaks down within months.

When to choose it. Windings of pulse transformers, induction drivers of moderate power, Qi charging pads, production coils for compounding, almost any dry electrical machine up to 155 °C — anything where the environment is neutral, the system is qualified for class F, and you need a wire that will «fuse» with the impregnation. Grades — Литц ПЭТ and ЛитцПр ПЭТ.

3.2 PTFE F-4EO: chemical inertness

PTFE is the champion of chemical resistance among polymer films. Indifferent to acids, alkalis, most solvents; it does not age in sunlight, does not hydrolyze. Temperature resistance — up to 200 °C continuous, up to 260 short-term. Dielectric strength is a bit more modest than PET (E_d ≈ 80 kV/mm), but for litz wire this rarely becomes the limiting factor.

PTFE has two systemic weaknesses, and both grow from the same physics — an extremely low surface energy. The first is slipperiness. Its coefficient of friction is one of the lowest in the world. For flat coils of induction heaters, where turns are held in place self-consistently by mutual friction, a bare PTFE sheath can «creep» — the turns slide apart under their own tension.

The second and more serious one is zero adhesion to compounds. Epoxies, polyurethane varnishes and thermally-conductive impregnants do not wet PTFE: a drop of compound rolls off the surface like water off a Teflon-coated pan. In the literal sense that is exactly it — Teflon (PTFE) and our F-4EO are close relatives. For coils that are meant to be vacuum-impregnated or potted, a PTFE serving creates a zone without adhesion: the compound does not hold, the protection comes out full of holes, and the mechanical fixing of the winding is broken. For induction reactors that work in an aggressive environment without impregnation, PTFE is ideal. For production transformers under potting, you need PET or a combined construction with an outer layer the compound sticks to.

When to choose it. Electrochemistry, wave soldering (flux vapors), petrochemical drives, medical autoclave sterilization with peroxide. Grades — Литц Ф, ЛитцПр Ф, Литц 200 Ф, ЛитцПр 200 Ф and all combined ones with the кФ prefix.

3.3 Polyimide: mechanics, radiation, fantastic dielectric strength

Polyimide film (in the catalog — ПИ, from the family of aromatic polyimides; the best-known trade name is the legendary DuPont Kapton) is the strongest thin-film electrical insulation material of all. The numbers look fantastic and deserve a separate discussion. Per the Kapton HN datasheet (DuPont, test ASTM D-149, 60 Hz):

Thickness	Dielectric strength Ed
25 µm	303 kV/mm
50 µm	240 kV/mm
75 µm	205 kV/mm
125 µm	154 kV/mm

For the 25-micron films typical in winding production — more than three hundred kilovolts per millimeter. For comparison: PTFE — 80, standard PET — 150, transformer oil — 20. A 25 µm polyimide film withstands breakdown at 7.6 kilovolts AC — at a thickness less than a hair.

The temperature range is, without exaggeration, unique among polymers: from −269 °C to +400 °C. DuPont states these bounds directly: from cryogenic (liquid helium, scientific instrumentation, MRI magnets) to continuous operation in industrial furnaces. Radiation resistance — the best in the polymer class; Kapton survives in open space for decades as the thermal blanket of spacecraft and solar sails.

Specifically for modern SiC and GaN inverters, DuPont makes Kapton CRC (Corona Resistant) — a film with radically increased resistance to partial discharge. The physics is nasty: a steep switching front shorter than a hundred nanoseconds creates a traveling voltage wave along the winding, the end turns are limited in inductance, and the local voltage between adjacent turns jumps two to three times above the average. In micron-scale air gaps, PDIV (Partial Discharge Inception Voltage) ignites — a partial discharge. It carries almost no current itself, but over thousands of hours it neatly carbonizes the insulation, turning it into a carbon track. Ordinary Kapton HN degrades in such an environment within years; CRC holds the same regime for decades. It is precisely on the 150FCRC019 film that DuPont makes magnet wires for the traction motors of railway locomotives and the turbo-generators of wind power. And separately there is the Kapton MT+ grade with a thermal conductivity λ = 0.8 W/(m·K) — four times higher than standard polyimide, for applications where you need to insulate and remove heat at the same time.

Ji et al. in IEEE Trans. Aerosp. Electron. Syst. (2023) showed that replacing the standard polyurethane varnish with a polyimide+BN-nanosheet composite raises the PDIV of an inverter winding from 500 to 900 V at the same insulation thickness. For an 800 V EV bus per IEC 60034-18-41, this is the difference between «degrades within 20,000 hours» and «runs its rated life with no issues». Hence the rapid migration of aviation and EVs to polyimide.

The Russian equivalents are the PM films (PM-1, PM-1E) per GOST 25924, close in characteristics to Kapton HN. In our catalog the polyimide serving comes in the grades Литц ПИ, ЛитцПр ПИ, Литц 200 ПИ, ЛитцПр 200 ПИ.

Polyimide, of course, has its weak spot too — the combination of temperature and humidity. In a dry atmosphere it lasts practically forever; in saturated steam at 120 °C continuous, it degrades within thousands of hours via hydrolysis. For induction coils that work outdoors in rain and snow, it is less suitable than PTFE. Where there is both heat and humidity — either a combined sheath with PTFE, or PTFE straight away.

When to choose it. Aviation and space motors, SiC/GaN inverters at 600–800 V, MRI coils, high-Q circuits, cryogenic windings (liquid helium, low-temperature magnets), anything that requires temperature resistance, radiation resistance and mechanical strength simultaneously.

3.4 Arselon: emergency temperature resistance and retro aesthetics

Arselon is a Russian meta-aramid, a sibling of DuPont's American Nomex. Unlike the film materials, it is a fiber: it does not wrap the bundle with a flat tape but is braided or woven onto it. An Arselon serving, like a polypropylene one, is done with multifilament yarn — so it inherits the same important property: it readily takes up impregnating varnish or compound during vacuum impregnation. The epoxy poured into a coil fills the space between fibers and binds the winding to the strand mechanically and thermally. This is non-trivial, because getting both 400 °C temperature resistance and impregnability at once is a rare combination, and it is what keeps Arselon on the market.

Continuous temperature around 250 °C, short-term — 400. Dielectric strength is low (E_d ≈ 20–30 kV/mm — the air inclusions between fibers are to blame), but for a litz-wire serving this rarely becomes a problem.

Arselon's main bonus is its behavior in a fault. If a turn-to-turn short occurs somewhere, the temperature at the breakdown point spikes by hundreds of degrees within seconds. PET in such a situation melts and flows, degrading the insulation of its neighbors and triggering a cascade breakdown. Arselon does not melt. It keeps its shape and insulating properties up to 400 °C, giving the protection time to shut the equipment down. For industrial equipment, where downtime from a fault is counted in millions of rubles, this is a fully valid engineering reason to pay more.

The second virtue is purely aesthetic. Braided golden Arselon looks like the winding on bobbin coils from the early twentieth century. For audiophile builds and designer induction cooktops this is genuinely in demand — and let's not pretend engineers are indifferent to it.

When to choose it. Industrial furnaces, drying chambers, aircraft cabins (AS/FAR fire-resistance requirements), retro audio, decorative coils. Grades — Литц оА1, Литц оА2, Литц хА, Литц 200 оА1/оА2/хА and all combined ones with Arselon.

3.5 Polypropylene: cheap protection and excellent impregnability

Polypropylene fiber is a simple mechanical protection for the bundle. Continuous temperature 105 °C (not PP itself but its tendency to soften), moderate chemical resistance, medium mechanical strength. Where the only thing required of the serving is to protect the strand from abrasion against the bobbin or against a neighboring turn, PP covers the task.

In the catalog, PP comes in two forms: a serving (оПП1, оПП2 — one or two layers) and a braid (хПП). The braid is denser, the serving is cheaper.

An important property of the polypropylene serving rarely mentioned in catalogs: it is made of multifilament yarn — many thin fibers twisted into a loose cord. After being laid down, this cord on the bundle becomes a capillary system: during vacuum impregnation the varnish or compound is drawn in between the fibers, fills all the voids, and after curing turns the serving into a monolithic frame tightly bound to the strand. As a result, a coil wound with Литц оПП2 and impregnated with epoxy is no worse in mechanical strength than one compounded on PET, and on thermal contact it even wins — the compound fills the space between fibers far more completely than it lies on a flat film.

When to choose it. Consumer appliances, low-power SMPS, prototypes, pilot samples, low-power wireless chargers, production coils for vacuum impregnation on a limited budget. Grades — Литц оПП1/оПП2, ЛитцПр оПП1/оПП2, Литц хПП, ЛитцПр хПП.

3.6 Extruded polyethylene and polyurethane: solid moisture resistance

The grade Литц фПЭ stands apart. It is the only construction in the catalog with an extruded polyethylene sheath over the PTFE film. The result is a tube — continuous, seamless, without regular gaps. For underwater equipment, outdoor antennas, inductors freely suspended in air, this is essential: a serving has seams, a braid has gaps between yarns, and water will find any crevice. Only extrusion gives a hermetic seal.

The temperature is limited to 105 °C — that is what polyethylene is as a material. But for applications where moisture resistance is decisive, this boundary usually does not get in the way.

An important nuance here. The extruded sheath and the polyurethane extrusion are the only solid sheaths in the catalog. Everything else — PET, PTFE, polyimide, PP, Arselon in any form — is discrete films and yarns with process seams. For moisture resistance and for galvanic isolation the difference is fundamental.

3.7 Summary

Material	T cont., °C	T peak, °C	Ed, kV/mm	Chemistry	Impregnation	Price	Leading application
PET (film)	155 (in system)	200	150	Medium	Excellent	Low	Production coils, class F system
PTFE F-4EO	200	230–260	80	Maximum	None	High	Chemical aggression without impregnation
Polyimide (25 µm)	240 (400 Kapton)	400	303	High	Medium	Very high	Radiation, cryogenics, high dV/dt
Arselon (serving)	250	400	20–30	Good	Excellent	High	Temperature resistance + impregnation
Polypropylene (serving)	105	130	50	Medium	Excellent	Minimal	Cheap + impregnation
Extruded PE	90	105	20	Medium	—	Medium	Moisture resistance

PETProduction coils PTFEChemistry PolyimideSiC/GaN, space ArselonEmergency margin PolypropyleneBudget + impregnation Extruded PEMoisture resistance

Six axes: continuous and peak temperature (logarithmic scale), dielectric strength, chemical resistance, mechanical strength, economics (farther from center = cheaper). Click the legend chips — the selected material is highlighted, the rest are dimmed. Plainly visible: no single material covers all axes, hence the combined grades (кФоА, кФоПП).

4. Axis 3. Cross-section shape — round vs rectangular

4.1 Window-fill geometry

The core window is a rectangle. Round litz wire does not pack into a rectangle in the best way: air gaps remain in the corners, between turns, along the edges of the flanges. The theoretical fill limit for round elements in a regular close packing is:

This is the ideal case, a hexagonal packing of identical cylinders. In practice, accounting for the bobbin, the interlayer insulation and the inevitable bowing — the bulging of the middle turns under winding tension (the classic description is chapter 4 in McLyman's «Transformer and Inductor Design Handbook») — the real fill of a rectangular window with round litz wire comes out to 0.50–0.65. Roughly speaking, half the window is air.

A rectangular cross-section changes the arithmetic. In theory it fills the window 100%, in practice with tolerances — to 0.75–0.85. The difference from the round equivalent is 15–25 percentage points. That is where the «+20% window fill» figure on our home page comes from. It translates into engineering economics in two ways: either fix the window and add 20% copper to the cross-section (and reduce the winding's DC resistance by the same amount), or fix the winding and shrink the window — getting a smaller, lighter, cheaper core.

4.2 How we do it: one pass, together with the insulation

Our technology is not the one described in most litz-wire publications. The classic description (the same Sullivan APEC 2014, the Elektrisola and Rubadue catalogs) depicts a multi-stage process: first the bare copper bundle is drawn through several roller stands with intermediate forms, then the outer insulation is applied to the already-formed rectangle. We do it differently.

The compaction is done in a single pass, and — this is important — together with the already-applied insulation. The assembled litz wire with all sheath layers — be it a film, a multifilament-yarn serving, or their combination — enters the profiling tooling round and exits rectangular with the specified cross-section. The copper bundle and the sheath deform simultaneously, as a single body.

Three practical consequences follow from this approach.

Tight fit of the sheath to the strand. When the film or serving is formed together with the bundle, it lies onto the new cross-section without bubbles or gaps. If the sheath were applied to an already-rectangular bundle, air pockets would inevitably form in the corners: the film would not hug the sharp corner as tightly as it does a flat side. In our process the sheath is «drawn into» the corners during forming and fits the strand across its whole surface. For thermal conductivity, breakdown strength and the hermeticity of the finished product, this matters more than it might seem.

Short production cycle. One operation instead of two. In series production this is both time and fewer points where something can go wrong. For pilot samples and small batches — the same: faster launch, a lower threshold for experiments.

Demands on the insulation's elasticity. Not every material survives single-pass forming together with the strand — the sheath must endure substantial deformation without tears or microcracks. Here the materials in our catalog are chosen precisely for such a process. PET in a thin gauge (25–50 µm) is a biaxially-oriented, stretchable film; it passes without trouble. Polyimide is one of the mechanically strongest polymers (MIT fold endurance up to 285,000 cycles), and deformation is a normal mode for it. PTFE is plastic by nature and does not crack. Multifilament servings (PP and Arselon) — here it is even better: the individual yarns redistribute over the new shape without breaking, because there are many of them and they lie loosely to begin with. The limit is strongly elongated rectangles with an aspect ratio above 3.5:1, where the deformation becomes technologically incorrect (more on this in section 4.5).

The price of the technology is a small increment in strand length inside the bundle. Per our TU (clause 1.3.8), the standard coefficient K_Σ for round wire is 1.03, for rectangular — 1.05. These two percentage points also account for the forming effect: under compaction the strands inside the bundle stretch slightly, the helical lay path changes, and more material is used per meter of finished product. The result is +2% copper and +2% DC resistance per meter. Compared with the 15–25% gain in winding-window fill, this is a disproportionately small price.

Round Rectangular

Window fill: 60 %

Toggle the shape — see how the same volume of copper packs into the same window. With round strands ~40% of the window is air in the corners and between turns; a rectangular cross-section fills the window to 75–85%. The practical gain: η_rectangle/η_round ≈ 4/π ≈ 1.273 ideally, in reality +20–25%.

4.3 Why a catalog of rectangular cross-sections does not exist

Round litz wire has a standard range of cross-sections — five series in our catalog, 199 ready-made constructions. For rectangular wire such a catalog cannot exist, and this is not an oversight but a physical feature of the task.

Rectangular litz wire exists precisely to fit as densely as possible into a specific window of a specific core. And the windows of cores are all different. There do not exist two transformers whose winding window matches to the millimeter: one has an EE-core, another an EI, a third an ETD, a fourth a custom one for its own board. And even within a single size, a customer may have bobbin flanges of different thickness, their own interlayer insulation, their own requirements for clearance to the core. Each task has its own geometry, and the optimal cross-section of rectangular litz wire must match exactly that one, not a neighboring one.

If we kept a catalog of twenty or thirty «standard» rectangular cross-sections, the customer would have to choose one of two things: either fit their winding geometry to our catalog (and lose part of the fill advantage for which they take the rectangle in the first place), or order a custom item. In both cases the catalog turns out to be not a helper but a middleman between the customer and their real task.

That is why for us every rectangle is made to order, with a cross-section calculated precisely for the customer's window. And this works not thanks to some abstract «production flexibility» but thanks to one specific engineering decision: when we designed our profiling machine, we deliberately made the forming tool simple to manufacture and replace. Tooling for a new rectangular cross-section is, for us, not an expensive separate operation but a quick interchangeable part. A custom rectangle geometry therefore costs no more than a standard one, and we see no point in artificially narrowing the customer's choice with a catalog of «popular sizes».

For the customer this means a simple thing. If you are designing a winding for a specific core — send us the window dimensions, the flange thickness, the interlayer-clearance requirements. We will select the optimal cross-section of rectangular litz wire for maximum fill of your geometry, with no «do you have a 2.8 × 1.4 mm profile?» queries and no compromise toward the nearest catalog size.

4.4 When to choose round, when rectangular

Round litz wire (no Пр prefix in the grade) is your choice when:

• The core is toroidal. A toroid has no rectangular window — the winding cross-section shape gives no advantage.
• You need maximum flexibility. A round wire bends in all directions equally; a rectangular one only in its plane.
• The batch is small and the fill requirements are routine. The profiling operation is a separate process step, and for single meters it rarely pays off.

Rectangular litz wire (the Пр prefix) — when:

• The window of an EE-, U- or EI-core and the winding density is critical. The typical case — SMPS transformers, PFC and LLC inductors, EV OBC inductors.
• The winding is concentrated in an electrical machine. Today in traction motors hybrid assemblies — a hairpin winding in the slot plus rectangular litz wire in the end turns — are starting to work as a standard (MDPI, Appl. Sci., 2023).
• Height clearance is limited. The flat coil of a Qi charging pad is the canonical case of an elongated rectangular litz wire.

4.5 The rectangle's aspect ratio is a separate engineering task

Once the cross-section shape is chosen, the question remains — what proportion? In the part number the profile is given by a pair of dimensions after the strand diameter: 2×2 — an isotropic cross-section with a 1:1 aspect ratio (colloquially, a «square»), 3×1,5 — a moderately elongated 2:1 rectangle, 3,5×1 — already close to the practical forming limit. The choice of ratio is not cosmetic: it changes three properties at once.

Mechanical anisotropy. An elongated cross-section is sharply asymmetric in bending stiffness. A flat 3:1 rectangle bends around the short axis (in the «plane») almost as easily as a thin bundle of round strands, but around the long axis — like a wide plate, i.e. stiffly. For the flat coils of a Qi charging pad this is convenient: the winding holds its shape in the plane of the substrate and does not require a rigid bobbin. For a coil that must be wound with a fold (toroidal geometry, radial leads) — the opposite, the closer the ratio is to 1:1, the easier it is to lay.

Thermal anisotropy. In a strongly elongated cross-section the heat path from the center to the surface along the short axis is several times shorter than along the long axis. For a winding with active heat generation this means that the thermal time constant in the «plus-minus thickness» plane is sharply lower than in the «plus-minus width» plane. For flat coils on a ferrite base with a heatsink this works in your favor: heat is shed efficiently into the core. For bulk concentrated windings of electric motors, where cooling comes uniformly from all sides, the isotropic square profile is more efficient.

Electromagnetic anisotropy. The proximity effect in a rectangular cross-section depends on the orientation relative to the local magnetic field of the winding. A field perpendicular to the long side induces eddy currents over a larger area than a field perpendicular to the short side (Ferreira, IEEE 1994; Lotfi & Lee IAS 1993). The difference in F_R can reach tens of percent at the same copper cross-section. For correct design this means that on planar coils the elongated cross-section should be oriented with its long side parallel to the winding's magnetic field lines.

Putting it all together, three characteristic aspect-ratio ranges:

Aspect ratio	Typical application	Why exactly this
1:1 (square)	Concentrated windings of electric motors with rounded slots, EE/EI-cores with a near-square window	Isotropic stiffness and heat removal; turns lay symmetrically; FR independent of orientation
2:1…3:2	Typical windows of SMPS and LLC EE-cores, PFC inductor windings	A compromise between window fill and winding flexibility
3:1…3.5:1	Flat Qi wireless-charging coils, planar transformers, pancake coils of induction heaters	Low winding thickness in one plane is critical; closer to the practical limit of the forming technology

Above 3.5:1 is the practical limit of our forming. Beyond this boundary, strain concentrators appear on the long sides, the insulation gets a non-uniform load, and the copper-cross-section fill becomes uneven. We do not make such rectangles, not on principle but out of engineering honesty: the result stops being predictable, and the quality stable. If a task truly requires a flat ribbon (10:1 and above) — that is no longer litz wire but rather a flat cable with a different manufacturing technology, and it needs separate tooling.

5. How the axes connect — PDIV, thermal path, price

Geometrically the three axes are independent. But there are three cross-cutting effects that link them in practice and force the designer to make decisions with an eye on all of them at once.

5.1 PDIV and the traveling wave in an inverter winding

For windings fed by modern SiC and GaN inverters, the critical parameter is PDIV, the partial discharge inception voltage. We covered the physics in detail in the polyimide section: steep switching fronts shorter than a hundred nanoseconds create a traveling voltage wave along the winding, local peaks between turns reach three times the rating, and in micron-scale gaps partial discharges ignite, turning the insulation into a carbon track over thousands of hours.

PDIV grows with the thickness and dielectric strength of the outer insulation. The strand thermal class hardly works here — the thin enamel on the strands does not hold hundreds of volts. For an 800-volt bus per IEC 60034-18-41 the engineering choice is a polyimide film on the outside: its 303 kV/mm at a 25 µm thickness gives a PDIV margin of 2–2.5 kV, and the winding gets a chance to survive its rated life without a discharge season inside every microgap. If the application is especially harsh — the traction motor of an electric locomotive, a 1.5 MW wind generator — you already need Kapton CRC or a Russian equivalent with increased partial-discharge resistance.

5.2 The thermal path — why the strand class matters more than the sheath class

The thermal conductivity of winding materials, per Kyaw, Sullivan, Schanen et al. (IEEE Trans. Power Electron., 2020):

• copper: λ = 400 W/(m·K);
• air: λ = 0.026 W/(m·K);
• polyurethane varnish: λ = 0.2 W/(m·K);
• polyamide-imide (PAI) varnish: λ = 0.27 W/(m·K);
• PTFE: λ = 0.25 W/(m·K);
• polyimide (ordinary Kapton HN): λ = 0.12 W/(m·K); the special Kapton MT+ version — up to 0.8 (six times higher than standard);
• Arselon and Nomex servings: λ = 0.07–0.1 W/(m·K) (much air between fibers).

The numbers suggest a conclusion. Between the center of the bundle and the surface of the winding there is no straight thermally-conductive copper but a layer cake of varnish and air. The outer sheath is only the last, thin layer. The effective transverse thermal conductivity of an assembled litz-wire winding (Liu, Gerada et al., IEEE Trans. Ind. Electron., 2020) is 0.3–0.6 W/(m·K). Ten to twenty times more than air, a thousand times less than copper.

The consequence — the inner turns of the coil are 30–60 °C hotter than the outer ones. This is not a theoretical correction; these are direct measurements on NASA's HEMM motorette (2019): at a calculated outer-layer winding temperature of 120 °C, the strand at the center operated at 160–180. Hence the discipline: the strand class must cover the inner hot spot, not the coil average and not the outer surface. Class F (155 °C) is appropriate in coils where the calculated temperature of the outer turns is no higher than 100–110 °C. Anything that runs hotter — class 200.

There is also a subtlety about the sheaths themselves. A solid extruded sheath insulates heat better than a breathing serving: extruded PE is worse for heat removal than the same PTFE film over a free bundle. For induction heaters and PFC inductors with active heat generation, breathing constructions are preferable. The extruded sheath goes where moisture resistance is more important than heat removal.

5.3 The economics of the three axes

Approximate price ratios at equal length and equal strand. We take the polypropylene serving as unity:

• PP — ×1.0
• PET — ×1.1
• Arselon (serving) — ×1.3
• PTFE F-4EO — ×1.6
• Polyimide — ×2.0
• Combined ones like кФоА2 — ×2.2

Going from class F to 200 (strand) — plus 15–25% to the price. The profiling operation (round → rectangular) — plus 10–20%. If you move «up» on all three axes, from Литц ПЭТ to ЛитцПр 200 кФоА2, the total cost at the same copper multiplies 3.5–4 times. Each axis is an independent contribution to the price. Do not overpay where the margin in a property is not repaid by service life or reliability.

6. Algorithm — from application to grade in five steps

We assemble the three axes into a working sequence. Each step is one decision.

Step 1. Series — by frequency

Frequency sets the maximum diameter of the elementary strand: d ≤ δ for single- and double-layer windings, d ≤ δ/2 — for multilayer ones (more in the calculator documentation, the Sullivan optimum section). From this comes the catalog series:

• 80–150 kHz → the 0.071 mm series (at 100 kHz δ = 0.21 mm). 41 constructions.
• 30–80 kHz → the 0.1 mm series. 40 constructions.
• 10–30 kHz → the 0.15 mm series. 40 constructions.
• 3–15 kHz → the 0.2 mm series. 40 constructions.
• up to 10 kHz → the 0.25 mm series. 40 constructions.

Step 2. Strand count and layout

We compute the copper cross-section as S = I_RMS / J, where J is the allowable current density: 3–5 A/mm² with natural convection, 6–8 with forced air, up to 12 in oil. The strand count N = S / (πd²/4) — and we round to the nearest standard construction from the series table. The layout by stranding steps (the first step — up to 60 strands in a true bunch, then multipliers only from threes, fours, fives) we covered in detail in the article on transposition.

Step 3. Thermal class — by the hot spot

The decision is made not by the average winding temperature but by the calculated temperature at the center of the bundle, accounting for the internal 30–60 °C above the outer surface:

• T_center ≤ 100 °C — you can use Литц фПЭ (class 105, if moisture resistance is critical) or standard class F without a prefix.
• 100 < T_center ≤ 135 °C — class F. The main body of the catalog.
• 135 < T_center ≤ 180 °C — class 200 (grades with the 200 prefix in the name).
• T_center > 180 °C — class 200 plus a high-temperature sheath (ПИ or Arselon).

Step 4. Outer insulation — by the leading environmental factor

Leading factor	Insulation	Typical grades
Nothing special, price matters	No serving or PP	Литц, Литц оПП1
Mechanical abrasion	PET	Литц ПЭТ
Chemical aggression	F-4EO	Литц Ф, Литц 200 Ф
Radiation, high dV/dt (SiC/GaN)	Polyimide	Литц ПИ, Литц 200 ПИ
High temperature + emergency margin	Arselon	Литц оА2, Литц 200 хА
Chemistry + temperature at once	Combined кФо*/кФх*	Литц кФоА2, Литц 200 кФоА2
Solid moisture resistance	Extruded PE or polyurethane	Литц фПЭ
Decorative application	Arselon braid	Литц хА, Литц 200 хА

Step 5. Cross-section shape

• Round cross-section — for toroidal cores, flexible bundles, small batches, prototypes, applications with modest window-fill requirements. Size — from the standard catalog range.
• Rectangular (the Пр prefix) — for EE/EI-cores with critical fill, flat coils, concentrated motor windings. The cross-section is not selected from a list but calculated for the customer's specific window: for your core we pick rectangle sides at which the turns lay as densely as possible. Price — +10–20% over the round equivalent, the gain — 15–25% in window fill.

• Application Frequency, current, environment, size
  80–150 kHz → series 0.071

  T_center ≤ 135 °C → class F

  • Standard: Литц ПЭТ 125×0,071

  • SiC/GaN + high dV/dt: Литц ПИ 125×0,071

  • Rectangle for an EI-core: ЛитцПр ПЭТ 125×0,071 [b×h]

  T_center > 135 °C → class 200

  • Standard: Литц 200 Ф 125×0,071

  • Emergency margin to 400 °C: Литц 200 кФоА2 125×0,071
  30–80 kHz → series 0.1

  • Class F, dry environment: Литц ПЭТ 315×0,1

  • Class F, vacuum impregnation: Литц оПП2 315×0,1

  • Class 200, petrochemical: Литц 200 Ф 315×0,1
  10–30 kHz → series 0.15

  • Induction heater: Литц ПЭТ 170×0,15

  • Moisture-resistant, in the rain: Литц фПЭ 170×0,15
  3–15 kHz → series 0.2

  • Power welding inverters: Литц 200 Ф 144×0,2
  up to 10 kHz → series 0.25

  • EV chargers, large currents: Литц 200 кФоА2 255×0,25

Click the nodes — a branch will expand. The tree reflects the logic of the algorithm: first frequency → series, then hot spot → thermal class, then environment → sheath, finally shape → the Пр prefix or not. The leaves are examples of ready order lines per TU DSGT.685660.019.

Full example: 15 kW induction heater, 30 kHz

1. Series. At 30 kHz the skin depth δ = 0.38 mm. We choose the 0.15 mm series (d/δ ≈ 0.4, strictly less than half of δ — a good margin per Sullivan). Series 0.15.
2. Layout. RMS current 15 A, density 5 A/mm² with convection → cross-section 3.0 mm². The nearest standard construction is 170×0,15 (cross-section 3.0 mm², 170 strands, layout 34 in the primary bundle + five bundles in the second step).
3. Class. Calculated winding temperature 110 °C (outer layer), center — 140–150. Class F (155 °C) — with a 5–10 °C margin. Sufficient, though without much comfort.
4. Insulation. The environment is industrial, there is no water cooling, mechanical loads are moderate. For a balance of price and reliability — PET.
5. Shape. An EI-core with a 12×40 mm rectangular window; for this geometry the optimal litz-wire cross-section is 1.8×1.8 mm (a dense two-row layout along the window height with a margin for the bobbin flanges). For this cross-section we make the corresponding forming tool to order.

The resulting order line: ЛитцПр ПЭТ (ПЭПЭТ-Пр-155-М) 170×0,15 1,8×1,8 ДСГТ.685660.019ТУ.

7. When the catalog is not enough

Clause 1.2.2 of our TU explicitly allows the customer to configure their own construction, basing it on any of the standard ones. This is not a formality. It is working practice. Here are the scenarios where the ready catalog does not cover the task and we make non-standard constructions.

Non-standard strand diameter. Five series cover the range from 0.071 to 0.25 mm. But at 400 kHz, by the Sullivan optimum, 0.05 or 0.063 mm may be needed. For very low frequencies with large currents — 0.315 or 0.4. All of this we make as custom items; the corresponding catalog section already counts 78 combinations, and it grows regularly with projects.

Non-standard sheath. The catalog has, for example, no «polyimide plus Arselon» combination — yet physically we assemble it without trouble. For special applications that need both the radiation resistance of polyimide and the emergency thermal margin of Arselon at once, we make custom constructions.

Experimental constructions. For a research task — a dissertation, a prototype of a new converter, an HF generator at a non-standard frequency — we are ready to make a small batch, from tens of meters. In some cases, when a new construction extends our own experience (for example, a very thin 0.04 mm strand with hundreds of strands in the bundle for gigahertz applications), we make a trial run at our own expense. In return we ask for the chance to measure the finished product and add the results to our database — a fair exchange.

Any rectangular cross-section. As discussed in detail in section 4.3, a catalog of «standard» rectangular cross-sections does not and cannot exist for us — each task requires its own geometry. Send the core-window dimensions and the winding requirements — the tooling for your cross-section is inexpensive for us, and manufacturing requires no separate complex changeover.

Optimization for your geometry. If you have a specific core with a specific window, air gap and loss requirements — the optimal construction often falls between the standard catalog items. A first approximation is given by the calculator. Then we refine the construction for your geometry and produce exactly that.

How it works in practice: you send a specification or oscillograms → we run the calculation and propose one or two recommended constructions with a justification → we agree on the specification → we put it into production. Lead time for standard grades — from seven working days; for custom ones — three to six weeks depending on complexity.

Check your choice in the calculator

The three axes of this article are assembled in the online litz-wire winding calculator. Enter the frequency, current, voltage, window dimensions, number of turns — and get the ideal calculated construction and the nearest standard catalog item, with a ready part number per TU. For non-standard tasks — describe the spec in a request.

References

1. TU 27.32.11-019-59596873-2025 (ДСГТ.685660.019ТУ). High-frequency transposed winding wires. Technical specification. DIADA LLC, 2025.
2. GOST 8865-2023. Electrical insulation systems. Thermal evaluation and classification (harmonized with IEC 60085:2007).
3. IEC 60216-1:2013. Electrical insulating materials — Thermal endurance properties.
4. ASTM D 2307. Standard Test Method for Thermal Endurance of Film-Insulated Round Magnet Wire.
5. Sullivan C.R. Optimal Choice for Number of Strands in a Litz-Wire Transformer Winding. IEEE Trans. Power Electron., vol. 14, no. 2, pp. 283–291, March 1999.
6. Sullivan C.R., Zhang R.Y. Simplified Design Method for Litz Wire. IEEE APEC, Fort Worth, 2014, pp. 2667–2674.
7. Kyaw P.A., Delhommais M., Qiu J., Sullivan C.R., Schanen J.-L., Rigaud C. Thermal Modeling of Inductor and Transformer Windings Including Litz Wire. IEEE Trans. Power Electron., vol. 35, no. 1, pp. 1–14, January 2020.
8. Liu X., Gerada D., Xu Z., Corfield M., Gerada C., Yu H. Effective Thermal Conductivity Calculation and Measurement of Litz Wire. IEEE Trans. Ind. Electron., vol. 67, no. 4, pp. 2667–2677, April 2020.
9. Delhommais M. et al. Practical Guidelines for the Litz Wire Selection and AC Copper Losses Estimation. Bodo's Power Systems, 2022 (PREMO R&D).
10. Ji Y., Giangrande P., Zhao W., Madonna V., Zhang H., Galea M. Inverter-Fed Electrical Machines Exposed to Prevailing Electrical Stress. IEEE Trans. Aerosp. Electron. Syst., vol. 59, no. 9, pp. 9368–9378, 2023.
11. NASA/TM—2019-220220. Thermal Analysis of Potted Litz Wire for High-Power-Density Aerospace Electric Machines (HEMM), 2019.
12. Dowell P.L. Effects of eddy currents in transformer windings. Proc. IEE, vol. 113, no. 8, pp. 1387–1394, August 1966.
13. Mc Lyman C.W.T. Transformer and Inductor Design Handbook, 4th ed., CRC Press, 2011, ch. 4.
14. Elektrisola. HF-Litz Wire Technical Information. elektrisola.com.
15. Lopez G. High-Performance Polymers for Aeronautic Wires Insulation. Recent Progress in Materials, vol. 3, no. 1, art. 005, 2021 (SAFRAN Tech).
16. Ferreira J.A. Improved analytical modeling of conductive losses in magnetic components. IEEE Trans. Power Electron., vol. 9, no. 1, pp. 127–131, 1994.
17. Lotfi A.W., Lee F.C. A high frequency model for litz wire for switch-mode magnetics. IEEE IAS Annu. Meeting, vol. 2, Oct. 1993, pp. 1169–1175.