EN

Modifying an induction motor winding

How changes of voltage, frequency, poles and winding type affect the turns and the wire section

Modifying a winding is easier than calculating one from scratch: if the motor ran well, its turns and wire already tell you what that iron and those slots can take. Proportions from the original are enough, which is why the first step is to record its data properly before stripping it. If you do not have the data, start from Turns by stator.

The two rules of any modification

Rule 1: the iron must keep working the same way

The voltage of each phase is balanced almost entirely by the back EMF. This is the formula that sets the flux:

\[ E = 4.44\, f\, N\, \Phi\, k_w \]

If the flux rises, the iron saturates and the no-load current shoots up; if it falls, torque is lost. To keep the same air-gap flux density, the turns have to change like this:

\[ \frac{N_2}{N_1} = \frac{V_2}{V_1} \cdot \frac{f_1}{f_2} \cdot \frac{2p_2}{2p_1} \cdot \frac{k_{w1}}{k_{w2}} \]

Rule 2: the copper in the slot does not change

The slot is the same, so about the same copper fits. If c is the number of conductors per slot and S the section of each one (the sum of the wires in hand), their product stays constant:

\[ S_2 = S_1\, \frac{c_1}{c_2} \]

At the same current density, the phase current changes just as the section does. More turns with less current, or the other way round: the ampere-turns do not change, so neither does the torque, as long as the flux density is the same.

Before rewinding: is it necessary?

Sometimes connecting it differently is enough:

Summary of effects

k is the factor of the change. The table keeps the flux density, the copper in the slot and the current density:

ChangeTurnsSectionCurrentTorqueSpeedPower
Voltage × k, same frequency× k÷ k÷ k===
Frequency × k, same voltage, rewinding÷ k× k× k=× k× k
Frequency and voltage × k, same winding====× k× k
Poles × k, with more poles× k÷ k÷ k≈ =÷ k÷ k or somewhat less
Winding factor from kw1 to kw2× kw1/kw2÷ kw1/kw2====
Parallel paths × k× k, per coil÷ k, per wire= line===
From star to delta, same line voltage× 1.73÷ 1.73= line===

These are starting proportions and do not replace the final checks. Iron losses, ventilation and mechanical limits do not follow these rules so neatly.

The motor used in the examples

We use the motor from the turns calculation guide: 4 poles, 400 V in star, 50 Hz, about 5.5 kW and 11 A. It has 36 slots in a single layer, 30 turns per coil (180 per phase) and 3 wires of 0.95 mm in hand, 2.13 mm² in total. Each slot holds 64 mm² of copper. For the slot we take the case from Turns by slot: rounded, h = 20 mm, W = 8 mm and 0.3 mm paper.

Changing the voltage

At the same frequency, the turns follow the phase voltage and the section does the opposite. Power does not change. What counts is the voltage across the coil, not the nameplate figure: always compare phase voltages.

Example. Changing the motor from 400 V in star to 400 V in delta so that it can be started in star-delta.

  1. Phase voltage: from 231 V to 400 V, that is, × 1.73.
  2. Turns per coil: 30 × 1.73 = 52. Per phase, 312.
  3. Section: 2.13 × 30 / 52 = 1.23 mm². A single 1.25 mm wire gives 1.227 mm².
  4. Phase current: 11 / 1.73 = 6.35 A, about 5.2 A/mm², the same as before. The line current stays at 11 A.
  5. Slot: 71 conductors of 1.25 mm fit and we need 52. Turns by stator confirms the result: its 1082 at 9500 G and 400 V becomes about 1867 with 690 V in star, and 1867 / 36 ≈ 52.

Changing the frequency

Frequency changes both flux and speed, so there are several routes. For a 400 V, 50 Hz motor moving to 60 Hz:

What is doneFluxSpeedTorquePowerKeep in mind
Nothing: 400 V at 60 Hz83 %+20 %Rated 83 %; starting and maximum, close to 70 %Almost the sameIt may not start demanding loads
Raise the voltage: 480 V at 60 Hz100 %+20 %Same+20 %More iron losses and more mechanical stress
Rewind: 400 V at 60 Hz, N × 5/6, S × 1.2100 %+20 %Same+20 %Current rises 20 %: check protections and cables

In all three cases the driven machine turns 20 % faster, and a pump or fan then demands 73 % more power. The motor’s own fan also uses more, although it cools better.

The opposite route is the dangerous one. A 60 Hz motor at 50 Hz with the same voltage receives 20 % more flux: it saturates, the no-load current shoots up and it can burn out with hardly any load. Lower the voltage in the same proportion, from 460 V to about 380 V, or rewind with N × 1.2 and S ÷ 1.2. Either way, power falls to 5/6 and the motor cools worse.

Example. For 400 V and 60 Hz with the same flux, the turns go from 30 to 25 and the section needed is 2.13 × 30 / 25 = 2.55 mm². With 1.05 + 1.05 + 1.00 mm you get 2.52 mm², 5.2 A/mm² with the new 13.2 A. 34 groups fit and we need 25. It gives about 6.6 kW at about 1800 rpm. The result matches Turns by stator at 60 Hz.

With a frequency inverter no rewinding is needed: it keeps V/f up to the base frequency, and above it the flux falls because the voltage cannot rise any more. If you rewind a motor that will run on an inverter, use wire suited to those voltage spikes.

Changing the poles

With more poles, each one covers less of the air gap, so the flux per pole is inversely proportional to the poles. If the flux density is kept, the turns are directly proportional. But that flux passes through the yoke, and that is where the limit lies:

\[ \Phi = \frac{B_m\, \pi\, D\, L}{2p} \qquad B_c = \frac{\Phi}{2\, h_c\, L\, k_{Fe}} \]

Bm is the average air-gap flux density, D the bore diameter, L the stack length, hc the yoke thickness and kFe the stacking factor, about 0.95. With D = 120 mm, L = 100 mm, Bm = 0.6 T and a 20 mm yoke, the yoke flux density is 1.49 T with 4 poles. With 6 poles it would be 0.99 T, and with 2 poles 2.98 T, which is impossible: as a guideline, it is best not to exceed 1.5 to 1.7 T.

More poles, from 4 to 6. There is yoke to spare. As kw goes from 0.960 to 0.966 (q falls from 3 to 2), N = 180 × 6/4 × 0.960/0.966 = 268. With 6 coils per phase, 45 turns per coil. The section falls to 2.13 × 30/45 = 1.42 mm²: 2 wires of 0.95 mm, and 60 groups fit. Current is 7.3 A, at 5.2 A/mm². Torque is kept and speed falls to about 1000 rpm. The theoretical power would be 2/3, about 3.7 kW, but it ends up around 3 to 3.5 kW because the power factor worsens with more poles and the motor cools worse at lower speed.

Fewer poles, from 4 to 2. With 1.6 T in the yoke the flux can only grow 8 %, from 0.00565 to 0.00608 Wb, rather than double. That gives about 170 turns per phase, almost the same: 29 per coil with 2.2 mm² of wire. Speed doubles, but torque halves and the power stays almost the same. The maximum air-gap flux density, about 4900 G, is far below the usual 6000 to 9000 G for 2 poles: that lamination is not meant for it.

Catalogues confirm it: IEC frame 132S usually offers 5.5 or 7.5 kW with 2 poles, 5.5 kW with 4, 3 kW with 6 and 2.2 kW with 8. For your lamination, the second graph of Turns by stator gives the flux density in the tooth and yoke with the new poles.

When changing poles, also check:

Changing the winding type or the pitch

Going from one layer to two or shortening the pitch changes the winding factor, and the turns must compensate for it. q is the number of slots per pole and phase, and y/τ the pitch relative to the pole pitch:

\[ k_w = k_d\, k_p \]

\[ k_d = \frac{\sin 30^\circ}{q\, \sin(30^\circ / q)} \qquad k_p = \sin\!\left(90^\circ\, \frac{y}{\tau}\right) \]

q123456
kd1.0000.9660.9600.9580.9570.956

A single-layer concentric winding occupies the same slots as a single-layer full-pitch lap winding, with the same distribution per phase: magnetically they are equal and the turns do not change. Only the end windings and the wire length change.

Example. Moving to a double layer with 7/9 pitch: kp = sin 70° = 0.940 and kw = 0.960 × 0.940 = 0.902. N = 180 × 0.960/0.902 = 192. There are now 12 coils per phase, so 16 turns per coil and 32 conductors per slot. The rule asks for 2.13 × 30/32 = 2.0 mm², but if they fit, keep the same 3 wires of 0.95 mm: 40 groups fit and the motor will run cooler. Leave room for the separator between layers, which the tool does not deduct.

Changing the parallel paths or the connection

With a parallel paths, the series turns per phase are those of all the coils of the phase divided by a. To keep N, each coil carries a times more turns; each path carries 1/a of the current, and the wire can have 1/a of the section. This avoids a wire that is too thick or too many wires in hand. The number of paths must divide the groups per phase exactly: with 4 poles in one layer there are 2 groups per phase, so a can be 1 or 2.

Example. With 2 paths, the base motor moves to 60 turns per coil and 1.06 mm². A 1.15 mm wire gives 1.04 mm², and 84 conductors fit. The copper in the slot is almost the same, but winding twice the turns takes more work, and the two paths must be equal so that no current circulates between them.

Rewinding from star to delta at the same line voltage is the case of the first example: N × 1.73 and S ÷ 1.73. Delta has another effect: it lets third-harmonic currents circulate inside. If the iron is working close to its limit, this shows in the no-load current and the temperature.

Other changes: power and material

In single-phase motors

The rules are the same, applied separately to the main and auxiliary windings, so that their turns ratio does not change. The capacitor changes with the square of the voltage, if the flux is kept:

\[ C_2 = C_1 \left(\frac{V_1}{V_2}\right)^{2} \]

From the result to the wire, with the tools

  1. Calculate the new turns and round them. Multiply conductors by section to get the copper per slot of the original, and divide it by the new conductors to get the target section.
  2. Look for wires with that section in IEC equivalences or NEMA equivalences. If you have them in the other standard, use the IEC / NEMA converter.
  3. Check that it fits with Turns by slot. Divide the result by 2 for a double layer. Leave some margin, because it does not deduct the wedge or the separator. Many thin wires fill worse than a few thick ones: if the turns rise a lot, the enamel takes up more room and the calculated section may not fit.
  4. If you do not know the original or want to cross-check, calculate with Turns by stator.
  5. If the poles, pitch or layers change, generate the layout in Three-phase windings or Single-phase windings.

Turns are whole numbers, so the flux density changes a little: new B = original B × calculated N / actual N. If you are between two values and it fits, choose the one with more turns: the motor will run slightly cooler, with a little less maximum torque. In the first example, 52 turns against 51.96 change nothing.

If you work with AWG gauges, each gauge is 26 % of section, and 3 gauges are double or half. For example, going from 230 V to 460 V doubles the turns and halves the section: from AWG 17 to AWG 20.

Checks when finished

The proportions in this guide start from a well-designed, undamaged motor. If the lamination stack has hot spots or the original motor was already overworked, the modification inherits those problems.

Related guides