How Neodymium magnets are made

Posted: 17th April 2018 by iulian207 in Projects








The method for manufacturing Neodymium Iron Boron magnets is as follows:

  • MELTING OF THE ALLOY UNDER VACUUM The Neodymium metal element is initially separated from refined Rare Earth oxides in an electrolytic furnace. The Neodymium, Iron and Boron and the other chemical elements are weighted  and put in a vacuum induction furnace to form an alloy. The mixture is melted due to the high frequency heating.


  • CASTING AND MILLING In simplified terms, the “Neo” alloy is like a cake mixture with each factory having its own recipe for each grade. The resultant melted alloy is then cooled to form ingots of alloy or cooled down by the so called „strip casting process“. The alloy ingots or the strip cast faces respectively are then broken down by hydrogen depreciation (HD) and then jet milled down in a nitrogen and argon atmosphere to a micron sized powder (about 3 microns or less in size). This Neodymium powder is then fed into a container, and oriented by a magnetic field in order to allow the pressing of the so called green compacts of the magnets.


  • PRESSING AND ALIGNMENT IN MAGNETIC FIELD There are three main methods of pressing the powder: isostatical pressing, axial pressing and transverse pressing. Die pressing requires tooling to make a cavity that is slightly larger than the required shape (because sintering causes shrinkage of the magnet). The Neodymium powder enters in the die cavity from the container  and is then compacted in the presence of an externally applied magnetic field. The external field is either applied parallel to the compacting force or perpendicular to the direction of compaction. Transverse pressing gives higher magnetic properties for the NdFeB sintered magnets. The other method of pressing is isostatic pressing. The NdFeB powder is put into a rubber mold and is put into a large fluid fled container which then has the pressure of the fluid increased. Again an external magnetizing feld is present but the NdFeB powder is compacted from all sides. Isostatic pressing gives the best possible magnetic performance for Neodymium Iron Boron. The methods employed vary depending on the grade of “Neo” required and are decided by the manufacturer. The external magnetizing field is created by a solenoid coil set either side of the compacting powder.

The individual powder particles of the NdFeB powder align with the magnetizing field that is applied – the more homogeneous the applied field, the more homogeneous the magnetic performance of the Neodymium magnet. As the Neodymium powder is pressed by the die, the direction of magnetization is locked in place – the Neodymium magnet has been given a preferred direction of magnetization and is called an-isotropic Rare Earth magnets exhibit uni-axial magneto-crystalline anisotropy i.e. they have a unique axis crystal structure corresponding with the easy axis of magnetization. In the case of Nd2Fe14B, the easy axis of magnetization is the c-axis of the complex tetragonal structure. In the presence of an external magnetizing field, it aligns along the c-axis, becoming capable of being fully magnetized to saturation with a very high coercivity.


  • SINTERING AND ANNEALING The  Neodymium magnet is then sintered to give it its final magnetic properties. The sintering process is carefully monitored (a strict temperature and time profle has to be applied) and occurs in an inert (oxygen free) atmosphere (e.g. argon). If oxygen is present, the resultant oxides destroy the magnetic performance of the NdFeB. The sintering process also causes shrinkage of the magnet as the powder fuses together. The shrinkage gives a magnet close to the required shape but the shrinkage is usually uneven (e.g. a ring may shrink to become an oval). At the end of the sintering process a fnal rapid quench is applied to rapidly cool the magnet. This is to minimize the unwanted production of other magnetic phases. A rapid quench maximizes the magnetic performance of NdFeB. Because the sintering process causes an uneven shrinkage, the shape of the Neodymium magnet will not be to the required dimensions.


  • MACHINING The next stage is to machine the magnets to the required tolerances. Because machining is required, the Neodymium magnets are made slightly larger when being pressed e.g. larger outer diameter, smaller inner diameter and taller for a ring magnet. Standard magnet dimensional tolerances are +/-0.1mm although +/-0.05mm is achievable at extra cost. The possibility of even tighter tolerances depends on the shape and size of the magnet and may not be achievable. For note, the Neodymium magnet is very hard. Trying to cut holes in NdFeB with a standard drill or carbide tip will blunt the drill bit. Diamond cutting tools (CNC diamond grinding wheels, diamond drills, etc.) and wire cutting machines (EDM) have to be used. The NdFeB swarf powder produced during machining needs to be cooled by liquid otherwise it may spontaneously combust. For Neodymium block magnets, there may be cost savings in using much larger magnet blocks made by isostatic pressing and cutting them into smaller Neodymium blocks of the desired size. This is done for speed and for mass production (where enough cutting and grinding machines are present) and is known as “slice and dice”. COATING Once the final dimensions for the magnet has been met by machining, the Neodymium magnet is given a protective coating. This is usually a passivization of the surface or a Ni-Cu-Ni coating. Other coatings may be possible. It is not recommended to use the magnet without any protective layer.


Prototype production phase done. Testing phase starting. Production phase started.



1. Introduction:

First BLDC motor was developed in 1962 by T.G. Wilson and P.H. Trickey unveiled what they called “a DC machine with solid state commutation”, it was basically because solid state thyristors appeared on the scene.
A permanent-magnet synchronous motor (PMSM) uses permanent magnets embedded in the steel rotor to create a constant magnetic field. The stator carries windings connected to an AC supply to produce a rotating magnetic field. At synchronous speed the rotor poles lock to the rotating magnetic field. Permanent magnet synchronous motors are similar to brushless DC motors.
The difference between BLDC and PMSM is that BLDC is driven by square waveform and PMSM is drive with a sine wave current.
 These motors require a variable-frequency motor inverter, ESC,  Frequency inverter.

With a strong background from Polytechnic University, Faculty of Electrical Engineering, electrical machines discipline, I started  10 years ago to modify some BLDC motors, and then designing my own. 

After making my own electric car conversion Opel Agila:, i got enough experience to understand what are the motor needs for an electric car.

A motor should have high range of RPM and high peak capacity to be able to keep the car in one (gear) second or third, and have 120km/h at 7000RPM for example.
Many simulations were done until the final version (after about 50 simulations with various magnets sizes, angles, dimensions, stator tooth sizes and shape, and different winding configuration. 
To be able to have a very small cogging torque and a sinewave back EMF the motor is consisted in three rotor slices with 3 degrees skewing; this also reduces harmonics and eddy currents losses in the magnets. The motor will work with sensorless sinewave controller or using resolver/ encoder.  

2. Motor sizing, slot pole count:

To decide what is the best motor for your application you need to take in account many of aspects.

  • You start from the power, speed and torque needed.
  • There are many type of motors good for a car conversion ( DC motors, AC motor, Permanent magnets motors (PMSM), Hybrid PMSM ( less magnet material and increased use of reluctance torque), Switch reluctance motor (no magnets, no clogging torque high efficiency, some disadvantages not high start torque, noise and vibrations )
  • For example BMW i3 has a hybrid design between PMSM and switch reluctance motor, by still using some magnets, they where able to create one of the best electric motors on the market.
  •  Higher RPM motors native offer higher power density, because for higher speed you do not need in increase the size of the motor, only for the torque and since the power of a motor is torque multiplied by speed you gain power easy in this way.  Mechanical power at the motor shaft  equation : P[W]=Speed [RPM] * Torque [Nm] / 9.55
  • If a motor has 65kw of mechanical output power the electrical power needed is bigger because of the efficiency. For a car a 60-65kW is enough to have good acceleration.
  • I decided to use 60 slots and 10 poles because this combination is offering 5 symmetries, and high winding factor. 0.966), if you use 12 poles then the frequency becomes to high and the motor eddy currents losses become to high, GM is using the same slot pole combination Chevrolet Volt, also many other producers.
  • We can say that for a higher torque we need more magnets, but also keep the frequency lower than 400-500hz, because higher magnets count will create higher frequency.
  • Motor Frequency calculator : f[hz]= Pole count (magnets nr.) x RPM / 120 Example: this motor has 10 magnets and 7000rpm then f=10×7000/120 = 583Hz. 
  • For this frequency is imperative to use high frequency lamination in the stator like NO-20 or equivalent. Thinner lamination will offer lower eddy currents losses. The thickness for this frequency should not be higher than 0.3mm
3. Choosing the Materials: 
  • Even if in the last 50 years the PMSM motor theory remained basically unchanged, the big improvement was in the materials area, especially in the magnets.
    This days the magnets are very powerful, and this allowed for very high motor densities to become a reality. Also with the use of thinner and thinner silicon steel laminations the frequency of the motors was increased 10 times versus of a conventional AC motor that runs at 60 or 50hz, this resulting in very high RPM and many pole pairs. For example regarding this motor a NO-02 material in the stator (0.2mm thickness) compared with M235-35A materials witch is 0.35mm thick, will give an improvement of 0.5% in efficiency, which is quite a lot can mean 400w of less power dissipation.
  • Neodymium Iron Boron is an alloy made mainly from a combination of Neodymium, Iron, Boron, Cobalt and of other transition metals and with varying levels of Dysprosium and Praseodymium. The exact chemical composition within NdFeB depends on the grade of the NdFeB magnet. Dysprosium and Praseodymium are added to improve the Hci (Intrinsic coercivity) of the “Neo” magnets.
  • Stronger magnets produce more torque so more power in the end. I decided to have N42UH magnets with phosphate coating, Stronger than N42 can not sustain high temperature operation so i stop at N42.

4. Motor topology, calculations and simulations: 

  • To be able to calculate and determine all the parameters of the motor a simulation software is needed, they are extremely expensive unfortunately.
  • Even with a tool to simulate the motor, you still  need to know and understand all the parameters, to determine best motor topology for certain application, so is almost pointless for somebody to try the software if there is no university background for electric motor like me for example.
  • The software can determine many parameters, but it can determine them well if you feed with the correct data, correct materials and correct assumptions.
  • Underneath is the 60 slot and 10 pole model with transparent core to be able to visualize the coils and magnets.
  • Mechanical analysis is done in a separate program to evaluate the centrifugal forces that tend to throw the magnets outside the rotor.

This test is very important because you want to make sure the magnets will not fly outside, but you also want to have them as close to the exterior with bracket as thin as possible.

My test bench setup

Skewed magnet representation: 

Stress analysis. A force of 2700N was applied to each magnet in outer direction to simulate the centrifugal force produce by the rotor rotation at 7000 RPM. The limit force for the material is about 250 Mega Pascal. So we are way under the limit.

Motor simulated at different parameters and loads


Main Electrical characteristics:

  • Peak Power …………………………………….130Kw, 1min.
  • Nominal Power(S1 continuous) :……… 65 kW
  • Max speed: ………………………………. 8000 RPM
  • Supply DC bus Voltage: ………………… 280Vdc
  • Supply DC bus Voltage: ………………… 280Vdc
  • Peak Torque :………………………………………… 200 Nm
  • Very low cogging torque (Zero current)………….. 2,5Nm (lower than 1% of the motor torque)
  • Ac Supply……………………………………. 151 VAC
  • Turns: 2, parallel paths: 4

  Mechanical Characteristics:

External diameter 292mm

length : 190mm

Weight ~ 36kg

The video represents the flux polarity, is displayed radially. As we can see the field is moving form coil to coil in front of the magnets.

The motor has big peak capability ~2x and enough iron not to saturate the core. At continuous operation the flux in the tooth and back stator is not higher than 1,5T

    Winding distribution and parameters:

65kW Permanent magnet synchronous motor

  • Connection Type : Star
  • Number of parallel paths: 4
  • Number of turns: 4
  • Wire diameter 0.61mm
  • Number of strands in hand 23

Calculations: Conductor area is 6.75mm2 per each path. Using 0.61mm diameter wire we end up with 23wires in parralell.
Since we have 4 parallel paths total conductor area will be 6.752 x 4 = 27mm2 resulting a current density of 9,63Amp RMS/mm2. With a water jacket around the motor to cool down, the motor temperature will be about 90 degree celsius from the simulations.

The efficiency map in Non linear mode took 5 hours to simulate with core i7 7700 processor. In practice we expect some variations due to the materials, and motor construction tolerances.



Rotor balance evaluation