Customised Brushless DC (BLDC) motors

March 2015 (last updated: 10 March 2015)

Rewinding larger brushless DC motors in the 30mm to 75mm diameter range.

I will focus on the larger ‘drum’ outrunner type motors commonly found in photocopiers and larger printers (lots has already been written about floppy drive and CD-ROM motors). I’ll also favour practical over theoretical, as there is loads of theory on the web but not so much practical advice for larger motors.

When working on motors of this size, it is very useful to have access to a metal lathe. You will also need a digital caliper and a digital multimeter, and some way of measuring rpm.

This page is a work in progress, mostly notes to self…

The main parts of this page:

  • Assessing a stock motor
  • Modification experiments
  • List of disassembled motors

Why BLDC outrunners?

Brushless DC motor advantages:

Brushless DC motor disadvantages:

Outrunner advantages:

Outrunner disadvantages:

Large BLDC motors

  • outrunners
  • smaller motors have 9 stator arms, larger ones have 12
  • motor shafts are 6mm to 8mm diameter, are splined where they are pressed into the top of the rotor, and feature a helical gear – larger motors feature a brass ‘boss’ into which the motor shaft is pressed
  • motor shafts run on miniature deep groove ball bearings, one at the top of the stator and one below (e.g. 696ZZ) – these bearings are good for up to 36k rpm however as they are deep groove they are not suitable for high axial loads
  • rotor cans range in size from 30mm up to 75mm diameter and more (a CD-ROM motor is typically 20-25mm)
  • the control electronics are mounted on a PCB next to the motor, and the whole assembly is pressed in to a steel plate – for larger motors the steel plate is replaced with a useful ~3mm cast aluminium plate which can be re-used in a customised motor (the steel plates are less easy to re-use)
  • the control IC is usually either not marked with a part number, or where marked the datasheet for the IC is not available (i.e. it is a proprietary IC)
  • powered with 24V DC, controlled by a DC motor controller board (a separate board buried in the donor machine)
  • as wound and controlled, they are designed to operate in the 2,000-2,500rpm range

Online resources

Assessing a stock motor

Read this before pulling the hapless victim lucky contestant apart. It is easier to work out how to get where you want to be, if you know where you are now (I am so freakin’ zen).

Measuring motor constants

You will need to repeat these steps after you have modified and rebuilt the motor. Before disassembly take the following measurements using the methods described here.

For accurate measurements, you will need to de-solder the motor connections from the PCB and attach some fly-leads. There are usually access holes in the base plate so you don’t need to disassemble anything.

  • Kv or Volts per rev
  • Io or no load current (note: you don’t need two multimeters as described in the link above, you can do it with one, just be quick about changing from volts to amps)
  • Rm or winding resistance

After disassembly you can move on to:

  • stator dimensions: number of stator arms, wire diameter, number of winds per arm, winding configuration, stator thickness
  • rotor dimensions: internal and external diameter

Disassembling a stock motor

Before disassembling the motor, make sure you follow the steps above to assess the donor motor. This guide assumes that you are not intending to use the onboard controller circuit, and that you may wish to re-use the rotor and the backplate.

Remove the rotor:

First check if there is a circlip securing the rotor in place. Look down the motor shaft from the output end – if there is a circlip then remove it.

Hold the motor assembly by the backplate in one hand, and a small hammer in the other. Tap the end of the output shaft with moderate force until the rotor falls out. If you see no movement after a few taps, check again for a circlip!

A support plate between the rotor and the backplate would be a better method of pressing out the shaft, however there is usually very little space in there and motors do not follow a consistent design, so a new plate would need to be made for most motors. Also the shafts are a close sliding fit in the bearings, so a lot of force is not required. The hand-held method at least minimizes damage to the assembly, as there is a natural reluctance to bash your own hand with a hammer.

Remove the stator:

After removing the rotor you should see three screw at the top of the stator. These secure the stator to a boss that is pressed into the backplate (for larger motors the boss is a cast part of the backplate).

Undo these three screws. You should now be able to move the stator slightly in an upward direction off the boss – this will make the next step easier… Cut the coil wires that are terminating on the PCB – use a pair of scissors to get in there, thicker wires on larger motors are more troublesome.

Unwind the wire:

A pleasingly soporific activity: relax and unwind.

Larger motors have a two-part plastic insulation sleeve. Smaller motors have the more usual resin coating.

BLDC and electric motor design

Power and torque

Torque is the amount of rotational force a motor exerts. It is expressed as: force x distance. You can generate a torque of 1 Newton-meter (Nm) by pulling with a 1 Newton force on the end of a bar 1 meter long. Power is torque x rotational speed (Watts). Torque may decrease with higher speeds, but power will increase due to the higher rate of delivery of the (lower) torque.

Don’t confuse the output power of a motor with the input electrical power, which is also expressed in Watts. The ratio of these two values is the motor efficiency (this is somewhere in the region of 75% for our BLDC motors). Consumer-grade equipment manufacturers always quote the higher input electrical power – your Dremel 3000 may be advertised as “130W” but its universal motor is less efficient that BLDCs especially at lower speeds, so you are likely getting between 50%-60% efficiency (65W to 78W output).

The key requirements for a motor are usually expressed as a specific torque delivered at a specific speed (this may be expressed as a range). These requirements will depend on the application – for some examples look at calculations for matching propeller size to motors within RC forums and sites.

The power output of a motor is largely incidental, just a function of these two requirements. However, many of the calculations you see online will use this power figure – it’s not wrong to do this, however when designing a motor it’s useful to keep in mind that power is torque and speed.

Critical design elements affecting performance

Listed in order of importance. There are many, many more design elements that affect performance, but these are the top six that we can adjust in the real world of modifying stock motors.

1/ stator volume – sheer brawn

The amount of power a motor can produce is proportional to the physical volume of the stator (for our BLDC outrunners) – i.e. the mass of copper and iron. You can tweak other design elements of a motor to improve output (significantly), but all other motor elements being equal: a motor with twice the stator mass will output roughly twice the power.

2/ heat dissipation – it’s all about keeping cool

For a given motor size (see above), the most critical design element affecting power output is cooling. The more current we put into a motor, the torque it will produce but also the hotter the windings will become. There are physical limits as to how much heat/current we can put into a motor, too much heat and the wire insulation will burn or the magnet ring will be demagnetised (esp. for high-power neodymium), and the motor will literally burn out.

RC brushless motors output high power for their physical size because they are cooled with a howling gale of prop wash (or cold water in the case of boats). There are few other applications where such an open design is acceptable.

Cooling is heat dissipation, or rate of energy dissipation. It’s an arms race: the amount of heat the windings generate vs the amount of heat we can dissipate.

3/ winding wire volume

The more copper, the less heat generated (copper loss) for a given electrical input power, and the stronger the magnetic field. As the size of the stator slots is fixed, winding wire volume is just a function of wire thickness and number of turns per stator arm. The goal is to fill the slots in the stator as full as possible, this is called winding density.

Assuming we have achieved optimal winding density, more winding volume requires a physically bigger stator – both as a physical frame for the copper and also to increase the magnetic flux capacity. This is simply increasing the volume of the stator, see point 1 above.

4/ winding wire thickness and number of turns per stator arm

For a given winding wire volume, we can tweak the characteristics of a motor by varying the wire thickness and number of turns. Here is where a base design (physical size) can be adjusted to fit the motor design requirements.

In practice, when modifying an existing motor, we are constrained by the number of turns of a given wire thickness that we can physically fit on the donor rotor. There is an excellent practical discussion on winding here including good tips for technique.

Assuming a fixed winding wire volume (which in practice and with due care in selection of wire thickness is what we have with the stator from a donor motor):

  • thicker wire (fewer turns):
  • yields more power and higher RPM, at the expense of lower-speed torque
  • higher motor efficiency (important for battery-powered applications)
  • requires higher voltage to achieve the higher speed
  • (note: if high efficiency was critical to the design but not high RPM, you could simply not rev the motor to its maximum)
  • thinner wire (more turns):
  • yields more torque at lower RPM, but lower overall power due to lower top speed
  • lower motor efficiency (this is a less important consideration for mains-powered applications)
  • requires a lower voltage as top speed is lower

This is just a brief and practical summary, to guide design choices – there is a great deal more to it than this!

5/ winding pattern

Delta vs star/wye.

6/ magnet power

Affects strength of magnetic field – stronger field means higher torque which translates to more power and efficiency.

I don’t claim to understand how magnetic power is measured or even how relevant it is to BLDC motors, however some understanding of the relative strength of different magnetic materials is useful to know. I looked at 25x10x3 rectangles on this website, and collected the cost per magnet, strength and also got the temperature information from here.

In the temperature column, the first value is the maximum operating temperature in Celcius. The next value is the temperature coefficient – I don’t understand this completely but ferrite loses magnetic strength the fastest when exposed to heat (with -0.2 value).

Neodymium (N42)3.4kg£1.58150 / -0.12
Samarium Cobalt (SmCo26)2.04kg£1.96300 / 0.04
Ferrite0.24kg£0.11300 / -0.2
Flexible magnet (???)unknowncheap!unknown

Rare earth magnets are expensive, and only come in a set range of sizes. If you can compromise on efficiency, and produce the required power output by changing other design elements (e.g. motor size and cooling), then you may end up with a more economically efficient motor (cheaper), at the expense of electrical efficiency.

7/ rotor diameter

Affects torque. Short and fat vs long and thin. Link to stator volume.

Modification experiments

Proper scientists change one thing at a time and then observe changes in results. I aspire to be a proper scientist.

I wanted to see in the real world what effects changing different design elements would have. To conduct these experiments I used four identical donor motors – so the motor physical size will be fixed and I will change other variables (so, a real world scenario when modifying stock motors).

Given that we have fixed donor motors, there are only a few things we can change to affect performance. Cooling will be the limiting factor in actually running the motors, as the stock rotors do not have cooling holes, however by measuring motor constants we can fairly accurately predict higher-speed performance. I will use a standard hobby RC motor ESC and a 24v 60A benchtop power supply to power the motors.

Things to change:

  1. magnet strength
  2. winding wire thickness
  3. winding pattern

I have omitted winding wire volume from this list, as we know from the wealth of information online that more is better.

I will need a combination of two rotors (one with standard ferrite ring, one with neodymium magnets), and four stators (thin & thick wires wound in delta & wye patterns). I am not expecting this to be a 100% comparable test, as the combinations of wire thicknesses and winding patterns may result in different wiring volumes.

The selection of thinner & thicker wires is largely based on guesswork. The stock wires were

Stock motor assessment

Motor constants. Physical dimensions.

Physical dimensions:

  • wiring: 0.405mm diameter (26AWG), wye pattern, 83 turns per stator arm, 3225mm of wire per stator arm (415.5mm^3)
  • stator: 9 arms, 12mm thick, 43.0mm diameter
  • rotor:
  • magnetic ring: 2.6mm thick,

Test 1: increase magnet strength


Test 2: thinner wires, delta pattern (standard magnet)


Test 3: thicker wires, delta pattern (standard magnet)


Test 4: thinner wires (neodymium magnet)


Test 5: thicker wires (neodymium magnet)


3/ thicker wires



List of disassembled motors

BH80FT43-02 (Japan Servo)

  • source: Canon photocopier

BH60FT20-03 (Japan Servo)

  • source: Canon photocopier
  • rotor: 58.85mm ID, 61.6mm OD (2.75mm thick)
  • magnet ring: TBC
  • shaft: 6mm diamater, helical output, splined and pressed directly into rotor, bearings 696ZZ
  • stator: 19.0 ID, 51.5mm OD, 16.0mm depth, 0.62mm OD wire, 34 winds,

One thought on “Customised Brushless DC (BLDC) motors

  1. Ash Cain

    I am interested in making custom brushless motors. I would appreciate if you could send me the results for your experiments. These results would be very useful in creating future motors.

    Thanking you,
    Ash Cain


Leave a Reply

Your email address will not be published. Required fields are marked *