Projects:2020s1-2430 The Ball Bearing Motor Mystery

Project team

Honours students

• Fengyuan Zhan
• Liam Martin

Supervisors

• Prof Derek Abbott
• Dr Andrew Allison

General project description

The ball bearing motor is a mystery because to this day no engineer knows how it works! No one understands the physical principle at all. The project attempts to solve the ball bearing motor mystery by doing some experiments and investigating the theories on why it rotates. It may not be useful for large motors, but it may be interesting for micromotors and micropumps that have numerous applications.

Abstract

The ball bearing motor, also known as the Huber motor, was first discovered in 1959. It is simply two ball-type bearings connected to a conductive shaft as shown in Figure 1. The motor will continue to spin when a power supply is connected to the bearings and given an initial spin. This motor can spin in either direction and will work off of an AC or DC supply. The motor is also known infamously for being unpractical and self destructive as it does not produce much torque and cannot run for a long time due to arcing and immense heat generated in the bearings which causes them to deteriorate quickly. The motor was therefore considered useless for a long time until recently where it is believed it could have some potential application in the field of micro-electromechanical technology. Before the motor can be developed for this field, it first needs to be understood on why it works.

Figure 1: Ball bearing motor. After ref [1]

Aim

This aim of this project is to find out more useful information on the ball bearing motor and try to relate the characteristics of the motor to one of the suggested theories on how the motor works. Several motors shall be built and tested in different environments with their speed and torque measured. Additionally the motor will be simulated in the simulation software COMSOL that will allow for testing in a virtual environment where bearings will not deteriorate.

Background

The principles of the motor are still a mystery in academia and there are still many disputes on how the motor works. These three main theories are the most prominent.

Electromagnetic effect

The electromagnetic theory was raised by Gruenberg in 1977 [1]. It states that each ball has primary current density J0 and primary magnetic field B0. When the ball is moving, a new current density J1 and related magnetic field B1 will be produced [1]. The constant interaction between J0 and B1 and between J1 and B0 produces a force on the ball that causes the cycle to repeat and rotation is therefore sustained [1].

Based on the analysis of the electromagnetic effect theory, the motor torque is proportional to the angular velocity and the squared value of the supply current [1].

Figure 2: (a) Zero order current and magnetic fields (b) First order current and magnetic fields. After ref [1]

Plasma discharge

The plasma discharge theory was proposed by Polivanov, Netushil and Tatarinova in 1973 [2]. They described the Huber effect is due to the plasma discharge (sparking) which occurs at the points of last contact in the balls and the race when the motor is spinning [2]. A force is produced as a result of the interaction of induced currents within the magnetic field disturbing the symmetry of the current and flux distribution established by previous contact points. This interaction is repeated and sustains rotation.[2]

As a result of investigation, they summarized the rotating force is proportional to the squared value of currents [2].

Figure 3: (a) Current distribution of stationary ball (b) Current distribution of rotating ball. After ref [2]

Thermal expansion

The thermal expansion theory was proposed by Marinov who indicated that “the ball bearing motor is not an electromagnetic motor but a thermal engine” [2]. When the current travels through ball bearing, the contact points between balls and races will become hot due to the high resistances [2]. This leads to expansion in the ball bearings at the contact points and changes the shape of balls to elliptical [2]. With an initial spin, the elliptical points will push against the races with a sideways component of force and hence rotate. This cycle repeats itself and hence rotation is sustained [2].

Figure 4: (a) Ball ball stationary (b) Ball bearing rotating

Simulation

A simple version of the motor was created in the simulation software COMSOL as the regular ball bearing motor is too complex to simulate. The purpose is to simulate the motor in a virtual environment where the motor can be tested without being subjected to bearing deterioration that affects accuracy of the results. The simple version eliminates bearings from the design and instead considers two conductive discs in their place.

Figure 5: Motor in COMSOL with red arrows indicating path of current

Experiment

Motor size

The motor built for these experiments is believed to be the smallest version ever built. A ball bearing motor typically has a shaft diameter size of 10mm whereas this ball bearing motor has a shaft diameter of 1mm. The reason for scaling down the motor design is to scale down the destructive currents experienced in the bearings which may help prolong their service life. A housing was made for this motor to ensure it had good alignment. Two metals screws were drilled into the housing and would make contact with the bearings. These screws acted as the terminals for the bearings and would provide the current to the bearings.

Figure 6: Typical ball bearing motor and experimental ball bearing motor

Measuring equipment

An Arduino Mega board along with code downloaded from the internet specifically written for measuring angular speed and torque was used to measure this motor's characteristics.

To measure the angular speed of this motor, an encoder and photo interrupter are used. An encoder is simply a disc mounted on the motor shaft with some sections of it cut out or drilled out, so that light may pass through that section and the rest of the encoder will not pass light through. When the encoder is turning, it will periodically obstruct the photo interrupter. The photo interrupter will sense when it is being obstructed and send a signal to the Arduino board. The Arduino board uses this combination of the obstructed signals over a period of time to calculate angular speed.

To measure the torque, a prony brake and load cell are used. A prony brake is simply a clamp and arm. The clamp is mounted to the shaft tightly but not too tight as the shaft needs to be able to turn while the clamp and arm are stationary. The load cell is a steel bar and strain gauge which can detect forces applied to it. When the motor is turning, the prony brake will want to move from the friction of the motor shaft. The load cell is placed under the arm of the prony brake which apply a downwards force due to the friction from the turning shaft. The load cell signal is then amplified and sent to the Arduino board and calculates the torque. A potential problem with this method is that the ball bearing motor is known to only produce little torque and this setup may stall such a small motor, in which case the prony brake should be taken off and only angular speed is measured.

Figure 7: Arduino Mega board and LCD screen

Figure 8: Experiment setup

Experiments

This motor was to be tested in three different environments; normal conditions, submerged in demineralised water, and submerged in nitrogen gas. The purpose of submerging the motor in demineralised water and nitrogen gas is that they will suppress arcing in the bearings and act as a heat sink for the motor which will help prolong its life. The reason the water needed to be demineralised is because normal water conducts electricity whereas demineralised does not.

Liquid metal motor

Another experiment has been designed the eliminates the bearings from the design and instead considers two conductive discs submerged in mercury in their place. Power is supplied to the discs through the mercury. If this motor works, it will allow for prolonged testing and may additionally disprove the thermal expansion theory. It should be noted that bearings are still used in this design but only the secure the motor in place.

Figure 9: Liquid metal motor design

Power supply

A power supply 30V 20A power supply was used for this experiment. The ball bearing motor typically runs on a much larger current but due to its reduced size, this power supply was an appropriate choice.

Figure 10: Power supply used in experiment

Results

Simulation

The group is still performing simulations at the time of writing this and no results have been made yet.

Practical

This motor was operated from a range 0.5A to 3A, The motor was only increase to a max current of 3A because at this current the bearing shape deformed and became unusable, likely due to immense heat generated in the bearing. Unfortunately this motor failed to sustain rotation on its own for these currents. Without a current applied, the motor shaft had good rotation but as soon as the power supply was connected and turned on, the motor became more difficult to turn. When the power supply was disconnected the shaft rotation would become better again.

It was speculated that the reason for poor rotation was heating and expanding in the bearings making them difficult to turn. Therefore, the motor was tested in demineralised water which would help cool the bearings. However, the motor still failed to turn and would become more difficult to turn when the power supply was connected.

Due to the failed results from the motor in air and demineralised water, the nitrogen experiment was cancelled as it is highly expected it will yield similar results.

The mercury motor experiment has not been commenced at the time of writing this and hence no results or conclusions can be made. This is due to delays and difficulties brought on by the COVID-19 pandemic delaying the project and staff with expertise on handling and using mercury and laboratories equipped with measures for using mercury not being available.

Conclusion and discussion

Simulation

The group initially had trouble acquiring the software as the group encountered several delays in the project. Additionally due to social distancing, the group had difficulty getting training for using the software as this had to be done virtually. These factors severely limited the groups ability to use the software.

Practical

There are several theories as to why this small motor failed to operate.

First is the ball bearings bulging and making the bearings ineffective. It is known the motor does not produce much torque and it may not be able to overcome the bearings bulging too much. This theory seems the most likely due to retarded rotation the motor experiences when a power supply is connected.

Secondly is the quality control of the bearings. These bearings were bought online through eBay due to them being substantially cheaper than ball bearings of the same size from local bearing suppliers. However, it was noticed that these bearings were not identical to each other and some of them had a slightly larger inner diameter than others. Some bearings were also better at turning than others. For the experiments, bearings were compared to make sure similar sized and good turning bearings were chosen for the experiment but it is possible that due to poor quality, these bearings do not work well for the motor. This may also be related to the first theory.

Thirdly is motor misalignment. A housing that the motor would sit in tightly and be secured down was built for the purpose of making sure the motor is aligned, however it is possible that the motor is still not perfectly aligned due to subtle bends in the fragile shaft and unequal balances on the encoder that are not noticeable.

Finally, whatever sustains rotation for this motor may not work when it is made this small. As the motor is still a mystery, it is hard to say if it could work when it is this small.

Due to this motor failing to sustain rotation, it is a possible indication that this motor may is not be a good choice for a micromotor.

Future work

This project still leaves the ball bearing motor mystery unsolved. Future projects into the ball bearing motor could focus on:

• Can the ball bearing motor work when it is small? - An investigation should be taken to determine if the motor can possibly work when it is small. A variety of different sized bearings and shafts should be used to see how size affects the performance of the motor and at what sizes the motor will work.
• Simulating the motor - A simulation study should allocate more time to simulating the motor with computer software because COMSOL is a complex software the requires a lot of experience in order to use it effectively.
• Using a bigger motor - The same experiments in this project could be repeated with a bigger motor that works. Recommended shaft size of at least 3mm as they have been known to work in past projects [2].

References

[1] Gruenberg, H. "The ball bearing as a motor" American Journal of Physics, no. 46 (1978): 1213-1219.

[2] Choo, J.L Soong, W.L, Abbott, D. "Toward Characterization of Huber's Ball-Bearing Motor" Proceedings of SPIE, no. 5649 (2005): 700-720

[3] Thompson, B. Shen, Y. Tay, K. Soong, W.L. Davis, B.R. Abbott, D. "Investigation of the Huber effect and its application to micromotors" Proceedings of SPIE Electronics and Structures for MEMS, no. 3891 (1999): 178-183

[4] Lauterbach, A.P. Soong, W.L. Abbott, D. "Investigtion of small motors operating under the Huber effect" Proceedings of SPIE, no. 4236 (2001): 306-318

[5] Hayward, J. Saing, R. "The ball bearing motor mystery 2019" The University of Adelaide, November 29 2019, Accessed on: March 17 2020. [Online]. Available: https://projectswiki.eleceng.adelaide.edu.au/projects/index.php/Projects:2019s1-142_The_Ball_Bearing_Motor_Mystery