Projects:2019s1-142 The Ball Bearing Motor Mystery

Supervisors

• Prof Derek Abbott
• Dr Andrew Allison

Honours students

• Jayden Hayward
• Richard Saing

Abstract

Based on the Huber Effect, the ball bearing motor can be made to continuously rotate in either direction when supplied with either a DC or AC supply. This phenomenon was first observed in 1959 and has since motivated a number of theories to explain the underlying principles behind the motor's operation. This projects aims to test the validity of some of these theories by taking a modular approach to testing the ball bearing motor. An attempt to evaluate the electromagnetic behaviour is made with the use of the a simulation software called ANSYS Maxwell and the relationship between angular velocity and motor torque are also obtained by measuring a physical motor with the use of a load cell and tachometer. The completion of this project hopes to assist future research into the application of the ball bearing motor in micro electrical-mechanical systems whilst leading further research into the Huber Effect in the right direction.

Aim

The main goal is to find out a little more about this motor. Specifically, the aim is to establish the relationship between the angular velocity and the torque because there are actually two theories based on the principles of electromagnetic that are similar but one predicts a linear relationship between speed and torque while the other predicts a squared relationship. By establishing this relationship, the theory that is more likely to be correct can be seen and hopefully future research can be led in the right direction. A secondary goal is to test out a modified version of the ball bearing motor. Instead of applying electricity through the ball bearings, electricity is applied to a solid disc instead. If this works, it will allow for prolonged testing as it will not deteriorate, expand or arc like the ball bearings. This modified motor will then be used to achieve the main goal. The behaviour of the electric and magnetic fields will also be observed by simulating the modified motor using ANSYS Maxwell.

Background

Ever since the Huber effect’s realization, many researchers have made hypothesis’ on how they believe the Huber effect works. Nevertheless, there has been no definite evidence to support any of these theories despite the diverse experiments, consequently the principle of the Huber effect remains unknown. The three main theories hypothesizing this effect are: the electromagnetic force effect, the thermal expansion effect and the electromechanical effect, all of which will be discussed here.

Electromagnetic force effect

The cause for the Huber effect that it uses the electromagnetic force effect, was first proposed in 1977 by Gruenberg - an electrical engineer in Canada. He states that inside each ball bearing there exists an initial current density, J0 with its corresponding magnetic field, B0. When the balls rotate with angular velocity ω, then a new current density, J1 will be generated due to the motion displacement that occurred in the initial magnetic field, B0. A new magnetic field B1 is produced as a byproduct of J1. As the ball continuously rotates then so does the process continuously repeats. The interchanges of between J0 and B1, and J1 and B0 is what causes a torque to occur. One remark is that the initial current density, J0 and the initial magnetic field, B0, is not sufficient to produce a starting torque, meaning that the motor is not capable of self-starting. A torque can only be produced once the motor is manually given a spin [1].

In the ball’s initial state without any external forces, the zero-order fields around the ball: J0 and B0 take shape as shown in part (a) of Figure 1. Once exposed to an external force causing the ball to spin, the fields around the ball in the bearing to take shape as shown in part (b) of Figure 1. Notice how the current density is now shifted to go along the “Y” axis and the magnetic field is shifted to go in the “Z” axis. This orientation of the fields is believed to be the cause of the continuous ball rotation. Through detailed analysis, Gruenberg concluded the relationships between torque, current and angular velocity was a squared one [1].

Figure 1: Current and magnetic field distribution of each ball in the ball bearing for zero and first order fields. Ref [1]

Thermal expansion effect

Thermal expansion is the increase in volume of a material as temperature is increased, typically expressed as fractional change in volume per unit temperature. Marinov hypothesized that the Ball Bearing Motor is a thermal engine, where it is able to produce motion from electricity without the use of magnetism. He states that the torque (or the main cause of the rotation) is from the thermal expansion of the balls in their bearings. That the contact points between the ball bearing and the races is where it heats up the most, causing the ball bearings to physically expand, yielding a small “bulge” at that point. Noting that not the whole ball becomes heated but only contact points of the ball as shown in Figure 2 with red dots. At these points is where the ohmic (junction between two conductors with linear current) resistance is much greater than the resistance across the entire ball [2].

Figure 2: Contacts of a ball with inner and outer races

This is where there ball dilates slightly, and since both the balls and races are made of firm steel; a slight dilation of the ball will create large torque. The motor will then rotate in the direction of provided initial torque. The motor undergoes its rotational movement through the repetitive heating of the ball bearing contacts with its electrical contact with the races. During the rotation, the ball’s “bulge” moves from one position to the next and the radius of the “bulge” becomes equal to the radius of the ball. This radius becomes bigger and bigger with more torque produced, resulting in a mechanical motion. There is not enough evidence to support Marinov’s theory, however the fact that the ball bearings become significantly hot and deteriorate, give reason to not neglect such a theory.

Electromechanical effect

In 1973, the physicist, Polivanov, proposed that Electromechanical was cause of this Huber effect. That the discharge takes place at the contact points between the ball bearings and the races, producing a force sufficient enough to sustain rotation. This force is produced by the applied voltage with an assert being that the current is prone to flow in the opposite direction to the applied voltage. The interactions between the current and the magnetic field causes a disturbance to the symmetry of the current and flux confined by previous contact points. As the points of contact are being continually established through rotation, the force is sustained [3].

In Figure 3, when the ball starts spinning, the old contact point shown in the green circle of part (b) is no longer the contact point. This in turn emerges a new current distribution without neglecting the old distribution, therefore the symmetry is disturbed. Hence the ball will continue to rotate from the momentum of the event of plasma discharge.

Figure 3: (a) Cross-section of current distribution in a stationary ball. (b) Cross-section of new current distribution when the ball is rotated

Design

Previous works have shown that the ball bearing motor exhibited several self destructive behaviours when subjected to high currents. This has historically limited the measurements of the ball bearing motor as heating of the metal components expands and seizes the ball bearing races. One way to minimise these self destructive effects is with the utilisation of a liquid metal that submerges solid disks. This metal liquid will act as a slip ring that allow for current to be supplied to a solid disk. Not only will this method limit the heating of the ball bearing races but will allow for the Huber Effect to be investigated when the electricity is not being applied to a rotating but stationary component. The liquid metal will also act as a fluid heat sink to minimise the heating and expanding of the metal disks.

Liquid Metal

Gallium is chosen for this experiment due to its relatively low melting point (30 degrees Celsius) and its non-toxic behaviour which allows it to be safely handled. It is highly corrosive to other metals, however, it has also been proven to be an effective cooling agent with low viscosity making it very adequate for this project.

Galinstan, an alloy consisting of gallium, indium and tin, was also considered due the similar properties it shares with pure gallium. It was, however, dismissed due to its high cost when compared to gallium.

Solid Disks and Rods

A list of materials including; tungsten, copper, aluminium and stainless steel were tested against gallium to see which would be best suited for the manufacturing of the solid disks. Copper and aluminium showed obvious corrosion when placed in liquid gallium for 30 minutes. Whilst copper exhibited clear signs of becoming brittle, the aluminium rod that was tested completely snapped in half. Tungsten and stainless steel on the other hand showed little to no corrosion at all. Ultimately the solid disks were made out of stainless steel due to its relatively low cost and availability but the rods were made out of tungsten.

The following video was made to show how tungsten, copper and aluminium reacts with gallium: https://www.youtube.com/watch?v=nglK-_yMVz4

Measurement Method

To measure the torque of the motor, the shaft of the motor were fitted with a peg like structure that will push against a load cell when the motor is operated. The peg (also referred to as the torque arm) were designed in Ansys Spaceclaim and 3D printed using polylactic acid (PLA). The load cell is a small steel bar with strain gauges attached onto its sides. When the load cell is bent, the strain gauges elongates causing its electrical resistance to change. This change in resistance causes a change in voltage which is amplified and measured using an Arduino board. The speed of the motor will be measured using a laser tachometer. To increase the accuracy of the tachometer, an encoding wheel also designed in Ansys Spaceclaim and printed using PLA was also used. The encoding wheel was fitted onto the shaft and the tachometer is pointed directly at it.

Power Supply and Current measurements

A large amount of current is required to operate the ball bearing motor, hence a welder that is capable of supplying 78 volts and between 40 to 130 amps (AC), as shown in figure 4, was used. The welder used is relatively old, hence to measure the current going through the motor, a Hioki multimeter was used.

Figure 4: Welder used to operate the ball bearing motor

Results

The below setup is what was used for both experiment 1 and 2 (except when the position of the ball bearings were changed). Numbered are the different aspects of the setup: 1) Two clamped ball bearings on a metal shaft 2) Two solid metal discs partly submerged in gallium 3) Copper wire wrapped around tungsten rods that are connected to terminals of a welder 4) Torque arm 5) Encoder wheel.

Figure 5: Setup of gallium experiment

As a proof of concept, a smaller ball bearing motor was built and recorded. Video can be seen here: Ball Bearing Motor Construction.

Simulations

Ansys Maxwell was used to simulate a simplified, stationary ball bearing motor. Results show similarities to electromechanical behaviour described by Polivanov. Figure 6 shows that the current density at the point of conduction is at its highest and decreases as it moves closer to the center of the disc.

Figure 6: Magnetic field and current density behaviours. Left - Magnetic field around rail, Middle - Current density in solid disc, Right - Current behaviour described by Polivanov [4]

Experiment 1 - Modified Motor

To test the modified motor we submerged two discs in gallium as well as two tungsten rods. The wires are wrapped around the rods so that current can be conducted through the gallium and the discs. However, it was found that the modified motor did not spin. It is believed that there are two possible reasons for this results. One possible reason for this is that the disc does not move relative to the point of conduction and it does not push against a hard surface unlike the ball bearing which pushes against the inner and outer races. If this was the case, this result slightly supports the electromechanical theory as the basis of the theory requires the motor to move forward and not just rotate. Another possible reason is that it was assumed that discs and ball bearing races of similar sizes would work the same. This experiment suggests otherwise and it is possible that the torque of the disc could not turn the large shaft and ball bearings. It is possible that the motor would have worked if the size of the motor was reduced. If this was the case, the electromagnetic theories still holds merit.

The ball bearings were then moved in between the two solid discs and containers of gallium as shown in Figure 7. Applying the current through the gallium and giving the shaft an initial torque did not cause the discs to continue spinning. One important remark to take away from this is that the current needs to flow through the ball bearings for it to work, it simply cannot pass past the bearings.

Figure 7: Setup of the Gallium experiment with the ball bearings on the inside

Experiment 2 - Regular Ball Bearing Motor

The welder was connected up to the ball bearing motor and set to 58.5A. The welder was then turned on and the shaft was given an initial torque, the shaft then rotated at high speeds. The torque values and rpm values were recorded as described in design section. Both these sets of values were then plotted against time as well as against each other. This was repeated for the currents: 60A, 60.9A and 62.8A. It was found that the torque increased with current. The angular velocity, however, was inconsistent at higher currents. The relationship between the torque and angular velocity was found to be too inconsistent to support either electromagnetic theories.

Figure 8: Top left: Relationship between torque and time, Top Right: Relationship between angular velocity and time, Bottom: Relationship between angular velocity and torque

Future Work

To eliminate the possibility that experiment 1 did not work due to the low amount of torque, the experiment can be repeated with a smaller motor. Recommended is a shaft of 1 mm diameter, instead of ball bearing races, wrap the ends of the shaft with teflon and have steel disks with a smaller diameter. The simulations can also be repeated using other programs for validity. Recommended programs are COMSOL and Altair. Recent papers disagree about the importance of the material used as rails when testing the Huber effect. Experimenting with different materials could provide insight on electrical and magnetic field behaviours corresponding to the Huber effect and could help explain why experiment 1 did not work.

Conclusion

In conclusion, the experiments conducted in this project supports Gruenberg's electromagnetic theory as the modified motor did not work and the relationship between torque and angular velocity was found to be a squared one. The result of the simulations and from testing the modified motor have given reasons to not neglect the electromechanical theory suggested by Polivanov instead. However, it is possible that the modified motor did not work due to the size of the motor. Future testing of the modified motor should take this into consideration.

Project YouTube videos

• Ball Bearing Motor Mystery
  • https://www.youtube.com/watch?v=P-OtHLq0KVU&t=1s
• Reactions of metals with Gallium
  • https://www.youtube.com/watch?v=nglK-_yMVz4&t=1s
• Investigation of the Ball Bearing Motor
  • https://www.youtube.com/watch?v=BsAD6xZrKqM&t=1s

References and useful resources

[1] H. Gruenberg et al, "The ball bearing as a motor," Am. J. Physics, 46, pp. 1213-1219, 1973.

[2] S. Marinov, "The intriguing ball-bearing motor," Electronics & Wireless World, April 1989

[3] K. M. Polivanov, A. V. Netushil and N. V. Tatarinova, "Electromechanical effect of Huber," Elektrichestvo, No. 8, pp. 72-6, 1973.

[4] Integza (2019). DIY Torquemeter - How to measure torque! [Arduino & 3D Printed]. [video] Available at: https://www.youtube.com/watch?v=NIpspXaPVcs [Accessed 1 Jul. 2019].