“There is no authority who decides what is a good idea.” ― Richard Feynman
This page showcases some of the developmental activities that were carried out by me, mostly out of interest or curiosity. Some of them assisted my research work, while others led to deeper understanding.
Many of these activities can be replicated by an individual and could be of great assistance in teaching the concepts related to the electromagnetics.
Development of MIEDW Prototype
Motor integrated electrodynamic wheel (MIEDW) is an energy conversion device having two airgaps participating in electromagnetic interactions. It consists of an inner stator, a rotor constituting an array of permanent magnets (PMs), and a conducting plate. When the windings of the stator are suitably excited, torque is developed on the rotor, causing it to rotate. This rotation results in interaction of time-varying magnetic field with the plate in the outer region. Eddy currents are induced in the plate which interact with PM array magnetic field to develop electrodynamic forces. These forces can be utilized for electrodynamic suspension.
A 12-slot stator of an existing commercial motor is used. This stator had delta connected windings and also the slot fill-factor was poor (29%). For direct measurement of phase emf, star connected windings are advantageous. To address this, the stator is rewound with 16-turns of SWG 26 wire on each tooth, achieving a slot fill-factor of 42%. This rewound stator has three-phase double layer winding. The photographs of the original and rewound stator are shown in Fig. 1 (a) and (b), respectively.
Fig. 2 shows different versions of the MIEDW rotor, developed by me. Fig. 2(a) shows the first version of rotor, where PM blocks are glued on a non-magnetic (wood) rotor support to form an 8-pole EDW. This rotor is connected to the shaft of a DC-motor for further application. It was during this experimental activity, the idea of integrated motor struck to the scholar.
A straightforward approach for the integration would be to attach the EDW PMs on the outer periphery of an outer-rotor motor. To realize this, a commercial outer-rotor motor is used, which has 12-slot stator. Eight PM blocks are glued on the outer periphery of this rotor, to realize an 8-pole EDW, as shown in Fig. 2(b). Nylon threads are used to hold the PMs against centrifugal forces. It may be noted that the form-factor of this EDW–motor assembly would be smaller than that of a shaft-mounted arrangement. To improve the performance of the rotor shown in Fig. 2(b), additional magnets are introduced in the space between the existing magnets, as shown in Fig. 2(c). It can be noticed that the EDW PM array resembles a 2-segment Halbach array, with higher flux density in the outer region. During this stage of the experimental activity, the idea of ‘coupled’ MIEDW emerged.
To realize a coupled MIEDW, one such rotor is fabricated using 3D-printing. This rotor is shown in Fig. 2(d). The requirement of a non-magnetic rotor structure is fulfilled by the 3D-printed material, which is a PLA based polymer. The process of buiding such rotors is involved. The main reason being the calibration requirements of the 3D-printer. The correct calibration is obtained after three–four iterations.
It may be noted that the shunt PMs in the rotor, shown in Fig. 2(d), are wider than the required. In final version of the rotor, shunt PM blocks are machined as per the requirement. Figs. 2(e) and (f) show the final version of the rotor. It can be seen that the slot for the shunt PMs are open before their assembly (Fig. 2(f)). This resulted in fabrication difficulty in fixing shunt PMs in place. This is due to the repulsion forces exerted by the radial PMs. As a solution to this problem, t-shaped slot-liners are inserted to close these slots and then the shunt PMs are placed. This point should be kept in mind while fabricating higher power-rating MIEDWs.
Machining of PM Pole Pieces
A setup consisting of a mechanical vice to hold the PM pieces and a rotary cutter is developed by the me, as shown in Figs. 3(a) and (b). Initially, attempts are made to cut the magnetized PM pieces. Another motivation for doing so is to observe the loss of magnetic strength after machining. During machining, the PM block is cooled using soap-oil-water mixture to extract the local heat. It is found that the PM remanence remained almost the same. Fig. 3(c) shows the machining procedure, while Fig. 3(d) shows failed PM pieces after the machining. The yield of this machining was about 30%. Though, there were no difficulty in using the ferromagnetic cutter disk, the PM blocks were demagnetized before machining.
Development of a Setup for Characterization of MIEDW
For characterization of the levitation performance of the MIEDW, lift and thrust forces exerted on the plate are required. This requires two load-cells. Further, in order to measure the thrust, the plate must be free to move in the direction of thrust, while the levitation gap should be constant for measurements at different speeds. In addition, there should be a provision to vary this levitation gap. Following subsections provide details on the development of a test-bed, satisfying these requirements.
Initial Test-bed
Fig. 1(a) shows the test-bed developed to verify the initial models. The EDW is rotated using a DC motor, controlled by a DC-DC converter. The plate is connected to the lift load-cell using a linear guide. Fig. 1(b) shows the arrangement where MIEDW is used. The integrated motor is driven by a DC-AC converter. It may be noted that the form-factor of the MIEDW assembly is reduced drastically as compared to that shown in Fig. 1(a). There were two issues with this test-bed: first, the plate tend to tilt and hit the rotor, even for small asymmetric placement of the EDW below the plate, and second, the adjustment of the levitation airgap was difficult.
Improved Test-bed
Fig. 2(a) shows the improved test-bed, where two linear guides are used, and threaded screws are used to adjust the levitation gap. This setup is found to be very susceptible to vibrations, corrupting the force measurements. Therefore, the attachment is modified using springs and spacers on the either sides of screws, as shown in Fig. 2(b). The effect of vibrations was marginally suppressed, while the precise adjustment of the levitation gap was tedious and not reproducible. Finally, the plate and linear-guide assembly are rigidly fixed to the lift load-cell. Whereas, to vary the levitation gap, the entire MIEDW setup is lifted up and down using spacers under the motor support. This final arrangement is shown in Fig. 2(c).
Efforts to Reduce Vibrations
During initial testing, it was noticed that at speeds > 10.000 rpm, the test-bed vibrations caused the plate to jitter with the thrust load-cell. To reduce these effects of vibration, several efforts are made. In an initial effort, the test-bed space is divided into two regions, RI and RM (see Fig. 3(a)). In the first trial, only small connections are retained between these two regions, as shown in Fig. 3(a). This arrangement could not reduce the vibrations in the setup.
In the second trial, the connections between the regions RI and RM are removed, and, these regions are aligned using acrylic-glass with cushion in-between them. This arrangement did not give satisfactory results either. Another variation, as shown in Fig. 3(c), is tried, wherein an additional wooden base is provided on the other side of the acrylic glass. This arrangement, on contrary, increased the vibrations in the test-bed.
Finally, an attempt is made to eliminate the source of the vibration! One of the sources could be the unbalanced mass distribution in the rotor. A simple arrangement is made to balance the MIEDW rotor, as shown in Fig. 3(d). Some mass of the rotor is removed to reduce the mass on the heavier region, while paper tapes are added on the lighter region to increase the mass. Upon several trials, the rotor was manually balanced. The vibrations and its effects were substantially reduced after rotor balancing.
Proof-of-Concept of Axial-Flux Motor for CPU Cooler
An attempt is made to develop a toy axial-flux motor and understand if it's drive requirements are different from the radial-flux ones. Fundamentally, as long as the terminal electrical parameters of the axial-flux motors are comparable to that of the radial-flux motor (for which the drive is available), drive requirements would not change substantially.
In this work a small motor was developed from scratch, scavenging the electrical steel from an old inductor and manual handwork. The photograph of the stator assembly is shown in Fig. 1(a). There are six teeth in the stator, and alternate tooth are wound with 25 turns of copper wire (SWG-26). Such a winding layout is called single layer winding.
The structure of the rotor was realized using 3D-printed disc with slots for magnet blocks. For this rotor NdFeB magnets are used. The rotor configuration is of 4-pole rotor. The photograph of the rotor is shown in Fig. 1(b).
Fig. 2(a) shows the frame of an existing CPU-fan motor in which the developed stator is retrofitted. Similarly, the existing magnet-ring of the fan was removed and the developed rotor was attached inside the fan-hub. The photograph is shown in Fig. 2(b). The phot graph of the assembled prototype is shown in Fig. 3 along with the off-the-shelf drive (electronic speed controller - ESC) used for testing the motor. Such ESC's are widely used for driving radial-flux motors.
The per-phase inductance of the assembled motor was obtained in the range of ~30 mH. The back-emf constant was ~ 60 mV/krpm (very low!). To obtain this parameter, the motor was driven at no-load (with fan removed from the rotor) using the ESC drive-voltage of 12 V. The motor could approach 18,000 RPM at no-load.
To test the loaded condition, the fan was attached to the rotor. The motor could operate at 5000 RPM, however, there was significant heating of the stator in few tens of seconds. This is because of poor heat-dissipation since the stator was isolated from the frame using a poor thermal conductor. Ironically, the CPU Fans are meant for better thermal management of CPUs!
It could be verified that the existing ESC could satisfactorily drive the axial-flux motor using the existing control algorithms (sensor-less back-EMF zero-crossing detection based in this case).
Estimation of PM Strength using Indirect Approach
Remanence of PM (Br) is an important parameter required for accurate modelling in FEA environment. In the absence of a commercial tesla-meter, the following method was used to estimate the Br of the magnets.
Using Semi-Experimental Approach
The Br of the PM's is extracted using a combination of experimental measurements and 3D-FEA simulations. First, the force of attraction between the PM blocks and a ferromagnetic sheet is measured using the setup shown in Fig. 1. Schematic of this arrangement is shown in Fig. 1(a). Since the PM magnetic fields are likely to interfere with the weighing scale, a lever type arrangement is used for measuring the force. Figs. 1(b) and (c) show the photograph of the actual setup during the measurement of the weight of PM block and its force of attraction with the sheet, respectively. The same arrangement is also simulated in 3D-FEA tool, to predict the force of attraction. This model is shown in Fig. 2(a). The Br of the magnet is varied in this 3D-model, and its relationship with the attraction force is obtained. This is shown in Fig. 2(b). Using this relation and the force measurements from the experiment, Br is estimated. For a class of PM blocks, average value of Br is estimated to be 0.94 T. This is in good agreement with the estimated value of 0.96 T, which is obtained using an in-house developed tesla-meter.
Development of Impulse Magnetizer for Small Permanent Magnet Blocks
This section describes the efforts toward the development of a magnetizer setup intended for small PM blocks. This setup also enables partial magnetization of the PMs and study of their characteristic such as permeability as a function of magnetization state
To magnetize a typical NdFeB PM block to its saturation flux density, magnetizing field of about 2-3 times of its intrinsic coercivity (Hc) has to be applied. For example, N35 grade of PMs have Hc of about 800 kA/m, and therefore, to magnetize it 1600–2400 kA/m of magnetizing field is required. Further, the magnetizing field should be as uniform as possible in the PM region, and should be present for sufficiently large (1-2 ms) duration. This is to ensure the penetration of magnetizing fields inside the PM material against the induced eddy currents [104]. These requirements can be fulfilled by an impulse magnetizer setup which mainly consists of core-less coil (called magnetizing coil) and an arrangement to discharge a charged capacitor into it.
Fig. 1 shows the schematic of such a setup. The photograph of magnetizing coil and the entire setup is shown in Fig. 2. The ‘to be magnetized’ PM blocks are placed at the centre of the coil (Fig. 2(b)). To estimate the peak flux density inside the magnetizer coil, the peak current through the coil is required. Since the current sensor probes available in the laboratory were capable of measuring up to 70 A, the current through the coil is predicted as follows: The entire setup is modelled in a circuit simulator using the measured values of the magnetizer circuit parameters. In order to verify the simulation model, the impulse current through the coil was measured for the capacitor charge voltage of 50 V, and then compares with simulated results. A very good correlation was observed. Using this simulation model, the impulse current due to the rated maximum voltage of 350 V is obtained. The results are shown in Fig. 3. Peak current of about 400 A is estimated through the coil. Corresponding to this current value, the magnetic field plot is shown in Fig. 3(b). The peak magnetizing field of 1740 kA/m could be observed using FEA, meeting the initial requirements.
A Study on Thermal Demagnetization of NdFeB PM Blocks
There was a need to demagnetize the PMs before machining. In this work, they were demagnetized using heat. Another objective of this work was to observe the loss in magnetic strength of the PM block with temperature. The arrangement has been shown in Fig. 1.
Fig. 2 shows the variation of PM block temperature and its estimated field strength over time. It can be seen that the PM loses all its strength at around 220◦C. It is interesting to see the variation of the PM field strength with respect to the temperature. It is shown in Fig. 3. Once the heat supply to the PM block is removed, its temperature reduces, and the PM block regains some of its magnetism. When the block is heated again, the temperature starts increasing, but the permanent demagnetization occurs only after the temperature to which it was previously subjected to. This is interesting because the demagnetization of PM block can be seen as a function of the temperature alone and not the heat generated/supplied in/to it. Therefore, the PMs used in electric machines can be subjected to the expected ambient temperature. This would harden them against such high temperatures during actual operation, without the loss of their predicted performance.
Method for Estimation of Permeability of NdFeB PM Block
This work provides details of the estimation procedure. Furthermore, since it is reported in the literature that the relative permeability varies with magnetization state and can be as high as 20 in demagnetized state, its estimation for different magnetization state is also included in this work.
Inductance Measurement
Fig. 1 shows a coil wound on a rectangular former. Three PM pole pieces are merged to form a PM block. This block can be placed inside the rectangular coil former in two orientations: in the first orientation, the preferred magnetization direction is parallel to the axis of the coil; and in the second, the preferred magnetization direction is perpendicular to the axis of the coil.
The PM block with different magnetized state is placed inside the coil and the inductance of coil is measured. This is done for both the orientation of the PM block. As the PM is magnetized to its saturated state, it starts behaving like air, and the inductance of the coil approaches to that of an air cored coil. Fig. 2 shows the variation of the coil inductance for various values of magnetization. It can be confirmed that the inductance of the coil with PM block in the preferred direction changes significantly. However, there is negligible variation in the non-preferred direction, indicating non-isotropic nature of the PM block.
Estimation of Relative Permeability
To extract the value of µr using the inductance variation, the following numerical approach is taken: The rectangular coil is modelled in 3D-FEA environment, and a ferromagnetic block is placed inside it. The dimensions of this ferromagnetic block are same as that of the physical PM block used for measurements. The permeability of this modelled block is varied in the FEA, and the inductance of the coil is obtained. The variation of the inductance for different values of relative permeability is shown in Fig. 3. It can be seen that the value of inductance for µr=1 is slightly higher than the measured value of the air cored coil. This small variation can be attributed to the coil curvature which is not modelled in FEA. To account for this variation, all the 3D-FEA results are scaled, such that the 3D-FEA inductance for air cored coil coincides with the experimental value. These values are plotted in Fig. 3.
The estimated permeability of the PM block is plotted in Fig. F.4. The following conclusions can be drawn from this study. The permeability of the PM block is not isotropic. Its value is close to that of air only when it is magnetized to its saturated state.
Design and Development of a Portable Conductivity Measurement Instrument
The accurate value of electrical conductivity of a material is crucial for modelling. This work presents a technique for the estimation of the same. For such estimation, four-point Kelvin connection based arrangement is used, as implemented using setup shown in Fig. 1 (a). Due to very high conductivity of the plate, the voltage drop across it is of the order of few millivolt. To amplify this, an instrumentation amplifier (INA128) is used. The gain of this amplifier can be adjusted using a single resistance, RG, making it suitable for this purpose.
Using this technique, the conductivity of an aluminium plate is estimated to be 2.60×107 S/m. It may be noted that the conductivity of pure aluminium is 3.63×107 S/m at 30 degree C. Therefore, significant error would have been introduced in the simulation result if the plate is modelled using the conductivity that of pure aluminium.
The above conductivity measurement technique is extended to develop a battery-operated instrument. Fig. 1(b) shows the photograph of this instrument. The detailed labelled photograph of this instrument is shown in Fig. 2. It can be seen that it employs good number of power electronic modules for different functionalities. For instance, a buck-converter, two boost-converters, and a Li-ion battery charging/discharging module. Therefore, development of this instrument can serve as a project-based teaching aid for undergraduates.
The accuracy of the developed instrument is tested for conductivity measurement of two thin-strips. The estimated conductivity of two representative strips, S1 and S2, using the resistance values obtained using the developed instrument are 43.17 MS/m and 2.32 MS/m. Conductivity values obtained using resistance values measured from HM8118 LCR Bridge (ROHDE & SCHWARZ) are 42.73 MS/m and 2.27 MS/m, respectively. A good agreement can be observed between the developed instrument and the commercial product, demonstrating its viability for conductivity estimation.
Methodology for Experimental Visualization of Eddy Current Paths in a Conductive Plate
A publication is submitted for review.