Linear Motor Test Bench

Linear motor post #1. Other linear motor posts

Introduction

Inspired by this YouTube video, I built a prototype linear motor. I based my version on this masters thesis, which I found in the comments of that video. I built a coreless tubular linear motor because it consists of only a tube of opposing magnets and spacers, and a series of coils around the tube; it only uses common cylindrical magnets and no electrical steel laminations are needed. Building a coreless motor also means that the motor will not have any cogging torque, so accurate servo motion is an easier control problem.

Making the Magnet Tube

I used a 10mm ID carbon fiber tube to structurally hold the magnets, which were 10mm diameter, 20mm long cylindrical N52 neodymium magnets. The magnets fit pretty tightly inside the tube, and I inserted them while alternating the orientations and with a 10mm spacer in between to create the motor poles. The spacers were just small 3D printed cylinders with 100% infill. Externally, it just looks like a CF tube. The internal organization is just an extension of the arrangement below.

The a spacer between two magnets. In the picture the magnets are facing the same direction, but inside the tube they are opposing each other.

Making the Coils

Making the coils was the most tedious part of this project. In order to get compact and neat coils, I 3D printed some spools to wind the wire around. I used 20AWG enamelled copper wire, and tried to wind each layer around the spool as neatly and evenly as possible.

I started each coil by wrapping one layer and supergluing it in place. At first I did the same for every layer, but that became too tedious, so eventually I just did it for the first and last layers.

The spools I made were flimsy because they were really thin, so I had to wind the coils by hand. I found that slight disturbances in the wire quickly lead to the layers becoming less and less flat, and therefore less dense. One thing that helped me get consistent layers of windings was to superglue each layer once I had arranged it neatly enough.

A single completed coil. The 3D printed frame slightly bent outward because of the tension on the wires as I wrapped them.

I eventually finished all 6 coils, which I assembled into a single cylindrical block to make a single motor. In order to be consistent as possible from coil to coil, I pre-cut the same length of magnet wire for all the coils and wound that onto a larger spool temporarily while smoothing out any bends or kinks. The dimensions of my coils were roughly 17mm ID, 42mm OD, and 9mm wide (with a 1mm spacer between each coil). The spools were 3d printed in segments and "welded" together with a soldering iron.

I got less and less careful with the coils as time went on, so some of them ended up messier than others.

After building this, I realized that the design was inefficient from a force-to-weight ratio perspective. All other parameters being constant, the force produced is roughly proportional to the number of turns times the current, so the larger outer turns of the wider coils have a lower force-to-weight ratio than the inner ones. The 3D printed spools were also a bit flimsy, and flexed outwards when I wound a coil and messed up the proportions of the coils. I ended up using this first attempt coil as a test platform for controlling linear motors and as a way of practicing construction methods.

Test Bench Assembly

Next, I built a test bench that I could use to test the electronic setup with. It consisted of a 48V DC power supply mounted, a carriage on a linear rail with the coils, mounting brackets for the magnet tube, and the control electronics, which I will cover in the next post.

All the parts of the test bench, but not in a fully assembled state.

Based on the assembly, I did some basic tests of the electrical properties of the motor. I measured that the resistance from one lead to another (passing through the coils of 2 phases) was about 2 Ohms. I also hooked up the motor leads to my oscilloscope and referenced them from the star terminal (the point where one lead from each coil were connected together) and moved the carriage back and forth. The resulting voltages should reflect the back-emf waveform of the motor, and I was pleased to see that it was very close to the ideal triple sinusoid:

Back-emf waveforms from a smooth motion from one side of the travel to the other. The amplitude is not perfectly consistent because the I made the move by hand, which resulted in inconsisted speed.