BuildIts in Progress: Battery Building, Motor Woes, and a Rideable Tricycle

Here is an unfortunately large update, due to my blogging laziness.

I spent way too much time inhaling soldering fumes while building two battery packs for the trike. As you may notice the cells I ended up using were the green flavored variety (given to me by Dane), rather than the cardboard covered ones I was planning on using. The green cells are 2.5 Ah, as opposed to the 2.2 of the others, and have 2/3 the internal resistance.

I started the packs by making Tetris-like blocks tacked together with hot glue, with cells soldered in parallel.

These blocks were then tacked to each other, and wired in series. Below is the lower half of one of the packs. The gap separating the three leftmost cells from the rest of the pack is for the drive shaft.

Each 12S3P pack is actually made from 2 6S3P packs connected in series. I did this because no hobby chargers (that I know of) can balance charge a 12S pack, and the charger I have been using can only charge up to 6S packs. Each half of the pack was packaged in "bottle armor" made from giant 3 liter soda bottles. When heated, soda bottles shrink like heatshrink tubing, and form a hard plastic case when they cool. The weird 3-cell extension off the pack could not be bottle-covered, so I used a combination of large heatshrink tubing and kapton tape.

For the power leads, I used XT60 connectors - one entire connector each for positive and negative.

To hide the pack, I made a cover out of matte black acrylic. In the process, I found a really good way of making clean bends in acrylic. I clamped thick aluminum blocks .25" above and below the bend line, and heated the gap with a heat gun. The blocks both stop the acrylic outside of the bend area from heating up and prevent anything but the bend line from deforming.

The result:

To stop the battery from banging around, I coated the inside of the battery compartment with soft foam tape.

To make the seat actually able to support a person's weight, I coated over the plastic frame with fiberglass. First, I covered the frame in electrical tape. This made a surface for the cloth to rest on, and epoxy does not stick to electrical tape.

Due to the particular epoxies and curing methods I used, the fiberglass turned an unpleasant orange-brown color. I started out using West Systems 105/205 epoxy, which is what I used on my bamboo bike. Some rust must have gotten into the hardener container, so the epoxy was bright red. To cure the epoxy faster, I decided to put it into the reflow oven at MITERS, and bake it at 200 F. Long story short, don't oven cure with 205 (fast) hardener! It foamed up and ruined the nice smooth surface. I continued with a much slower epoxy, which gave much better toaster-oven curing results.

I glued the fiberglass layer to the plastic frame with thickened epoxy. At the same time, I glued the mounting bolts in place.

After lots of sanding and gap filling with even more epoxy, the seat had a fairly smooth surface, which I spray painted black (I don't have any pictures of that yet).

To attach the external hall effect sensors that let the motor controller determine the position of the rotor, I used a Hall Effect Sensor Board and Sensor Adapter.

The motor controller, fuse and kill switch were all fastened above the motor, to the underside of the top plate:

With everything wired up, I began the process of finding the correct motor phase combinations, and aligning the hall sensors for minimum current draw, using this guide. This process started out remarkably smoothly. The motor spun the correct direction on the first try. I hooked up the controller to a bench power supply with a current readout, and moved the sensor board to the lowest current position. Things started getting weird when I reverted to battery power. For no apparent reason, the motor spun the opposite direction when I powered it on. I had written down the motor phase combination I used, so I know I didn't reassemble it incorrectly. I cycled phases until is spun the correct way again, and minimized current for a second time.

After this, the trike was actually rideable, and I brought it up to the IDC test track. Everything was working great until an overzealous sliding turn caused my knee to plant itself in the temporary throttle I found in a box of potentiometers. The potentiometer unfortunately did not survive.

I made my own thumb trottle out of a chunk of aluminum and an old game pad joystick, but could not get the trike to run smoothly with it. I thought it had to do with the throttle being too sensitive and causing the controller's over current protection to be tripped. After reprogramming the controller's throttle response and current limit to no avail, I replaced my throttle with a half twist hall effect throttle from an electric scooter, but the problem persisted. Sometimes the motor would run smoothly, but most of the time it drew way too much current, would not reach full speed, and produced so little torque it could be stalled by hand.

Not sure what to do, I enlisted the help of Shane to diagnose the problem. After spending hours adjusting phase combinations, sensor timing, testing the hall sensor outputs, trying another throttle, replacing the throttle cable, redoing the hall sensor wiring, trying a completely new controller, scoping the motor phases while running, and probably some other things I've forgotten, we were able to get it working smoothly again by swapping two of the hall sensor leads and going through the sensor timing process again.

However, when I went back and installed a the kill switch, the motor had reverted to its high-current, low-torque, generally-bad state. After even more phase and hall swapping, suspicion towards the motor itself was raised - and confirmed. The short between the windings and the case that I fixed ages ago had reappeared. It seems that the motor had been intermittently shorting, causing all the motor problems I had. Fortunately, Shane happened to have a brand new identical motor which he gave me. At some point I will try to revive the old motor again.

To prepare the motor, I chopped off the threaded side of the shaft, and milled a flat and a dimple into the drive side of the shaft, for the drive sprocket's set screws to grip into. It has worked beautifully since then, and seems to perform better than the old motor ever did, with both somewhat higher torque and lower no-load current.