Building an electric motor for Candy Cane, part 1: Background

2019.07.20

So a while back Candy Cane’s Atomic 4 started belching oil-burning smoke, and eventually failed to start. Given the timeline of that failure, it’s a pretty safe bet that the piston rings and the cylinder wall were no longer properly isolating the cylinder from the crankcase, and that the scope of work involved in fixing it would be close enough to a rebuild (also there were enough leaks, loose-fitting couplings, etc.) that it may as well get rebuilt.

I figured that for the cost to order all the parts I’d need for a rebuild, I could probably just get a Moyer rebuild and save myself a lot of time, but that was pretty expensive, and I’d had a hankering to repower to electric for over a decade. As luck would have it, I was soon to learn that I could probably get an Electric Yacht motor for about the cost of a Moyer rebuild — but I could also possibly put something together myself for a few thousand less (assuming that my time is worth nothing — more more accurately, that I’d enjoy the work enough for its value not to be an issue). On top of all this, I’d had an opportunity to take a look at an Electric Yacht motor, and found it to be a remarkably simple piece of machinery: just an electric motor driving a shaft via a belt.

So I consulted with a friend who has a bit of a hobby restoring shop machinery, and a couple of friends with professional experience with machining and the design and maintenance of industrial machinery, and came up with an initial design:

A 5KW 3 phase electric motor driven by a 48V solid state controller (Golden Motor Canada has such a motor and driver for a combined price of around $1500) , mounted to the aft face of a piece of rectangular steel tube (which would be mounted on the aft engine mounts by way of angle iron support flanges. It would connect to the prop shaft via a 7/8″ transfer shaft, and this shaft would be connected to the motor by a pair of V-belts with pulleys sized to provide a 5:2 reduction (simply a 5″ pulley and a 2″ pulley).

To hold the transfer shaft in the box-steel frame, my shop-tool-fixing friend recommended the use of a self-aligning pillow bearing, and I agreed (well except that I decided a flange mount made more sense than a pillow mount), since we overestimated how difficult it would be to align such a light motor (compared to an A4) to the prop shaft, and with a V-belt, keeping the motor shaft and transfer shaft parallel wasn’t quite as important as it might be with, say, a timing belt.

Needless to say, this motor ended up being something of a prototype and a learning experience (so I don’t feel as bad about not having any pictures of it).

The first major problem we had was when the lagoon where I normally moor my boat started to fill in with weeds as the summer progressed. It might seem like an advantage that a V-belt can slip if the prop gets fouled on, say, a mooring line, and maybe keep the prop shaft from getting bent. It turns out to be a disadvantage however, when the belt can also slip instead of transferring the necessary torque to cut through thick enough weeds fouling the prop.

The solution to this seemed simple enough: ditch the V-belts and their pulleys, and switch to a timing belt and get pulleys for it. One hitch was that we’d machined the frame to have a fixed distance between the motor shaft and transfer shaft (mostly because I did a really poor job of measuring the size of the bolt circle on the motor, and we ended up cutting the mounting holes as slots in more of an X configuration (that could take an assortment of 4-hole bolt circle sizes) than an H configuration (that would only fit one size of bolt circle, but allow the motor to be slid up and down while the bolts were loose). Naturally, this meant that there was no pulley and belt combination which both approximated the desired 5:2 reduction and, with the shaft spacing fabricated into the frame, would work with a belt size that divided evenly by any common pitch between the teeth (you can stretch a V-belt because it doesn’t have teeth and you don’t have to worry about them lining up properly with the pulleys — a timing belt OTOH…).

The solution for that snag also seemed reasonably straighforward: add an idler to take up slack in the belt. Of course we didn’t really leave any room inside the frame for a proper idler. Consequently, I just bought a stack of bearings that I could slide onto a bolt running through the frame as a sort of idler axle, and a (much smaller) stack of washers that I could put on either side of the bearings to keep them from slding fore and aft along their axle. Needless to say, this didn’t really work as well as planned, and once I brought the motor above about 20% power, the tension on the idler-hack knocked it out of parallel with the pulleys. Since it didn’t have any crown shape to it, getting unaligned like this caused the belt to want to drift from the higher tension side towards the lower tension side, until it had drifted so far in that direction that it tried to feed onto the flange of one of the pulleys instead of the teeth, and was subsequently ripped apart.

Another issue arose when replacing the belt: in order to keep the shaft alignment at all sane under belt tension, there are bearings to support it on both the front and back of the frame (after all, the fact that they’re self-aligning bearings provides absolutely no guarantee that they’ll self-align the way we’d ideally like them to, instead of, say, taking the alignment that minimizes belt tension. Of course this means that to install or remove a belt, the shaft has to be at least partially removed from the frame, and for this the pulley needs to be able to slide along it.

When we were using V-belt pulleys, the 2″ pulley on the motor was connected to the shaft by a set screw (and the 3/16″ key to transfer torque). The lower (5″) pulley was connected by a split taper bushing, but it was a small one whose tightening and loosening screws could be accessed through a hole that fit neatly under the forward bearing. With timing belt pulleys needing to remain secure much more reliably, the transfer pulley took a much larger bushing, and the motor pulley also took a bushing (in fact I was able to reuse the bushing from the 5″ V-belt pulley on the 2-ish” timing belt pulley). This meant a larger hole to access the bushing screws for the lower pulley and an entirely new hole to access the screws for the upper pulley. Both of these were hurriedly carved out in the middle of the night with a die grinder, in the hopes that I’d get the motor repaired well enough to make it to a race the next day.

By the time I got everthing sorted well enough to have at least some nominal auxilliary propulsion, it was apparenty that the accumulation of hacks to keep the existing design working was getting too large, and it was time to build a new frame to correct some of the design flaws (though the design flaw of continuing to use a self-aligning bearing was kept firmly in place).

Another glitch in the initial setup (but not really a design flaw) that caused me to have to begin the season with a tow, instead of the glorious maiden voyage of the new motor, was that I figured out that I could attach the existing cockpit control for the A4’s throttle to a linear potentiometer (i.e. a slider). What I didn’t know was that the controller supported two input modes for the throttle wires: Hall effect sensor, and potentiometer — and that the factory default was Hall effect. It was easy enough to change, though I had to buy a programming cable for the controller (but it came with the programming software, and the interface for the settings was reasonably intuitive).

All of these things having been figured out, a new off-season was upon me, with seemingly plenty of time fabricate a new frame, and rebuild the motor with the design flaws mostly removed.

Leave a comment