Remember that big motor from the Ford Fusion Hybrid? Yeah, that one that I build a housing for. Well it turns out that there’s a second, smaller one. This is a rather long post so I’m splitting it into two parts. Part one tells where these motors came from and why I’m interested in playing with them, and part two shows construction of a housing for the smaller MG1.
PART 1: Some Background on These Here Moters
I recently found this video (published by Ford)which shows a partial teardown of a New York City taxi (Ford Escape) with what I believe is a similar or identical hybrid drivetrain.
The big one is MG2 (Motor/Generator #2) in the Ford Fusion Hybrid transaxle. The small(er) one is MG1. If I had pulled the motors myself, I would have known that they were packaged together and coupled by a planetary gearset. I still can’t concisely explain why they use two separate motor generators, or why they’re coupled through planetary gears, but there must be a good reason for it since it seems to be the standard in hybrid vehicles across different manufacturers.
So how did I get my dirty paws on them? MG1 (the smaller one) has led a nomadic life. It started in a Mercury Milan (along with MG2) which was donated to MIT’s Electric Vehicle Team (EVT) sometime around 2010. They pulled the transaxle and replaced it with an induction motor from a bus. Their project page for the vehicle is here. Naturally, the first thing to do when converting a hybrid car to full EV is to remove the electric motors from it. It’s ironic on the surface, but dealing with these motors firsthand has helped me to understand why they may have made that decision. They’re very much sealed-in and designed to work alongside an internal combustion engine. When building a full-on EV, it almost certainly makes more sense to use one Big Ole electric motor (if it’s available) than to use the hacked up electric half of a hybrid drivetrain. Anyway, some MITERS folks grabbed MG1 and MG2 from an EVT cleanout sometime over the summer of 2013. The motors drifted around the floor of MITERS, and MG1 somehow wandered across the river to the electric vehicle team at Boston University. It wasn’t until mid fall of 2014 when I hacked up a housing for the big MG2 that it spun under its own power for the first time in 3+ years. After that one spun up, Bayley asked James over at BU to return MG1 since their EVT had also decided not to use it.
Finding Information on Hybrid Cars
There is a spectrum of information available on hybrid cars. On the one end there are technical papers and patents. On the other end, there are service manuals and repair resources for hybrid cars showing what goes wrong and how to fix it. Sadly a hard copy of a Ford Escape or Mercury Milan manual goes for $100 or more on ebay. That said, there are a number of repair videos around. Luscious Garage in San Francisco has this video showing the replacement of a burned-up Prius MG2 stator. In addition, there are a few teardowns of electric drivetrains on the net (like this one done by a couple of folks in a garage). I haven’t yet found much organized engineering-side information on these parts. While the papers and patents contain some valuable engineering knowledge, they tend to be written (understandably) more for the purpose of announcing or claiming than educating or sharing information. A notable exception is the work done at Oak Ridge National Labs. A search for “ORNL Prius” will turn up a bunch of excellent papers with incredibly detailed information on the capabilities of available hybrid car motors and inverters.
What is the point of this?
There are two main components to this project’s motivation cloud. 1 – To try to populate that gap in the knowledge spectrum mentioned earlier. More specifically – finding and compiling information on these motors. The more salient part of the motivation is to use the dang thing, leading to 2 – to play with big motors and motor controllers and put them on silly vehicles. With that out of the way, let’s make it go.
PART 2: Making the housing
This is a chance to repeat the process I went through for the bigger motor, using what I learned then to do it better this time around.
first up, layout. I made models of the stator and rotor just like last time.
I carried some papers around with me and took a couple days to think through the machining process and how I could make it better than Last Time. I considered approaching one of the local aluminum casting houses, as I heard that they’ll make you a few big motor castings for $150 or so (if you bring your own pattern). I ruled it out because it didn’t eliminate most of the tedious machining operations like boring the bearing housings and cutting the contour for the stator OD. It made sense to continue using scraps of aluminum because it costs less.
Major changes since the last rev:
-WELDING. The stator standoffs and tall verticals are welded to the front plate. Both bearing housings were welded to their respective endplates. The cap screws shown in the exploded view were used for rough positioning and fixturing while welding.
-stator is mounted to the front plate, fixing the nagging mechE bug of Last Time.
-tall verticals have the 180mm minor diameter of the stator milled into them. It is their job to repeatably locate the stator in the horizontal plane.
-tall verticals have spring pins for aligning the back plate. No more using screws to drag the rotor away from the stator.
-less clearance on the top and bottom of the stator windings. Might as well make it compact. That was a secondary consideration on MG2.
-bigger, sealed bearings. This is so that the motor 1. can be operated in dusty open air, and 2. can support radial loads from a belt pulley or something similar
Based on the output spline and the few pictures I’ve seen, I don’t think the motor is designed to experience much radial loading in its Natural Habitat. Cantilevering a sprocket or pulley off of the motor shaft justifies using bigger bearings, especially on the side nearest to the sprocket/pulley. In addition, the motor and its bearings used to live in a bath of filtered oil. They had no seals, and had a terrifically low rolling resistance. Because my design is open to the air, I purchased some sealed bearings. A pair cost $17 from The Big Ole Bearing Store.
I specified the ID according to the rotor shaft sizes (30 mm front and 22 mm rear) and increased the width and OD slightly to get a higher load rating. I picked the 62/22-2RS 22x50x14 @ $6.02 for the rear and the SS6006-2RS 30X55X13 @ $9.67 for the front. Both are doubly sealed bearings meant for ATVs.
I tried freezing the rotor and heating the original bearings to get them off.
I didn’t have a puller that fit, and my weak attempts at removal weren’t going anywhere, so I carefully sliced the old bearings off with a small cutoff wheel. I covered the rotor in painters tape to minimize the sticking of magnetic dust. It was a cool visualization of the 8 pole (4 pole pair) rotor. The dust pattern didn’t quite make sense given that I think this is an IPM (internal permanent magnet) rotor. The only way to be sure is to take apart the rotor. Just like the rotor of the larger MG2, it’s a stack of laminations compressed by two cast aluminum endplates and a big nut. I don’t plan on taking this one apart, since removing that nut would require pulling one of the bearings, grinding off the threads that they smash to stop the nut from backing off, then taking it all apart and successfully putting it back together. I’d do it if I had a spare rotor lying around, but I’d rather not risk messing this one up.
When I got close to the shaft with the abrasive wheel, I tapped the brittle inner race open with a cold chisel. It slides off easily after it is split. I would have saved the bearing races for making cutting tools, but I haven’t had much luck heat treating them to stay hard. The edges always fold even after a water quench. I have had better results with coil springs from cars.
The shaft ends were greased and the new bearings were carefully tapped on with a brass hammer. It is consistently amazing to me that I can buy hardened steel bearings very cheaply that are manufactured to tolerances of microns. wow.
OK LETS MAKE IT
I scrounged some cylindrical aluminum things to make the bearing housings. I used the one on the left for both housings. It is from a piece of optical equipment and is gummier to machine than 6061-T6, but seems to weld fine and do the job fine.
The stator standoffs were made on the mill.
The front and back plates are made of 3/8″ 6061 aluminum. I printed a drawing with a hole table and used the DRO on the mill to make quick work of the holes.
The tall verticals were machined in parallel. They were scrounged from some absurd piece of decommissioned X-ray microscopy equipment, a behemoth constructed of billet, stepper motors, and precision motion components.
I recently gained access to a shop with a nice TIG welder, a Miller Dimension 200 to be exact. Even when flooring it at 200 amps it took a while to heat up this plate enough to get the puddle going. Once the plate was hot, the metal flowed more easily. I had to stop after every few inches of weld to let the TIG torch cool down. I set the AC balance to 80% electrode negative to try to deliver more heat to the work and less to the torch. The “black pepper flakes” present in some of the welds means I could have used a lower AC balance for more cleaning action.
I goofed hard at one point and mistook stainless steel filler rod for the aluminum I should have been using. It’s easy to tell after the rod has been heated because the stainless shows the little rainbow of oxidation colors at the tip. I used a flap wheel to grind the steel out of the goofed area before adding a whole rod’s worth of (aluminum) filler to make up for my mistake.
The welds got a bit more difficult on the tall verticals. All surfaces were quite hot and there wasn’t a whole lot of room to maneuver the torch. I need to watch some more pro TIG welders.
All of the welding took about 8 hours spread over three days. It would have taken less time if I’d done it in one shot because a significant chunk of that time is spent in prep/cleanup/heating the stock.
Bak 2 the lab:
I set up the crazy welded plate on the rotary table and used a piece of precision ground stock to find an approximate center based on the locations of the three M8 tapped holes.
Now here’s the goofy part: cutting the big 180mm bore for a sliding fit over the minor diameter of the stator. I’m using a long 1″ end mill. It has about two inches of exposed cutter on the end, followed by another three of shank. It was the only cutter in MITERS long enough to reach down into this birdcage of metal. I gave this operation a good deal of consideration back in the design phase. I knew the (relatively floppy) tall verticals would chatter out the wazoo, but I figured it was whangle-able. This probably isn’t going to be a CNC job any time soon because I dampened the chatter by holding a rubber mallet against the uncut side of the vertical. I considered other methods of stiffening the verticals like gussets or a band around them, but I figured I’d try this way first.
I goofed with the nod and tilt on the head of the mill to try to get the cutter’s rotational axis more parallel with the rotary table axis. I swear I’ll buy a proper dial indicator one of these days.
The fit is gud. The verticals closest to the lobes needed a bit of cleaning with a file to clear the lobe fillet.
Next: the holes in the tall verticals and the front plate bearing bore. I used the Oh No Not This Again method of centering with a thin piece of brass wire in the drill chuck. I bent the wire and moved the mill table until the wire tip just barely scraped all the way around.
I spotted and drilled the holes for spring pins and screw connections. Conveniently, the hole sizes for a snug M4 tapped hole and a 1/8″ spring pin are both about 1/8″. I quit using the center drill and made all the holes with a sharp 1/8″ bit. It didn’t wander perceptibly with light starting pressure. The hole locations were eyeballed and recorded using the DRO so that I could repeat them on the back plate.
With the stator removed, I bored the housing for the lower bearing.
at MITERS we go hard: power tapping M4 holes.
The back plate got its bearing housing bored, and holes were drilled to match the spring pins and tapped holes on the long verticals.
A test fit with the rotor showed about 1.2mm of axial slop. I made the bearing bores on the deep side so I could take up the slop with a spacer after the fact. That’s easier than trying to make an existing bore deeper.
The big Clausing 6906 in MITERS was in use, so I went upstairs to visit Tinylathe and make the spacer.
OK OK LETS PLUG IT ALL IN
again, the trusty jasontroller is the sensorless Chinese ebike controller of choice when it comes to driving Random Ass (three phase synchronous) Motors without hall sensors.
WELL IT SPINS AREN’T YOU HAPPY??? hmm, it draws 1.5 amps at no load when running at 35 volts. That’s like 50W, like a lightbulb, kind of a lot. dang.
The weight of the motor + housing is a chunky 47 lbs (21 kg).
unlike the sensorless jasontroller, bayley’s hacked prius controller-controller needs sensor input. Here I tried the time honored tradition of gluing the hall sensors in amongst the stator teeth.
how to get the spacing right? check out the next in my MS paint mini-series: how do i figure out my hall sensor spacing?
This almost worked. This frustrating scope shot shows the purple and yellow sensors seemingly working fine, but the blue sensor is phased incorrectly and has some kind of consistent false triggering.
well that’s where we are now. There is more to be figured here.