Ford Fusion Hybrid Motor: How to get it to rotate outside of a Ford Fusion Hybrid

This big electric motor sat in MITERS for a while after being snagged from a cleanout of another nearby shop.

IMG_1801

I heard it was pulled from a Ford Fusion Hybrid.  My current understanding of the hybrid car situation is that many of the electric motors/generators in Ford and Toyota hybrids look and function about the same, and are made by a company called Aisin Seiki.

Specs:

Type: 8 Pole Permanent Magnet AC Synchronous Motor

Rated Power: 106 hp (79 kW) @ 6,500 rpm

Nominal Voltage: 275 V

Kv: 20 RPM/V

Kt: 0.5 N*m/A

Phase Resistance: 11.7 mΩ (assuming wye termination)

How is Moter formed?

The rotor is a stack of silicon steel laminations held captive between two cast aluminum endplates.  There are permanent magnets captive within the laminations.  This image, taken from this excellent Oak Ridge National Labs prius motor teardown and characterization report (SRSLY this is what national labs should be doing), shows what a single rotor punching looks like.

lams

I held a small magnet against the surface to feel for the number and orientation of the magnets within the rotor.  I count 8 alternating poles, that's 4 north-south pole pairs.  The magnets provide enough surface field strength to stick the rotor pretty well to most ferrous objects.  It will not hang on to a vertical wall, but it does require a pretty good tug to free it from an iron surface.  I was able to push it out of the stator without finger pinching or special tools, but it did require some HNNNGGG.

IMG<em>1803IMG</em>1807

A big nut on one end holds the stack together, and the threads are smashed in post-tightening to make it more difficult to remove.

IMG_1805

On the other end of the rotor is a lumpy shaped stack of laminations, measuring about 6mm thick.  It's got 90 degree rotational symmetry, same as the rotor magnet spacing.  It is almost certainly a position sensor.  I am guessing and have also heard mumblings that it is part of a variable inductance reluctance resolver.

IMG_1806

I wasn't present for the pulling of this motor, so I'm not sure what that part looks like.  It seems to be lost, so we'll have to figure out some other kind of rotor position sensor if we're going to turn this thing on.

Unfortunately, in shucking the motor from the car, the structure which holds the rotor concentrically with the stator was lost.  Since the rotor is a big stack of magnets and iron, and the stator is a big stack of copper and iron, they stick to one another unless they're held apart.  These stators are designed to plug into a big, machined aluminum casting that's part of the hybrid electric vehicle's powertrain.

Something like this (left: Toyota Highlander, right: Toyota Prius)

Highlander_MGR Prius Final Drive

(images from link)

perhaps that green thing in the bottom left of the prius picture is part of the resolver setup.

How make spin?

Well dang.  That's a hefty chunk of aluminum.  Its job is to hold the rotor in the middle of the stator.  It looks like the outer perimeter of the stator is a sliding fit against the casting.  There are three big cap screws that go into tapped holes in the casting to clamp the stator in.  The cup to hold the outer race of the back bearing is machined into the casting.  The cup that holds the front bearing is machined into the front cover, and that is located by some means and held down with a bunch of machine screws.

None of the bearings have seals on them, so I assume that the whole thing is sealed and bathes in lubricant.

I made rough models of the rotor and stator, then designed a structure to hold them.

stator rotor

one of these is fixed to either endplate and holds the bearings

frunt burrin house

ASSEMBLY

bayleyw had a pair of 3/4″ slices of 12″ round aluminum extrusion left over from a previous attempt at housing this motor.  I decided to include them in this revision also.  The verticals are made of about 4 feet of 1″x2″ 6061 bar.  The bearing housings are made from 1″ slices of 6061 tube with a 3.5″ OD and 2.5″ ID.  The metal totaled about 50 dollars from speedymetals, not including the big slices of round.  Fasteners were purchased from McMaster for about 30 dollars.

This design, apart from using a metric crap-ton of aluminum, would make the engineers weep.  The bearings aren't sealed, it requires a bunch of machined parts, and is rather silly looking.  Its purpose, as of writing this, is to let us spin up this dang motor.  In engineering terms, that means maintaining the rather tight airgap between the rotor and stator.

I did some rough calculations to see if the big M8 screws through the stator would be enough to keep it from shifting if the motor was say… dropped.  After all, the real designs hug the whole stator with a big hunk of CNC machined aluminum.  Assuming the screws are tightened to 80% of yield, and have a big fat washer on the top, and a pessimistic coefficient of friction of 0.5 between the stator and the mounts, I found that thing can tolerate in excess of 10x its weight in shear force between the top and bottom endplates.

Good enough.

It was already a CAD model, so I ran some derpy and terrible FEA using the number 22 kN, about 4500 lbf applied in uniform shear between the two endplates.  The whole thing weighs about 80 lbs when assembled.  So, when smashed under about 20x its own weight, computer sez it deflects about 0.3mm.  That's smaller, but the same order of magnitude as the size of the airgap.

FEA_TESTER

Again, not wicked stiff, but should be good enough to spin it up.

OK LET'S MAKE ONE

After tramming the vise and squaring up the mill, I set up a vise stop for drilling and tapping the holes in the three small and six big standoffs.  All holes are M8-1.25, power tapped by Our Lady Of Bridgeport.

IMG_1863

draw1IMG<em>1864IMG</em>1862

now on to the rounds.  I was originally planning on facing them with a flycutter, but despite being saw cut, they were exceptionally flat.  I used a hefty chunk of precision ground steel to rock over the surface, trying to feel the presence of any strong wonk.  After  being convinced of their flatness, I clamped them to the mill bed and went into CNC drilling machine mode, following the cartesian coordinates on my drawing with the mill's DRO.

IMG<em>1861draw2IMG</em>1866IMG_1867

I tapped the M6 holes using the only reasonable method of tapping M6 holes, the fabled MITERS DeWalt cordless drill.  It was obtained from a lab cleanout and zombied back from NiMH hell by the most versatile and venerable A123 12V7 battery module.

IMG_1880

next up are the bearing mounts.  I turned the sides of the stock square on the lathe and bored the inside to 58mm to get a reference for future fixturing.  The holes were done back over on the mill.  I didn't have a counterbore for M6 socket head cap screws, so I used a drill bit of diameter approximately 11mm with the depth stop set on the quill.

IMG_1873

With holes drilled and parts made, everything was screwed to the lower plate.  Liberal amounts of thread locking adhesive were applied and the screws were snugged.

IMG_1874

When the M8 cap screws were cranked to spec with a torque wrench, oil oozed from between the stator laminations.  Thats.. uh.. reassuring.

IMG_1876

next, the top plate was done up similarly to the bottom.  I goofed the alignment/handedness of the wire-exit hole in CAD, so one of the long vertical standoffs had to be left out.  I used the boring head to bring the output hole in the middle up to the 58mm dimension.

IMG<em>1877IMG</em>1879 IMG<em>1878 IMG</em>1883IMG_1888

IMG_1886

OK, here's where it gets silly.  The plan to do the final bearing housing boring is to indicate off of the inside of the stator, then do the boring based on that zero rather than dealing with tolerance stackup from all the other parts in my assembly.  I know the airgap is on the order of 0.5-1 mm, so I've got a fair amount of room to play with.  The big dial indicator didn't fit, so I used a piece of brass wire in the drilling chuck on the mill.  I put the mill in “neutral” to be able to turn the spindle easily.

IMG_1889

I bent the brass wire and moved the table around until it just barely scraped on all sides.  by tapping the wire at the elbow, I could feel whether there was a gap between it and the stator.  The fineness of my gauging was on par with the 0.01 mm resolution of the DRO.  That is, I could move the table by 0.01 mm in x or y (according to the DRO) and detect a difference with the brass wire feeler.  I picked up about that much runout in the vertical direction.  That means that the assembly was rotated a small, but perceptible amount (with respect to the mill's coordinate system) in the negative direction about the mill's X axis.  I called it insignificant for this application.

IMG_1891

Great, we're concentric with the stator.  Next, I attached the top plate with a piece of scrap attached in the bearing mount position to practice boring and gain faith in the setup before doing the big critical goofy boring operation on the bottom plate.  It took a few tries, mostly because the MITERS calipers differ by more than a tenth of a millimeter in their readings of inside diameters.  Knowing that calipers aren't really the best tool for measuring round bores, I looked for a telescopic feeler gage.  We didn't have one, so I made a makeshift gage of 1/4″-28 screws with the heads ground round and a coupling nut.

IMG_0083

It worked well, I was able to make confident measurements to within a hundredth of a millimeter.

IMG_1892

OK, I trust the tools.  In we go:

IMG_1893

I took a few 0.080″ passes, and the last pass turned out to be 0.010″ – about 250 microns.  I worked last summer in a shop making mostly metric stuff with imperial tools.  If there's one thing that stuck, it's that 0.040″ is darn close to 1 mm.

IMG_1895

Well there's that.  So far, the feeler gage is my only indication of whether this thing is the right size.  I scribed all the parts so that their relative orientation can be maintained if the thing ever has to come apart.

We're close now.  With the lower bore complete, I fastened the top endplate to the rest of the assembly.  The whole thing is still fixtured as it was for the machining of the lower bore.  I transferred that zero to the top endplate and its bearing housing, using the boring head set to about 58 mm on the diameter.

IMG_1896

The top endplate had to be flipped over to properly machine that bore, so I did that, indicating off of the just-machined surface.

IMG_1902

well here we go, let's see how that bore is.

IMG_1898

The rotor has some M6 tapped holes in the endplate.  I suspect that they have a jig for lowering it in at the factory.  I used some M6 screws and some nylon cord.

IMG_1899

a combined holy !(#^@$ and high fives were shared when the thing fit the lower bore OK.

IMG_1901

A second round of joyous expletives and high fives were shared when it was found that the upper bore also fit, and that the rotor could be spun smoothly.

IMG_1903

YEE

here's a shot from one of bayleyw's absurd cameras.  for more information please visit bayley taking pictures of things dot tumblr dot com

CF004255

Well it spins.  It's a “high voltage” motor, so shorting it while spinning it at modest speeds creates satisfying sparks.  If I knew more about motors I would have estimated the Kv by now.  Shane?  Ben?  Charles?

There still exists a nagging mechE bug that hasn't yet manifested itself:

a night or two before starting to make this thing, I figured that the stator should be attached to the top plate rather than the bottom one.  Doing so would shorten the structural loop between the stator and the rotor output splines and make the assembly stiffer at no additional cost.  I changed up the CAD but stupidly forgot to implement the changes when actually putting it together.

See below: structural path from stator to top plate

(left: how it is, right: how it should have been)

loop1 loop2

frustrating, but OK for now.  I'm glad that it spins.  Now let's give it some juice!

Next up: sensors.

Options include:

  1. trying to find the original resolver setup and figuring out how to extract information from it.

  2. adapting some kind of optical encoder.  this could potentially match whatever resolution the resolver provided.

  3. using hall sensors positioned close to the rotor.  this only works if the fringe fields on the rotor are strong enough to be reliably detected.  could be adapted to use additional permanent magnets hot glued to the rotor or something.  depending on how many magnets/ what configuration is used, this may only yield enough information for block commutation.  that's probably OK.

Wait, why hasn't this been plugged into a sensorless jasontroller yet?!  I have certainly spun another “300V motor” with a jason, albeit rather slowly.

which leads us to…

PART 2: UNSENSORED JASONTROLLING and motor characterization

Jasontroller is the name given to a particular strain of Chinese electric bicycle motor controllers by none other than charlesg.  Depending on the variety, they can typically be had for $50 or less, shipped from ebay.

The one shown here is of the 36V 500W variety.  It's running on an A123 pack of about 40 volts.  And yes, it spins this big ole car motor.  If Jason can do it with no sensors, why not Prius?

Next we took the motor over to the EE bench.  It was spun up with the jasontroller, then the wires linking the motor and controller were rapidly disconnected, and a scope shot of the back EMF was taken:

DS1Z_QucikPrint1

 

Hmm, looks fairly sinusoidal.  I remember reading in a paper somewhere that they put the magnets in that funny v-shaped configuration in the rotor to minimize THD in the BEMF.

 Kv/Kt estimate (probably wrong – just backdriving the motor doesn't account for reluctance torque)

That's 28V peak to peak at an electrical frequency of 40Hz.  This motor has 8 poles on the rotor (4 pole pairs), so the mechanical frequency that corresponds to a 40Hz electrical frequency is 40/4 = 10Hz.  10 Hz is 600 RPM, and 600RPM/28V is about 20 RPM/V.

Got Kv, let's make that into Kt.

when people make models for motors, they sometimes assume that IV = Tω

that is: electrical power in (amps * volts) = mechanical power out (newton meters * radians/second)  It works both ways.  In and out are interchangeable depending on whether you're motoring or generator-ing.

IV = Tω can be rearranged to make ω/V = I/T

In MKgS units, Kt and Kv are inverses.  Whoops, RPM isn't MKgS.  1 radian per second is approximately 10 RPM, so the Kv calculated earlier for this motor is 2 radians/Volt*seconds (when converted to MKgS).

The inverse of 2 is 0.5.  Kt is 0.5 N*m/A.

Next up, phase resistance.  We want this number so we know how much power is being dissipated in the windings when we jam a hundred amps through this thing.

IMG<em>1919 IMG</em>1920

Here you see a four wire resistance measurement, a good way to measure resistances that are similar or smaller than those in your multimeter's leads.  My cheap meter has leads with a combined resistance of about an ohm and I know the resistance I'm measuring is much less than that.

Here, I set the current controlled bench supply put out an amp, then used the cheap multimeter to measure the voltage drop across two leads on the motor with an amp of current running through it.  A drop of 23.4 mV at 1A corresponds to a resistance of 23.4 mΩ.

Assuming that the motor is Y-terminated, this means that each phase has a resistance of about 11.7 mΩ.  I'm not sure how great of an assumption this is.  I'm making it because I have observed that 3 phase motors designed for efficiency are generally terminated in the Y style… something about fewer losses due to eddy currents.

progress on This Huge Motor is currently on hold while we get to know its smaller (more reasonable?) kin.