Why only scoot when you can fly? 
The idea of waterskiing from the 8ft DANE-GER boat was postulated and after rumaging about, components were aquired to build a glorious 60-ish HP electric outboard. Last episode we painted and sealed the boat, and prior to that a 1kw dc outboard was born.  Read on for details of using a hybrid Buick Lacrosse three phase induction motor, a 600A induction forklift controler and a pile of miscelaneous parts for creating the glorious ULTRA-OUTBOARD.

Removing the Engine
Electric Motor Adapter Endurance Test Conclusion Image Directory


Remember the 8ft aluminum boat from last episode? [link] Imagine waterskiing behind it. 48 KW (60HP) OF DELICIOUS 3 PHASE in an outboard-shaped format attached to a small boat.

The Innagural test of the ultra thruster

WHAT CAN DO 600A at 80V do?


Why not just run a gasoline outboard?
Second hand gasoline outboards are actually fairly easy to come by, there are some disadvantages:
* Not permissible in certain lakes / resivours (generally there are no rules for electric propulsion, namely because there arent any high power options)
* Greater than 3hp gets fairly expensive quickly. A 10hp functional (used) outboard start at $300 in the Western Mass Area. Transom mounts for large engines get fairly heavy as well (to support large engine)s.
One of the main reasons for looking at electric was the first issue: Gasoline engines werent permitted, electric craft? Goto town.

The glorious alter-motor

Behold the alter-motor
The motor itself is a pretty excellent to behold, it is the size and shape of a normal truck alternator but there's much more buried underneath, the first two 'different' parts are the coolant lines, as this is outputting and absorbing quite a bit of energy, losses need to go somewhere. In this case the coolan lines enter a shroud which goes about the stator. Coolant to the rotor is only by proximity.
Wait what does it look like inside?
These images were from some promotional event for the BAS+ hybrid power train, shown is a really great cross section of the beast. Look at that stator copper. Serioulsy its beautiful. Note, this is an induction machine, no PM magnets on the rotor, so a bit more curious to drive.
Opening the interface up.
The end plate of this motor assembly was missing the mating sealed surface, and had a bent bolt, The actual assembly slides off after the outer bolts are removed. The housing slips vertically up, leaving giant copper lugs which fit into isolation mounts. Shown is the resolver in place.
Removing the resolver
The two-pin connector to the far right is the 'is my cover in place' connector. After sliding everything off a large in place o-ring is present, forming a face seal on the motor assembly. A large tyco connector forms the resolver and thermistor interface.
Adding a hall effect encoder
Going for 'simple' i went with adding a quick DIY quadrature hall effect encoder. The controller (shown below) consists of a
The three magnet hall position sensor. Note those are 2-56 screws tapped into a shaft collar. Tiny work indeed. One US Dime for scale. Initilly i had the wrong manual for the motor controller (below) and operated off the assumption that the controller only neede an incremental rotor position sensor, so, using a single hall effect transistor, mounted co-planar to the spinning magnet array, i should get three pulses per rotation. Curiously, after aquiring the correct manual, it turned out to require a quadratrue input, shoehorning two hall effect transisors didnt work as well as id' hoped so other options were evaluated. 
The encoders arrived!
I ordered two discs with correct sized shaft collars and a single reciever unit.  I used my favorite plastic bin [mcmaster] to sort my patrs, encoders screws, bolts and lugs. Stay organized if you can, it saves time!

Optical encoder maddness
As the hall effect incrememntal encoder approach wouldnt work in this application, i chose an encoder from US DIGITAL, low count per revolution but effective. The encoder wheel and encoder assembly sat on a 3d printed insert that mates to the existing locations that were previously used for the resolver. This worked great, magically we now had quadrature! 
The level shifting hardware, local linear regulator and wire interface to the encoder all fit neatly where the previous resolver mated to the motor. Note the quadrature out. The 6 wires, power, gnd, quad A, quad B, therm+, therm- all fed out the existing sealed connector to the motor casing assembly.
Disaster strikes

In the process of waterproofing everything, RTV goo snuck into the optical encoder assembly. Whats frustrating is at first we did not notice, the performance of the encoder just slowly degraded, and we got more and more confused. finally, after disassmbling everything we found our culprit, a fairly goo-ed in encoder assembly.
The ever excellent Krogen to the rescue
as you may imagine it was a late thursday night and US digital , SPECIFICALLY does not ship on Fridays. Comically something else needed to fill the void, for great glory was so close to existing. Peter spotted a module from a competitior on MOUSER saved the day, I rush shipped it to a nearby comrades home and waiting on their doorstep for the package to arrive that Saturday morning. One small bruce-lee-punch later the encoder was pried open and we were back to work.
Mounting the new encoder.
The replacement encoder had one benefit, a steel encoder disc, but, it also was for an ever so smaller shaft. A quick attempt at just drilling it larger ended poorly so some time with itty bitty reamers on Ben's excellent TinyLathe solved the problem, within a few hours we had an encoder disc and an encoder happily mated to the motor position sense portion of the shaft.
A new Seal
To prevent mistakes of the previous day, a 'full cover' mount for the encoder was printed up, this time helping to mitgate any of the issues from the RTV seeping into the encoder and killing the position feedback loop. Note the first photo shows the 'open top' enclosure with the new encoder and disc and the second is the 'full cover' test fit with the encoder inside.
Load testing the motor:
With 'actual quadrature encoder input' we tested the motor using 'an actual load' In this case a supercharger from Mike became an 'air load'. A matching pulley was born from aluminum round stock and the supercharger output was forced through a relativley restrictive wooden surface. This worked fairly well as a light load to make sure that integral windup wasnt too much of an issue.  The poor EE bench sat this way for a few weeks.
Some more shots of the test setup.

Note there are a PILE of interlocks that needs satisfying before the mighty sevcon roars. To simulate them, an array of tiny switches and gloriously large contactor was setup to simulate 'the boat setup'. Initially we used a low voltage 40V 70A EMCORE power supply for testing and slowly worked our way up to 80V, with an equally miters grade power supply. Finally after the amount of explosions was reduced, we switched to a single HV battery for testing.  Two 12S4P 'chibi atomic jeep' packs were put in series to allow for 'high power testing'

The Wunderous Sevcon ESPAC [80V 600A]

Behold the beautiful SEVCON espac,  80v 600A controller. These were donate to MIT's EVT team and they let me grab one for 'increased sciencing'. This is a curious beast. There are very, few off the shelf motors that this thing fits well. Low voltage high current induction motors are a bit rare. The insides of this beast are even more interesting. 
The disassembly begins! The fans are peppy and have feedback signals to indicate fan speed. This is passed through the ginormous heatsink back into the controller through a sealed grommet feedthrough.
Myself and the mighty Finberg go at the thing, attempting to pry open the controller. Its really well sealed. To open the controller, the heatsink-case seal need to be parted, which consisted of an RTV-like opaque goo.
Upon the innagural opening to peer whats going on inside, we find,  A LOT of mosfet, A LOT of capacitor and A LOT of bus-bar. This controller is curious, nominally in engineering you aim for design specifications and try to achieve them while minimizing cost. Generally, minimizing cost equates to reducing the number of components,  as every component costs board space and pnp placement time. This was quite the opposite. Arrays of parts as far as the eye can see. This does have the advantage of distributing the thermal load well across the heatsink. 
Its beautiful. Its like staring though the hubble deep feild and finding more galaxies than you ever imagined, but instead of galaxies, they were D2PACK mosfets, with INDIVIDUAL little totem-pole-esque gate drivers. The gloriousness knows no boundaries. Giant lugs were used to distribute the phases, huge LEM modules are used to measure phase amperes
This beast of mosfets is controlled with a pile of NXP parts. Curiously, 'shock mounting' the connectors is completed by using a flat-flex pcb soldered between two pcbs. The mylar board decouples manufacturing tolerances of the case without tweaking the main board, while also allowing the massive automotive connectors to be inserted without having huge tolerances on the waterproof connector cutouts. This is fairly brilliant. Holey moleys there is a lot going on in the main board. Shown on the far right is isolated power, logic and gate drive power management, and on the far left, way more CAN and NXP hardware than you would imagine would be required
Lets look at the 3 phase bridge topology
There's an enormous amount of DC rail capacitor in this controller, (or mabye i'm used to poorly spec'd foreign controllers). The three phases each sit between half bridges made up of eighteen 150V n-channel fets per leg. Part Number [SUM85N15].   Given that this controller is only rated for 80V this seemed incredibly high, but then, I checked. According to the system datasheet, the controller can handle over voltage approaching 140vDC. This is above and beyond the operations of a normal controller. The 18 Fets per phase leg suddenly made a lot of sense as they were higher impedance, higher voltage mosfets. 
Mosfets and Gate drives
Here's some more curious design details, A GATE DRIVE ONTOP OF A HIGH CURRENT COPPER POUR. This is really interesting, a Phase Leg (Phase Y) sits between its high side and low side N-Channel mosfets, and in between, nominally the copper pour that carries the phase current is a large area, generally undesturbed. This can have copper plates bolted to it, or reflowed in place. Even on a 'sinusoidal' controller, this area is NOISY, its high current switching. Impressvley, the gate drives are all sitting right atop the noise.
Shock Mount Your Mechanicals, even when your mechanicals are electricals
I learned this one from battlebots, large electrolytics have mass and can vibrate their tiny through hole board mounts off fairly eaisly. To mitigate this an RTV compound is used coating both sides of the capacitors to ensure they were dampened enough to not mechanically fail. I cant quite tell if these were done by hand or by machine, but it was a really good job, for anyone thats worked with messy epoxies or glues, keeping things tight and hitting all the right surfaces are hard.

From the sponsors:

Lets build pseudo-outboard mechanicals in an evening
With the motor controller responding, the rotor position feedback loop tuned and a procedure for satsifying all the interlocks, we now needed, er, a thing that the motor could mount to and apply propulsion to the water with. An outboard. Should be simple right???
I was gifted Two very excellent boat props from the ever-excellent Ted. The prop is huge and intended for a ~35hp boat. EXCELLENT. The prop-mate is a spline, and has a locating feature to allow it to transfer torque over a spline, be retained against a pitched locating feature, retained by a nut and then locked in place with a carter pin, or the like. Now to find the mate to this, er, beauty. .

A very long arduous ebay hunt ensued and a few days later, A SHAFT WAS FOUND. IT FIT. SAM, THE EVER EXCELLENT, located this part.  I didnt need to make a shaper tool or bludgeon my fingers with a j-head shaper attachment for a bridgeport or however these things are made.
Serpentine belt into a submerged

The Great outdoors: A test of great science

With the recently 'completed' craft, myself Peter, Sam, Birkel, Frederick, Ciarian, Jume, Austin and the remainder of miters headed out to the beautiful western waters.
The painted boat came out great and really got a chance to shine on its 'maiden' voyage. More about the painting and restoration process for the aluminum hull is available here.
The actual boat needed to now be retrofitted with all the electric and drive components. Notably are the outboard sub-assembly, the chilled water loop, the motor controller, battery pack and interface hardware.
The whole assembly was a bit precarious, the battery pack took up most of the mid-bay area while the motor controller and lever arm for the outboard took up the back portion. With 2 crew members it was a bit tight and ran fairly close to the waterline.


(There's other photos in the photo gallery)
Concluding Remarks:

Donate to Increased Science
Frustrated? Want to see a 60HP (peak) electric powered 8ft boat conquer the water (and possibly skies?) Donate to the build of the Mark II, not sure what it will be called but it will be curious.
Donations go directly to purchase of a new craft, electronics and replacing all the lost science.  
If you want to donate boat parts (like a big prop or a geartrain from an outboard) feel free to email.
Otherwise Paypal donate link below

From the sponsors:

If you have questions or comments, ask below or send over an email.
HTML Comment Box is loading comments...

(be careful, im not responsible for incredibly warm DC power cabling )

Rensselaer Polytechnic Institute 
Electrical & Electrical Power