August 16, 2014

It's Time

Thanks to Zach Both from Formlabs for taking some awesome pictures!

1:60 Scale.  Printed in 15 parts on a Form1+ and glued together with some extra black resin and a laser pointer.

Full-scale construction starting approximately now.

August 7, 2014

Chibi Atomic Jeep, the Power Racing Series, and the Detroit Maker Faire

Summers wouldn't be summers without frantically building something a week before a competition, and this summer is no different.

This year's event was the Power Racing Series at the Detroit Maker Faire.  Charles has been preparing for this event for a while, with the glorious Chibi-Mikuvan, but Dane spearheaded an effort for a brand new vehicle to race, just over a week before the event.  By the combined abilities of MITERS, we got the new cart, dubbed chibi atomic jeep, into a driveable state approximately 30 minutes before beginning our long drive to Detroit.

Here's some of the story of the fabrication of the Atomic Jeep, and its adventures in Detroit.

A frame came first.  Dane quickly welded up the frame out of some thin-walled steel tubing.

Photo credit to Dane.  I did a pretty poor job documenting the build, so most of the pictures are borrowed from other people.

I whipped up a brake disc and caliper mount or the back axle.  The brake disc is retained entirely by clamping force.  Keyways are for sissies.

Photo credit to Dane

The motor of choice was an old mikuvan alternator.  Alternators are pretty cool, because they have 3-phase windings, and a wound claw-pole type rotor powered through brushed slip rings.  When operated as a motor rather than generator, you put a current into the rotor to generate a magnetic field, instead of using permanent magnets.  Then you can commutate it like any other 3-phase motor.  Since you can control current to the rotor, you can on the fly change the torque constant of the motor by increasing or decreasing this current.  This effectively gives the motor a continuously variable transmission.  Start out at low speed with high field current for high torque, and then back off the field current as you speed up, so that you can reach high top speeds.  

We quickly got the alternator spinning using an RC car motor controller, but for the competition wanted something a bit more reliable and controllable.  The Power Racing Series limits you to 1440 watts by means of a fuse, so some form of battery-side current limiting is necessary to prevent you from popping fuses.

I had the old Kelly KBS48121 from my tricycle lying around, so I worked on convincing it to commutate the alternator.  Unfortunately, there was no way to cram hall effect sensors inside the alternator itself.  This meant externally attaching sensors somewhere to the motor shaft. 

I made the exciting discovery that CD drive brushless motors have the same pole count (12) as the alternator.  They also conveniently come with PCBs that have hall effect sensors already mounted in the correct locations.  Unfortunately, the hall sensors weren't the type that work with normal motor controllers, so I popped them off and glued standard halls in their place.  I also laser cut a mount to fix the CD drive motor behind the alternator:

Because of a slight lack of concentricity in my coupling between the alternator back shaft and CD drive motor, I mounted the CD motor on a laser cut delrin flexure, to accommodate the shaft wobbling around.  The delrin flexure was later made unnecessary by removing brushless motor stator and bushing, so that the rotor magnets and hall sensor board were not physically connected to each other.

By now I've memorized the important parts of wiring up Kelly controllers, so I hooked everything up and by some magic the alternator spun:

The cart then got steering and a real go-kart seat (borrowed form CapKart).  Thanks Mike for not letting us accidentally build reverse steering

Photo credit Dane

Dane provided some beautiful A123 Systems battery modules for the kart:

Photo credit Dane

Vice grips make the best steering wheels:

While the rest of the cart fabrication was going on, Rob Reeve was machining gears.  He volunteered to make a two stage herringbone gear reduction for the cart.  Unfortunately due to time constraints it became a helical gearbox rather than herringbone, but the gears are still some of the most beautiful objects to ever be produced on the MITERS bridgeport.

Here are the gears being machine.  The contraption on the left of the mill bed is essentially an indexing head coupled to the mill's X stage.  The liked rotation and translation allow you to machine helices.  Combining this motion with a gear cutter and a funny angle on the head of the Bridgeport allows you to machine helical gears.

Photo Credit Dane
Some of the resulting large gears:

Photo Credit Dane
 The gears were assembled into this glorious gearbox.  The gears were TIG welded to steel shafts by Mike.  The aluminum plates holding everything were waterjetted, and Rob machined the spacers from stainless steel.  I turned the aluminum spacers that hang the motor, as well as the giant steel shaft coupling to connect the alternator to the gearbox's input shaft.

Despite its helical-ness, due to a combination of slight lack of concentricity from welding, low-quality gearbox bearings, and possibly use of the wrong gearcutter on the larger gears, the gearbox was very noisy and had to be run in to spin smoothly.  This was done by slathering the gears in polishing compound and backdriving the gearbox with a giant DC motor coupled to the axle.

The gearbox was attached to the cart and I did some quick test driving a few hours before we planned to leave for Detroit.  And it sucked.  Despite vigorously spooling up on the bench, the alternator failed to produce any real torque when pushing a person.  An electronic shifter was set up for switching the rotor current with a rotary switch and some power resistors, and in the highest torque "gear", the cart would give a brief kick before topping out at around five miles an hour.  In faster "gears", the torque produced was pitiful, and failed to bring the cart to any reasonable speed.

With only a couple of hours to work, we decided to swap the alternator for the go-to MITERS motor, an 80mm RC airplane outrunner, like the trike motor but all black and 180 Kv instead of 130.  This particular specimen used to live on Brickscooter.  Unfortunately, no pictures were taken in the rush to do the conversion.  Performance was much more reasonable after the conversion, even when limiting battery current to 40 amps.

The cart was also fitted with the modified bodywork from a Jeep Power Wheels car.  For the drive, it was ratchet strapped to the roof of a rental jeep.

Photo Credit Bayley

The cart at the Racing Series track on saturday morning:

Photo Credit Dane
Atomic Jeep and Chibi-Mikuvan next to our pit:

3D-printed doge hood ornament and googly eyes:

The MITERS/ MIT Department of Silly Go Karts pit was very busy during and between races:

As soon as we started testing on the track we ran into problems with the Kelly motor controller.  Just as I experienced on the tricycle, before upgrading to the high speed version of the controller, the Kelly would spontaneously cut out.  This usually occurred when quickly going from full throttle to zero, back to full throttle again, and forced the driver to quickly cut power and reset the controller before starting up again.  Still, we qualified with a lap time of around 18 seconds (the fastest were around 15.5), seeding us 7th overall for speed.

After qualifying was a "Moxy Round", during which you are tasked with entertaining the crowd as best you can, and are judged by three small children selected from the audience:  We paraded all the small silly EV's we brought: Dane and Rob on the Atomic Jeep, Mike on the (really) tinykart, me on my trike, and Charles on eNanoHerpyBike.  This procession was followed by bribing the judges with Oreos, which earned us one of the highest moxy scores of the day.

For the first race, a 20 lap sprint, everything went wrong immediately, with the motor controller cutting out constantly.  Still, we somehow managed to qualify in the top half of cars for the next round of racing.  Lots of Kelly setting tweaking and more careful driving made the controller somewhat more reliable.

By the endurance race, we'd gotten good enough preventing the Kelly from cutting out, and resetting it quickly when it did, to perform reasonably well.  That is, until all our harbor freight wheels started exploding.  It turns out the super cheap hand truck tires are pretty terrible in every way for go-kart duty.  Mostly we burned through tires, but in a couple cases, turning forces caused the stamped sheet metal hubs on the front wheels to shear apart, leaving us wheel-less.  We managed to complete 132 laps, for 10th place in the endurance race.

Atomic Jeep, Chibi-Mikuvan, and Nimby Ferarri in the endurance race.

Photo Credit Bayley

Photo Credit Bayley
 There was plenty of down time for triking between races.  Apparently someone on the side was timing some of my laps, and recorded a 12.4 second lap.  2.7 times the power and 1/2 the vehicle weight of most Power Series cars means 25% greater speed (than the fastest cars) around the track.  The trike definitely needs a larger track to really make use of its power and gears.

The future of the Atomic Jeep:
Unclear at the moment.  Numerous improvements could be made:  Tires that don't suck, a high-speed Kelly to eliminate spontaneous cutouts, new gearbox bearings (the current ones got pretty sad from racing), better steering geometry...
Most likely, some of these changes will be implemented in the week or two before the New York Maker Faire.

Lots more pictures and videos can be found on Dane's much more thorough writeup.

July 17, 2014

Various Updates to the Tiny Go-Kart

Over the past several months, the one-day kart has seen lots of use and received a number of slight upgrades.  Its light weight, relatively low power, and low top speed make it an excellent vehicle to demo for other students and people visiting MITERS.  At least a dozen strangers have driven the kart around the hallways of N52.   User reviews include "that kart made my day!"

First, the 80/20 bar we used for handlebars was replaced with a real go kart steering wheel.  This was obtained from the cruftlabs cleanout earlier this year.  The real go kart wheel makes it even sillier and even more fun to drift around empty hallways.

Also, I implemented the Kelly controller's reverse and regenerative braking features.  A home-made brake trigger on the opposite side as the throttle enables braking.  Reverse is engaged by flipping the toggle switch on the steering wheel.

Outdoor riding (very not recommended on this vehicle) ground down the scooter wheels in front, so they were replaced with newer, slightly larger diameter wheels:

The kart's only real failure point has been its fall effect sensor board.  In its previous position, occasionally riders would accidentally kick the board with their heel and rip the connectors off.  To fix this, the motor and sensor assembly was rotated by 90 degrees:

Upgraded kart:

Stay tuned for more vehicular silliness soon.

June 6, 2014

Spontaneous Hub Motor

A couple weeks ago, Michael's 2-hour electric scooter made me sad about my lack of small, practical electric vehicle.  As a result of the piles of motor stators and magnets lying around, my recent acquisition of some scooter wheels, and quality 1:00 AM logic , I came to the conclusion that I should build a hub motor to fix this problem.

The first problem I ran into was that there were no steel pipes of appropriate diameter from which I could make the motor can.  This bit needs to be steel for its magnetic properties.  However, the MITERS drawer of Absurd Round Things had some steel gears in it, which were similar diameter and also had a face width equal to the length of my magnets:

I think this makes me a terrible person...
I machined the giant gear into a thin ring on MITERS's new Big Lathe.

I carefully glued the magnets in place with thickened epoxy.  I went with a 16 magnet configuration, rather than the usual 14 for a 12-tooth stator.  This gives the motor massive cogging torque, but should also up the torque constant a bit.  Anyway, human inertia should smooth out the effects of cogging.

I then bored out a 125 mm scooter wheel and pressed it onto the can.

I made endcaps out of a big section of aluminum round.  I could not find a matching set of appropriately sized bearings, so they are mismatched.  The endcaps have a small lip around their edges, which fits around the steel can.  The outer portion of the lip compresses the tire, while the recessed portion clamps onto the can.

Next came an axle to hold the stator.  The stator is retained to the shaft by a bolt at its edge acting both as a bolt and a key, as well as epoxy.  The ends of the axle are squared off, since the axle-frame interface has to withstand motor torque.

This particular stator was harvested from a large copier, and is almost exactly 1/3 of a 80-100 "Mellon" stator  in size.

The endcaps are clamped together by six M3 screws around the edges:

The stator was wound with 2 parallel strands of <> magnet wire.  The 12 slot 16 pole motor configuration means an ABCABCABCABC winding pattern.  Some rough calculations gave me an estimate of 30 turns per tooth to get the torque constant I wanted for 40V operation.  I ended up having a bit extra space on the teeth, so I put on 35 turns per tooth, with room to spare.

Here's the completely wound stator.  I originally wye terminated it, but the motor controller I was testing with had trouble spinning up, so I switched it to delta for now.

Next things on the agenda:

May 21, 2014

External Display from an iPad LCD, Part 2

Last post I showed how all the electronics stuff would work for the iPad screen.  After getting screen working it took me a few months to actually sit down and assemble everything.  Once I got around to it, it just took one evening to physically put everything together.

I first laser cut a bezel for the screen out of some thick black acrylic.  I then milled out a recess for the screen in the acrylic bezel.  I fixed the acrylic to the mill using this fancy milling fixture I machined a while back.

Rather than trying to desolder the Displayport jack from the LCD controller, I instead just cut and spliced a Displayport cable down to 8".  Same with the USB 3 cable.  Tip:  Splicing Displayport cables is terrible.  There's an absurd amount of shielding, along with the ~20 conductors you have to deal with.  Avoid if possible.

The thickest part of the entire assembly is connectors.  If this project was redone without internal connectors and with custom lower-profile electronics, it could easily be built to iPad-thickness.

For the back panel, I used the side panel of a PowerMac G5 case.  The 1/8" thick aluminum sheet is absolute overkill for this application, but I couldn't resist the anodized apple on the back.  A more reasonable material may be the back of a broken macbook screen, which would be much thinner and lighter.

I briefly considered just gluing the entire assembly together (it would have been a very Apple thing to do)  but then came to my senses and drilled and tapped some holes to screw it together.  Eventually I'll get some flat cap screws and countersink them.

To turn the Mini Displayport and USB connector into one piece, I 3D-printed a housing the two connectors press-fit into, at the appropriate spacing for my laptop's ports:

To attach the LCD to my laptop's built-in screen, I also went to my 3D printer and made these clips that friction-fit over my laptop's bezel.  This way the LCD just presses onto the side of my laptop screen, and can be removed and installed in a matter of seconds.  These printed clips were glued to the side of the acrylic bezel.

The assembly installed on my laptop: