This year, at Maker Faire Bay Area 2017 we raced a cardboard Toyota Corolla Trueno AE86 styled after the one from Initial D
Last year we had entered as an A-Team van built on top of a Razor crazy cart with the motor upgraded to the $80 hobby king 6374, but I broke all the speed controllers between events on the track. This year I wanted a better shaped body on the cart.
This is the 3rd revision of the belt reduction for the motor to the drive wheel. We are running the 192KV SK3 6374 motor this year, with a 15T HTD5m-15 pulley to 60T on the middle shaft, then 30T to the 72T on the wheel. There’s a back-side idler on both the short (400mm) and long (600mm) belts. The pulleys were bonded to the 10mm shaft with green Loctite 638 which worked much better than the cheesy and loose keyways we had previously, but might not be removable. Extra holes are drilled in the big 60T pulley to access the countersunk screws holding the KFL000 pillow-block bearings in place. The back-side idler on the short belt was 3d printed in Lulzbot green. On the long belt, the 3d printed one didn’t hold up in testing. After the event the 3d printed idler on the short belt also failed. The plates were water jet from 4mm or so thick aluminum from Big Blue Saw, and TIG welded by a Precision Welding next door to the hackerspace.
Here’s how the idler failed:
Here’s what testing the drivetrain, on an otherwise stock Crazy Cart looks like.
I spent a lot of time in blender converting the source model to look okay and fit our makehuman driver reasonably well. I tried to keep the overall cart length around 48 inches and width narrow enough to fit through doors. The resulting body is about 54 inches long and 32 inches wide.
The frame was designed in SketchUp and welded from .0625″ (1/16″) wall 1×1 inch square tubing. I didn’t think very hard about how to support it so I was surprised it turned out rigid enough, and more rigid than the stock Crazy Cart frame.
We forgot to take off the GoPro mount from when we first messed with this cart. It’s still there now.
Then the cart was tested with bare frame. Everything else was “attached” with hot glue, and maybe a zip tie, including the seat. Electronics are attached with vice grip.
One of my team mates asked, innocently, why don’t you just make it out of cardboard. So I bought some cardboard:
The Medium home depot moving boxes fit well into our 900×600 laser cutting area. 123d make was used to split the cart model into pieces and output DXFs for the laser cutter. It also helpfully added 1/4 inch dowel holes to align the parts. 123d make is dead now, this feature is now part of Fusion 360.
Cutting the pieces took 2 days, and gluing them also took 2 days. There’s 746 pieces on 182 “sheets” (1/4 box) for the body, and I think 48 pieces in 2 sheets for each wheel.
The body ended up too short and was cut and extended 4 inches to fit on the frame.
A wooden “body frame” was then cut, and hot glued to the cardboard. Sheet metal screws attached the wood “body frame” to the steel cart frame.
So, once the body was attached we tested it. The front fell off.
The cardboard wasn’t attached to the steel frame, so it was only the Elmers glue holding the front on, which failed. It was re-attached and the steel frame extended forward so the wooden “body frame” could be bolted on. The front has remained on since.
I built a throttle/brake mixer on a teensy that reads the hall-effect pedals (gas and brake) and outputs a single PPM signal using the Servo library. The teensy is also powered by the 5V in the VESC. It also had an input for the normally-open switch on the brake lever I planned on using to send the VESC into neutral.
The original plan had the Teensy at least outputting raw throttle and brake pedal position to another system I never got around to building. An arduino pro mini or similar would replace the teensy well to just mix throttle+brake but wouldn’t do USB as well.
The brake cable goes through the stem this year and has a delrin pull-handle I don’t have a photo of. This attaches to the stock cart wheel drum brake and isn’t very effective.
Batteries are in a smaller ammo can with XT90 hot glued in a hole in the one of the small sides.
The wheel+tire is from http://www.monsterscooterparts.com/in20sonrewha.html which were discontinuing the part before maker faire 2016 so I bought 3.
Rear casters are stock, but use cheap 52mm skateboard wheels instead of inline-skate wheels. Front casters were solid 2.5in casters from OSH with the wheel replaced with the same 52mm skateboard wheel, and side spacers from M8 washers.
Getting to Maker Faire
Swiss, our hackerspace’s dear leader, and cool dude with a pickup truck left for maker faire before the cart was completed so we had to find another way to get the cart to Maker Faire. It didn’t fit into the Honda CRV we were planning because it was too long to fit behind the folded up seats. It turns out, you can fit longer things in a hatchback than a CRV.
This is our second year running VESC and we didn’t do anything much remarkable, and didn’t have any trouble with them dying. We are running in BLDC mode with 12S Lipo, into a hardware 4.12 VESC. Our VESC is from hobby king, they mailed us a sample after maker faire last year. Our motor is the Turnigy SK3 6374 192KV and we’re running a 9.4:1 reduction to the wheel, so our theoretical top speed is around 26mph.
Initially we had the max current setting (motor max) set to 65A which worked well and with a lighter driver you could break the tire free and burn out. I lowered this to 35A because I don’t have cooling on the motor and it was much too hot to touch after the first 25 minute race on Saturday.
Back home after Maker Faire after increasing the current limits again, it’s pretty clear we are now hitting the default 80°C mosfet temperature protection which limits output power, so if we want to run with higher currents we need more active cooling. These limits are software adjustable so an easy ‘quick fix’ is to set them higher so they don’t start reducing power until the mosfets get to 90°C.
Saturday: First race went well, but motor was getting hot and we don’t have a fan in this cart. I dropped the max current to 35A (from 65A which all other videos show) to see if we could still run that slow and we proceeded to go sideways very slowly.
Sunday: swapped tire out for new one, and we proceeded to drift at slow rate. Single 6 amp-hour 12S pack lasted 45 minutes during endurance which made me happy. That works out to about 7 amps or 330W continuous which isn’t anywhere near the power limit of 30A or 1440W.
Also during the endurance race we lost the bolts holding the drift bar on. I found some spares and we got back into the race.
Overall we came in 3rd place in the Moxie category and I’m pretty happy with our improvement over not really completing any single race last year.
Next year, I have a couple more people interested and I’m starting a lot earlier. Here’s hoping we get done half of what we’re looking for.
08 Jun 2015
We need to give a big shout out to Milpas Rentals for the floor buffer and Mike Kapuscik, and volunteers John Craighead, Mike Ray, Dan Loman and Nick Winters for making the work happen to clean the floors at the new hackerspace. Without them all we wouldn’t be able to go from this
The floors were cleaned a few weeks ago. Now nearly everything is moved and we’re working on unpacking and getting utilities sorted before a grand reopening. Stay tuned for updates.
17 Sep 2013
After seeing the 3d printed Gear Bearing, Dan had an idea. Why not make skateboard wheels this way?
So I did.
They were printed to be taller than normal wheels, closer to 2.5 inches tall, and made out of ABS, printed at 0.3mm layer height with 2.2 width over layer height and 40% honeycomb infill. Each wheel took 3.5 hours to print, and I didn’t have trouble getting them rotating smoothly. They were greased with white lithium grease which makes them roll even more smoothly.
I have a video of us trying them…
They work, for a few minutes at least. They roll well but quickly the gear bearing gets more loose which doesn’t impact how they roll much during the maybe half our of riding they had. They weren’t printed with quite enough infill so the outermost layer of the wheel comes off like an old retread on the rear wheels, and they aren’t strong enough to take a kick flip.
So if you want to print them, use more than 3 perimeters, more than 40% infill, and tune your extrusion amount so that just the right amount of plastic is coming out. Maybe try polycarbonate instead of ABS so that your wheels are durable in polycarbonate ways?
The STL file and modified OpenSCAD file for this skateboard wheel design are on thingiverse.
At Maker Faire I was showing a clone of the Oculus Rift built out of an iPad Mini.
This version was built with a 3d printed case wrapped around an iPad mini. The software and controller are the same I wrote about on my personal blog previously.
The case was designed in Sketchup based on the dimensions in the apple case guidelines for iPad mini and printed on the RepRap at the SB Hackerspace.
It turned out every iPad mini I tried in the 3D warehouse had the wrong dimensions, so I just created a box of the correct dimensions from the a PDF with iPad mini dimensions from the apple developer site.
Before 3d printing I printed paper versions of each piece on paper, and I cut out foam core to make sure things were the correct size.
For the first printed part I quickly noticed my design would take FOREVER to print quickly, but this also gave me a chance to find that the 3d printed part was coming out too small. I corrected for this by setting a 1.03 scale factor when slicing future parts.
I intentionally made the gap for the iPad mini 2mm too large on each dimension so that there would be some tolerance if the prints are just a little too small, and so that the plastic doesn’t scratch up the iPad. I lined the inside parts that touch the iPad with adhesive felt from a craft store.
Building this for yourself
You will need everything listed below:
- iOS Developer subscription to deploy on a device. ($99/year)
- Mac with Xcode (I am using Xcode 4 on OS X Lion 10.7)
- Access to a 3d printer.
- iPad Mini
- 3d printed parts (Two sides, top/bottom parts (identical), lens holding piece)
- Nuts and bolts. 12 3/4″ long 1/4 20 hex head bolts, and corresponding nuts.
- 4 Lenses (http://www.frys.com/product/4383195)
- Adhesive felt (I bought a 9×11 sheet for something like $1.50 at a craft store)
- Arduino Uno
- RedBearLab BLE shield (http://www.makershed.com/Bluetooth_Low_Energy_BLE_Arduino_Shield_p/mkrbl1.htm)
- Sparkfun Joystick Shield (https://www.sparkfun.com/products/9760)
- Power Source. A 9V battery with a barrel connector cable works fine. I used a MintyBoost.
- Two LEDs, a small pushbutton, and two 330 ohm resistors are used for status lights and a “Start” button. One LED/resistor are used to indicate power (between 5V and ground) and the other is between A2 and ground to indicate that the controller is connected. The small pushbutton for “Start” connects pin 7 and ground when pressed. None of these are required for the application in its current state.
The Xcode projects, Arudino sketch, and 3d models are on github: https://github.com/sphereinabox/iPadRiftPrototype
- Arduino/BleJoystickShieldArduinoSketch is the arduino sketch to send the joystick information over BLE.
- iOS/BLEJoystickShieldMobile is a test application for the arduino sketch that shows the position of the stick and all buttons.
- iOS/CameraProcessing is the virtual application to run on the iPad.
- Sketchup/ contains the Sketchup file for the current case design, and the STL files in case you’re working outside SketchUp.
I plan on porting the native portions of this to Unity 3d so that I don’ t have to build a game engine myself. Unfortunately, this will require the pro edition (even after the free iOS announcement today) because of the render-to-texture required.
[Also, I should finally get to try a real Oculus Rift in a week or two to see how it compares.]