Motivation
I started getting the idea to work on an electric skateboard back in April 2016. I wanted a way to get to the train station that didn't require taking my bike and finding a spot on the ever-crowded caltrain bike cars. I did a quick benchmark of the available boards on the market at the time:
Benchmark | ||||
Company | Power (W) | Top speed (mph) | Range (miles) | |
Enertion | 4800 | 28 | 18.6 | |
Mellow | 2300 | 25 | 10 | |
Boosted | 2000 | 22 | 6 | |
Marbel | 2000 | 25 | 16 | |
Inboard | 1800 | 24 | 10 | |
Stary | 1350 | 18.6 | 9.3 | |
Evolve | 700 | 23 | 19 | |
Yuneec | 400 | 12 | 18 | |
commute = 3.5 miles one way | ||||
I never expected to make a more reliable or safe product as any of these vendors, but I thought I could make something more customized to my needs, and hopefully learn some things along the way.
Specifications
My plan was to take my current longboard and add motors/batteries/speed controller. Starting out, I didn't realize just how many variables there were to this equation. The hardest part was figuring out what I wanted. Here was my pass at spec'ing out my board.
The main two specs were speed and range. I wanted to get to the train as fast as I did on my bike, and I wanted to be able to get to the train and back without having to recharge the board. My commute was 3.5 miles to the train and another 0.5 miles from the train to work. My average bike commute was around 14 minutes, so the minimum specs became:
Minimum top speed = 15 mph
Minimum range = 8 miles
Charge time < 3hrs
With these things in mind, I started figuring out what I needed to build my custom board
The main two specs were speed and range. I wanted to get to the train as fast as I did on my bike, and I wanted to be able to get to the train and back without having to recharge the board. My commute was 3.5 miles to the train and another 0.5 miles from the train to work. My average bike commute was around 14 minutes, so the minimum specs became:
Minimum top speed = 15 mph
Minimum range = 8 miles
Charge time < 3hrs
With these things in mind, I started figuring out what I needed to build my custom board
Layout
I did some research on typical layouts of an electric skateboard. These forums helped a lot:
http://www.electric-skateboard.builders/
https://endless-sphere.com/forums/
In general, every board needs:
http://www.electric-skateboard.builders/
https://endless-sphere.com/forums/
In general, every board needs:
- battery
- LiPo (lithium polymer) - more energy per weight, but much less user friendly. If charged incorrectly or punctured, these can catch on fire.
- LiFePo (lithium iron polymer)
- LiIon (lithium ion) - safer, less energy dense
- speed controller
- VESC - vedder speed controller, designed for electric skateboard
- Hobbyking ESC - hobbiest general purpose ESC
- Brushless DC motors
- Hub - the wheel itself is the rotor (or rotating part) of the motor, and the truck acts as the stator
- belt drive - A separate motor drives the wheel using a toothed belt
- Controller
- 2.4 GHz RC remote
- Nyko Kama wireless nunchuck
- Enclosure for protecting the electronics
Speed Controller (ESC)
The VESC had a lot of customizations like motor parameter detection (basically spin up the motor and measure the voltage, speed, etc), speed limits and regenerative braking (feeding energy back into the batteries during braking) and allowed a master/slave VESC using a CANbus interface. I didn't do too much research into generic ESCs since the VESC went on sale and seemed like exactly what I wanted.
Motors and Batteries
Spec'ing the motor and battery went hand in hand, so I've lumped them into one section. I decided I wanted to use a LiPo battery since I already had a charger for this type of battery and I wanted to go for maximum range. I landed on using 4 of these in series and upgrading my charger so I could rearrange them into 2x 6S and charge at 120W each, resulting in a 1.5hr charge time.
As for motors, a large majority of my time from April to June was struggling with a vendor in Australia who had promised me high quality hub motors but after three months was having a litany of technical issues. By mid June it seemed like the motors were unlikely to ship any time soon. I cancelled my order and did some more searching. I wanted hub motors because I liked the idea of having motors integrated into the wheels. It seemed cleaner and had less mechanical parts that could fail. The one downside is they were less customizable, if anything happened to the motors or I didn't like the final speed/torque, I would be SOL.
After going up to SF and meeting with the person who makes these motors (and getting to ride his board around the streets of SF) I finally landed on a good set. Here are some of the considerations that went into chosing the motor:
As for motors, a large majority of my time from April to June was struggling with a vendor in Australia who had promised me high quality hub motors but after three months was having a litany of technical issues. By mid June it seemed like the motors were unlikely to ship any time soon. I cancelled my order and did some more searching. I wanted hub motors because I liked the idea of having motors integrated into the wheels. It seemed cleaner and had less mechanical parts that could fail. The one downside is they were less customizable, if anything happened to the motors or I didn't like the final speed/torque, I would be SOL.
After going up to SF and meeting with the person who makes these motors (and getting to ride his board around the streets of SF) I finally landed on a good set. Here are some of the considerations that went into chosing the motor:
Speed |
These motors have a Kv of 90, which is the conversion factor of voltage into RPM. For brushless DC motors (or any motors for that matter) voltage is proportional to speed and current is proportional to torque. To get from voltage to speed of the board, we need to know the radius of the wheels and the kV of the motor:
\[RPM = kV \cdot V\] \[v (MPH) = RPM \cdot 2\pi r_{wheel} \cdot 60\frac{min}{hr} \] Plugging in 83mm/2 for the radius, 90 for kV, and 48V for the voltage (12 cells at 4V per cell), we reach a top speed of ~40mph! |
Range |
Range can be calculated from the total energy stored by the batteries. My final layout utilized four 3S batteries in series, effectively making a 12S lipo with 8000mAh capacity. Total energy is calculated by:
/[ Wh = VI \cdot t /] Where Wh is the energy output by the battery in watt-hours, and I*t is effectively the battery capacity. For my batteries, the energy output comes out to 355 Watt-hours. To convert this to a range, we need an empirical conversion factor /[ Range = E \cdot k_range/] where k_range = 0.1 km/Wh. This claims a range of 35km or 20 miles! Plenty of range. |
Torque |
My commute was generally flat, so I didn't much care about torque or acceleration, but just for kicks I calculated what grade I'd be able to go up before the motors stalled, and how fast I could reach the top speed. To do this, we need to relate current to torque. By taking the recipricol of Kv and doing the following unit magic: V/RPM -> W/(A*rad/s) -> (J/s)/(A*rad/s) -> N*m/A, you get a conversion from current through the motors to motor torque. This works out to
\[Kt = 1/(0.1\cdot Kv)\] With Kv = 90, Kt ~ 0.1 Nm/A. The maximum amperage is determined by the discharge rate of the batteries and the maximum current that the switching circuitry on the ESC can handle. My 30C batteries could discharge at 30*battery capacity current = 30*8A = 240A! The ESC current rating is 50A, so this is the limiting factor in the setup. 50A times 0.1 Nm/A is around 5Nm of torque. Doing a free body diagram shows that this torque will go into two things, spinning up the wheels and accelerating the mass of the board. The inertial of the wheels is so small that we can assume it is neglible. Then, the equation used to determine the acceleration is just \[\tau = 2\cdot mr_{wheel}a\] With the factor 2 from the fact that we have two powered wheels. The acceleration comes out to 3.51m/s^2, or about 3.5g. This will get me from 0 to top speed in around 4 seconds. But how steep of a grade can I get up? To calculate this, we need our acceleration to overcome the acceleration coming from gravity, which is \[a=g\cdot sin(\theta)\] For a = 3.51m/s^2, theta comes out to 21 degrees, or a 38% grade. For sense of scale, the maximum grade on my commute is ~3% |
Enclosure
For the first week I was protecting the delicate electronics with lots of bubble wrap and duct tape. This quickly became impractical due to the shock and vibration that was imparted to the board on my commute. There are 5 speed bumps on my way from work, and at some point one of my batteries sprung a leak and quickly dropped its voltage. Luckily, the ESC monitors voltage and slows the board down to a crawl instead of throwing me off the board. After finding the battery, which was oozing liquid (not good) I decided I needed to focus on building a protective enclosure.
I debated two methods of building the enclosure. The first was to thermoform a sheet of ABS plastic. This had the appeal of being quick, easy and cheap. My second method was to do a composite layup with glass fiber. This had the advantage of being super strong and much lighter for the strength it provides. I eventually went with the glass fiber because it was a process that I wanted to learn about.
I followed this website for how to prepare the mold and lay the fiber. I prepared the mold by measuring the batteries and components and sketching the dimensions, making sure to leave a space for the electronics to live without getting squished by the batteries. To give easier access to the charging wires, I left a space for a cutout and hinged cover, which latches with magnets.
The mold was fabricated using scrap 2x4s that I glued to a piece of plywood and routed away until I had the shape I wanted. I sealed the crevices with wood glue and prepared the entire mold with wax and PVA. Then began the messy work of laying down layers of chopped-strand glass fiber and polyester resin. I ordered a bondo fiberglass resin repair kit, which came with just the right amount of fiberglass for the job. I did 4-5 layers of glass fiber, which was probably overkill but I wanted to make a super sturdy case. The process was somewhat stressful, the clock starts as soon as you mix the hardener into the resin and has a working time of 8-12 minutes, so I had to be sure everything was ready when I started. After finishing the layup and letting it cure for [?] I spraypainted the entire enclosure black to match the board.
I debated two methods of building the enclosure. The first was to thermoform a sheet of ABS plastic. This had the appeal of being quick, easy and cheap. My second method was to do a composite layup with glass fiber. This had the advantage of being super strong and much lighter for the strength it provides. I eventually went with the glass fiber because it was a process that I wanted to learn about.
I followed this website for how to prepare the mold and lay the fiber. I prepared the mold by measuring the batteries and components and sketching the dimensions, making sure to leave a space for the electronics to live without getting squished by the batteries. To give easier access to the charging wires, I left a space for a cutout and hinged cover, which latches with magnets.
The mold was fabricated using scrap 2x4s that I glued to a piece of plywood and routed away until I had the shape I wanted. I sealed the crevices with wood glue and prepared the entire mold with wax and PVA. Then began the messy work of laying down layers of chopped-strand glass fiber and polyester resin. I ordered a bondo fiberglass resin repair kit, which came with just the right amount of fiberglass for the job. I did 4-5 layers of glass fiber, which was probably overkill but I wanted to make a super sturdy case. The process was somewhat stressful, the clock starts as soon as you mix the hardener into the resin and has a working time of 8-12 minutes, so I had to be sure everything was ready when I started. After finishing the layup and letting it cure for [?] I spraypainted the entire enclosure black to match the board.
Some lessons learned from the process: use draft!!! Draft is basically angling the sidewalls to make it easier for the board to pull from the mold. I was eventually able to pull the part off, after much cussing and prying with a hammer. Another lesson learned is that a skateboard is curved. The flat bottom surface of the mold meant that the flange I put on the enclosure did not hug the board very well, and there was a large gap around the sides. I built a wooden frame with curved edges to put in between and take up this gap, but it would have been much easier to do it on the mold[pics]
The latch I 3D printed with ABS and used epoxy to secure the magnets. This allows me to unplug the batteries from the ESC and plug them into the charger. Not quite as easy as plugging in a wall adapter, but not that bad either.
The latch I 3D printed with ABS and used epoxy to secure the magnets. This allows me to unplug the batteries from the ESC and plug them into the charger. Not quite as easy as plugging in a wall adapter, but not that bad either.
Final Product
The best way to show off the final result is with a video, so without further ado:
Charge data
I keep a tracking log of how much mileage I get per charge so I can accurately predict the range I can get on a single charge. I'm getting ~15 miles per charge, which is less than the predicted 20 miles due to the fact that I'm having the ESC limit the batteries once the voltage per cell drops below 3.5V, which is fairly conservative but elongates the lifetime of the batteries. 15 miles is plenty.
BOM
TODO
Lessons learned
TODO