Electric Longboard V2 (AT)
A super safe custom built "all terrain" (AT) electric longboard with a lot of power, a very high top speed, and a relatively short range (~9 miles)
I starting this project to build a better product. Based on the knowledge gained and lesson's learned from building and riding my old custom board for about a year, I wanted to improve upon my electric longboard building journey with another iteration. My v1 build was incomplete in terms of aesthetics. This new board should look like a complete, well thought out, beautiful product. I took inspiration from Disney's Tron.
Concept art for an AT electric longboard
The goal of this project was to build an upgrade my own v1 board. I wanted to optimize my new board based on my own user experience. I found that I wanted to change aspects of my old board as follows:
Faster top speed
Safer (more stable, wider deck)
More cost effective (and easier to assemble)
Sleek, aesthetic, futuristic
There were several things that must be sacrificed for these optimizations. Some of these sacrifices include weight, height, and carving ability. Some expected specs from this board are:
Top Speed: > 35 mph
Max Running Amperage per Motor: > 80A (throttle and braking)
Voltage: 12s (44.4V nominal)
Range: > 8 miles
Deck Width: > 10" (and drop through)
Motors: dual 6355's (timing belt/pulley driven)
There were several things that worked out pretty well on my v1 board, and therefore I would attempt to keep relatively the same. These include:
Remote control response
Deck Length and flex
A concept drawing of what a final board could look like.
In order to accomplish these goals, I developed a BOM based on experience and several calculations. This list was revised a number of times. I started with a dream board. One that could be built with unlimited money. I then worked my way into a reasonable cost range for the specs I expected to get. The image to the right shows a screenshot of my final BOM revision. There are several implications that I understood, but I will outline here for the reader. Some other considerations were written down as well, and are also explained here.
Flipsky FESC Dual 6.6 Plus with TorqueBoards 218mm trucks and wheel pulley's
Charging Setup (with power supply, Lipo charger(s) and ammo box)
Evolve GTR All Terrain Conversion Kit
Microcontroller for Lighting
'2 Lipos' refers to two 6s 4500mAh 45C~90C Turnigy Nano-Tech Lipo packs (which, in series, provides a max of 4.5*45= 202.5 continuous amps, and 12s equivalent voltage). After many hours of research, I eventually decided on a 2x 6s configuration to get the 44.4V pack. These wasn't ideal in terms of space, since 3x 4s batteries wold allow for a much thinner design, however, since this board was to be built on all terrain wheels, the ground clearance compared to my old board was much higher, which relaxed the need for a thinner battery pack. 2x 6s batteries would be easier to manage and charge as well. The 45C nano-techs were chosen due to their low impedance and high amp output, which should allow for longer life, and quick and powerful acceleration and deceleration. Low impedance Lipo's allow for very high charging rates, up to 10C, meaning I can theoretically charge the series pack at 45 amps.
Lipo's were used instead of Li-ion's due to the strict power output I specified for myself. A configuration of Li-ions in a 12s2p arrangement of 18650's should give the range I would want, however it only provides 30A! Similarly, 4p outputs 60A, and 6p outputs 90A. At this rate, I would need an 12s11p pack to give me the desired amp output of 80 amps per motor. This means I would need to somehow strap 132x 18650 batteries onto the bottom of my board. Of course, then weight might be an issue... among other things. That's 13 Lbs of battery; my entire v1 board weighed around that much. The 2x 6s Lipo's that can provide 200 amps total weighs 3.3 Lbs, and should give me a reasonable range. Lipo's are a clear winner.
Lipo batteries require special charging. This is not ideal, but is a worthy sacrifice for the reason explained above. Additionally, being able to charge at 45A is a big plus. Most Li-ion's are recommended to be charged at around 1C, meaning they charge at around 3 amps per parallel group. A standard 12s2p (of comparable range to the Lipo configuration chosen) would only allow for a charging rate of 6 amps, for comparison. Not only does this affect charging time, it also affects the the braking power, since braking is accomplished via regenerative breaking, it is better to channel the energy back into the battery rather than wasting it by using dissipative resistor. BUT the braking power is then determined by how fast you can push energy back into the battery, therefore a 45A charging rate is safer than a 6A charging rate, as it allows for much much faster braking.
The Flipsky FESC 6.6 dual plus is a set of 2x electronic speed controllers (ESC's) that take controller input data and convert that into variable power output to the motors. This specific version was chosen due to its high amperage rating (100A continuous per motor) and compact design. A bonus was that it had a built in anti-spark power switch. With the future in mind, this was also chosen because it allows for a CAN bus connection to another FESC that could potentially allow for an all wheel drive vehicle (with 2x dual ESC's, and 4 motors).
In the v1 build, I wanted to design and fabricate my own motor mounts. With a lack of resources at the time (I had no readily accessible machine shop to work in), I just decided to buy mounts that were already designed and worked with the trucks I had in mind. These motor mounts allow for easy belt tensioning, and are solid, durable options. The trucks are not listed in my screenshot because I already owned the pair I wanted to use from a previous purchase (from the time of my v1 build).
The custom design of the casings will be discussed separately, due to their complexity. To 3D print these casings, I would use SCU's Maker Lab (which is not a machine shop). They charge only for the cost of materials. The total cost of ALL prints for this entire build did not exceed $20.
6355's were used in my v1 board. Two of them are very well sufficient for acceleration if they are provided the max amperage that they can pull: 80A each. At 44.4V, at 190kV, they will provide me with the speed I specified. These are brushless out-runner motors. They are very powerful, in short; I have previously done much research regarding motor types: brushless, brushed, in-runner, out-runner...
The GTR AT kit is a 'spare part' from Evolve Skateboards USA. They are a fantastic electric skateboard manufacturer, but due to cost cutting techniques related to that fact that they want to be an economical business, buying a skateboard from them at their price would be severely under-spec'ed compared to what I want to build for a similar price. BUT, all terrain wheels for a skateboard are hard to come by, and the All-Terrain (AT) kit from evolve includes a bundle of joy when it comes to buying extra components I need. These components include 4x 6" (diameter) hubs+tubes+tires (they are basically miniature bike tires), ceramic bearings (weather proof... doesn't rust), wheel pulley's (designed to fit the hubs) and timing belts. For the drive train, the only other thing I needed was motor pulleys, which, as I learned from my v1 board, I could simply 3D print, if it was designed well.
The nano-remote was used for my previous board. It is a simple handheld radio remote controller (like that from an RC car) that has two basic functions: you can turn the remote on/off, and you can rotate the throttle forward or backwards (with a spring return to neutral).
Tracking is listed in the BOM. One instance of tracking is in the remote itself, which does not come with tracking. I can use my knowledge and experience with "Tile" bluetooth trackers to integrate bluetooth tracking into the remote. This way, I can ring my remote from my phone within close proximity, and I can also track its recent locations. The other instance of tracking was imbedding this Tile functionality into the board itself.
Telemetry is included so that I can track my speed. It is difficult to integrate an HUD (or any display) on a board that sits under your feet. I can use a telemetry package, available on Ebay, and 'communicatable' via an app on the Apple App Store so that I can check speeds and voltage (and a few other stats) after stopping to look at my phone. Alternatively, I can check my Apple Watch while riding for my current speed, but that is difficult and unsafe, and checking my exact speed is not entirely that important while riding.
I wanted this board to be legal in California. After reviewing the California law AB-604, I noticed that my board can be legal if I am careful. Essentially, the board is legal to ride in public if it cannot exceed 20mph, it is within a certain size, it has to have a floorboard that you stand on, and you wear a helmet. The only part of this that wasn't legal was the top speed. In CA, the legal limit is actually 15mph for these types of vehicles, called 'electrically motorized boards', which, funnily enough, includes electric scooters, such as lime and bird scooters. With simple software limiting, I can use my phone to set an electronic (rather than hardware) speed limit. This is done via the telemetry communication.
To legally ride an e-skate board at night, there are several additional rules (paraphrased; most have sight/distance requirements as well, such as must be seen from ___ ft away).
The operator or board must have:
yellow or white side reflectors
a white forward facing light
a red rear reflector
Provisions for these were included in the design, which is why lighting and reflectors is included in the BOM.
Receive and Nano-Remote
Single 6355 installed
Close up of the Flipsky Heatsink
I stubbled upon an opportunity to buy a cheap, used longboard deck. My ideal deck was a 10" wide drop through deck of some sort. The cheap deck I purchased for $20, and included mounting holes with tapped inserts on the front and back, as well as slots through the top of the deck for cables to run through without being seen underneath the deck. I ended up stripping the paint off the bottom of the deck, and re-applying new grip tape. Ideally I would have etched a neat design into the grip tape using a laser cutter, but the SCU maker lab decided against it (since it has potential to emit dangerous chemicals, such as chlorine). This deck, however, was barely 8.75" wide, and was not drop through. For lack of a better deal, I went with this one.
The board layout happens to be fairly similar to most production boards that separate the battery from the ESC compartment, however the reason for this design choice was not directly influenced by production boards. This has much to do with weight distribution, balance, and ground clearance. In order to go over the tallest/narrowest hurdles (like speed bumps) without 'bottoming out', I wanted a arc-like shape under the board, like many off road vehicles feature. The top of the arc would be the bottom of the deck, and the bottom points of the arc would be the wheels themselves. The wheelbase needed to be wide for stability. I choose 218mm trucks, which resulted in a final wheelbase of 10.5" from center-to-center of the 2" wide wheels (c-t-c because the wheels are rounded, so the contact patch does not extend to the width of the board. For comparison, my v1 board had a wheelbase of about 10.75" from outside to outside, since the entire width of the wheel contacts the ground). Even with this apparently narrower wheelbase, this board, compared to v1, had much more space in between the wheels (which meant room for wider belts and bigger motors, if needed). In terms of weight distribution, it is good to evenly distribute the weight under a board for a number of reasons.
FIRE SAFETY: A huge danger to using almost any high energy density energy storage device is the potential to release all that energy in a very fast uncontrollable way. This can happen with fuel, lithium ion batteries, and lithium polymer batteries. Lipo fires can be caused in a number of ways, but essentially, the chemicals inside react very violently with air. This means any sort of rupture in the outer shell or casing can cause dangerous heat, smoke, and possibly fire. Ruptures can occur due to overheating, which causes the pack to expand. Overheating can occur when the battery is overcharged or discharged or charged too quickly (i.e. too high voltage or too much current in or out). These things can generally be avoided by carefully charging the batteries, and setting the current limits within the VESC software (the Flipsky FESC is VESC based, and there is a software available that can be used to fine tune many parameters for the motor controller). I charge the two 6s batteries at 20 Amps in parallel, meaning they share 10A of current when charging. This is well below the recommended 10C when charging the 4.5Ah batteries (effectively a 9Ah battery in parallel when charging).
I wrapped the batteries with the fabric from a fireproof Lipo-safe bag. This wrap never leaves the batteries, even when the batteries are taken out of the board. For maximum safety, the batteries should be removed from the board while charging and placed inside an additional Lipo-safe bag and/or metal container. This is especially important due to the energy capacity of these batteries.
Safety (part 2)
SPEED WOBBLES: One of the biggest dangers in e-skate is what is often called 'speed wobbles', and it is essentially when (I'm using recently learned mechanical vibrations theory from MECH 141 here) riding conditions and system properties cause flutter instability. This can also be analyzed in a controls theory kind of fashion, where a sort of 'control reversal' type of phenomenon appears to happen. What happens is that when too much weight is shifted to the rear of the board and a slight shift in weight occurs. This shift in weight causes the rear trucks to start turning the board toward the direction of the slight shift in weight. Then, since the rider has inertia, the rider's weight compared to the board is shifted to cause the board to turn the opposite direction. The weight is then shifted again and the board swings back under and past the rider, so that the weight is now shifted back in its original direction and position of application, however with greater magnitude. This increase in amplitude happens for a number of reason. I can summarize this by saying that the rear of the board swings outward more than the front. This cycle of natural 'correction', as well as the natural placement of the center of gravity, causes the response of the rear of the board to have a greater amplitude than the front, since the rear trucks are undergoing a cycle of continual self-excitement (plus the mechanics of rear wheel steering must imply that the rear wheels must swing wider in amplitude than the front trucks in order to correct the direction to straight; this is why most cars today have front wheel steering). This causes NOT LINEAR (as occurs in resonance), but EXPONENTIAL growth of the response. This means the ride will fall; this has happened to me once on my old board while I was going downhill and hit a 4' long dip in the road. This does not happen with the front trucks, since the trailing nature of the rear trucks dampens the oscillations.
A rider may be able to counteract this by shifting weight forward, but a good board design will naturally want to correct itself at high speeds. Various parameters of the board may be adjusted to accommodate for this, however it is usually a good idea to try to shift the board's weight towards the front anyways. Parameters that can be adjusted include damping (bushings) and range of truck motion (tightening the trucks). Additionally, the baseplate angle of the truck attack can be adjusted in both the rear and the front to control how much the wheels turn compared to how much the board is tilted. This is more difficult to accomplish, as it requires either specialized trucks or angled risers (which raise the center of gravity of the board; also not ideal). In any regard, usually shifts toward stability sacrifice carving ability (i.e. the ability to take tight turns).
The phenomenon of speed wobbles can occur in several natural states of riding, for example riding fast enough so that wind resistance shifts weight back, and going downhill, in which a rider may attempt to stay vertical, which naturally shifts weight towards the rear. This is difficult to naturally correct in the design of the board, and therefore rider knowledge of safety is paramount in these circumstances. It will never hurt to slow down when feeling unsafe conditions (in terms of speed wobbles or any other circumstance, really).
Padding, Rider Wear, and Visibility: An important part of e-skate safety is padding and a helmet. Since this board will potentially go upwards of 35 mph, I decided to invest in a full face helmet (built for downhill longboarding) as well as padding for my knees, elbows, and wrists. I typically don't wear the right wrist guard, as it interferes with my remote controlling, which is an unfortunate circumstance. Visibility on the rider and the board is just as important as any other safety equipment, especially at night. It is important to NOT wear dark colors while riding at night. Additionally, lighting and retroreflective surfaces can be very helpful in getting drivers to clearly spot the board and operator. Flashing lights often help draw attention that may not have been given otherwise, which can also be an important tool. I plan to incorporate lighting and reflective elements into both myself as a rider AND the board.
My helmet and padding with an early board iteration (with LED under-glow, but no forward and rearward facing lighting, and no retroreflective surfaces)
With my new-found supply of reflective tape, I also outfitted my pads with white and blue retroreflective stripes to match the color scheme of the board.
A failed solder joint on my v1 longboard cause a slow and intermittent cycle of death. I realized this was to two faults. 1) I did not use a quality soldering iron, and failed to do a good solder job for these critical joints. 2) The casing design allowed for too much movement and tension in this joint.
A first test of the tail light system. This was also the first time that this ESC was powered on. Two test batteries can be seen above in the intended final location. The wire outlet from these batteries can be seen to lead to the main power wires on the ESC.
The electrical work can be divided into sub-categories, major and minor. The major refers to electrical work that is designed to carry the full current specified, and the minor refers to everything else.
Electrical work for this build was fairly simple in both the major and minor senses. The major work involved soldering of thick wires and large pieces of metal. For this I needed a fairly powerful soldering iron, for which I visited the SCU Maker Lab. I first soldered essentially an extension wire from the front of the board, where the battery will lie, to the rear of the board, where the ESC and other electronics go. The battery side was soldered directly to a series connection (with 2x XT90 ports) so that the two Lipo batteries could be directly connected to make a 12s voltage supply.
The minor electrical work involved wiring the motor sensors to the ESC; a simple task. Another task was to power the lighting with a 5v source (discussed in more detail in the lighting section). This 5V power would also provide power to the microcontroller controlling the lighting, the Tile tracker charging circuit, and the radio receiver. The signal wire from the radio receiver was 'tapped' and connected to the microcontroller (to determine control state).
The 5v power requirement was also trivial, since the ESC has a BEC, meaning it has a built in 5V converter. The main concern with this was to not exceed the 2A limit provided by the BEC.
The tile tracker was disassembled (the single-use battery removed), and it was hooked up to a 500mAh rechargeable battery, charging circuit, and the 5v power supply. It is wired such that it charges while the board is powered on (so that the main battery is not drained during long periods of time during which the board is not being used), and is also powered by the 500mAh battery when the board is not on, so that the board can be tracked even when powered off. Since traditional Tile's have limited battery life (usually 1 year) and semi-replaceable batteries (as of last year), it was important that the tile was rechargeable and charged during normal use, so that no tracking maintenance need be done, ever.
A similar setup for tracking was planned for use inside the nano-remote. I realized that, since the remote already includes a 3.7v rechargeable lithium battery, I could directly connect the tile tracker to this battery. The Tile power drainage is negligible, and therefore does not affect the charging circuit, so that the battery can be charged while also powering the tile in series. In this fashion, the remote can also be tracked from my phone remotely, even when the remote is powered off.
The ESC software must be programmed for the specific motors and power given. While there are a number of interesting and important settings to configure in the VESC software (to mention a few, the motors are running FOC style, the hall sensors in the motors are set to be read and used for feedback in breaking and acceleration, and the max and shutoff voltages and max braking and acceleration currents are set), the main one that I would like to talk about is the throttle curve. This is simply the function that determines the reponse to the controller input. To make the throttle response smooth (in addition to using the motor sensors), I used the VESC software to set the reponse to controller input to vary the current fed to the motors. It is like a car throttle that varies power applied by the vehicle. The function for which input is translated to output, however, is not linear. In fact, this function is exponential, meaning increments of the controller input at low power allow for fine adjustments in output power, however when the controller input is very high, the same increment of controller input will make a large different in output current. This makes smooth acceleration and precise control possible, and, more importantly, natural and intuitive.
MOTOR MOUNTS: The motor mounts allow for my degrees of movement for customizable positioning of the motors. I mounted the motors on the rear trucks, facing rearward. This was done so that no clearance issues would arise from the motors contacting the underside of the board. The motor mounts rely mostly on very large set screws. In v1, this proved to be an issue, as the set screws would frequently come loose. This was remedied in v1 by using epoxy. In v2, lock tight and a very high amount of toque was used to clamp down the major set screws. (I used an Allen wrench in conjunction with a long PVC pipe, which I slid onto the end of the wrench for leverage).
3D PRINTING: 3D printing is an interesting topic in the construction of this board. I have received unwarranted criticism simply based on the fact that I am using 3D printed parts for important components. While I wouldn't use 3D printed parts on a production board, I believe that they can be an important tool in one time builds (meaning I don't plan to build more than a couple). While 3D printing is traditionally used for non-load-bearing applications, such as camera mounts, small brackets, decorative models, exterior shells, etc. I believe, if designed well enough, 3D printed parts can be used (and used very effectively) in many cases that often thought of as implausible, or ineffective. I used 3D printing in 4 parts of this board: Motor Pulley's, Lighting Fixtures, ESC Casing, and Battery Casing.
The most important thing to consider when 3D printing for structural components is the 'grain' direction. FDM 3D prints are printed in layers via a continuous strand of plastic. This means that it has great horizontal stress properties. The layers are then fused to each other, which results in a weaker bond, and the part will have weaker vertical stress properties (under tension). Another important design consideration has to do with stress distribution, and avoiding stress concentrations. The best way to describe this here is through examples. The first example will be the motor pulley's.
A close up of the motors and drive system
Another important design consideration with regard to 3D printing is material choice. Initially, all prints were going to be made from ABS due to its strength and impact resistance. Several attempts were made to print the large prints with ABS, however many setbacks made this task challenging. Sparing laborious detail, warping and other issues lead to the production of a PLA print for the battery casing. This was not ideal, so testing was in order.
Subtlely Integrated Headlights (more to be seen than to see in front of the board, but they are very bright)
Tail lights are subtly integrated into the ESC casing and tucked under the end of the boad
The tail lights views from the rider position are not visible
The headlights have a very wide viewing angle (almost 180° horizontal). The legal side reflectors are seen with camera flash on.
A worn motor pulley next to a new one shows the directionality of 3D printed layers and gradual wearing
The motor pulleys are the timing pulleys directly mounted to the motor shaft. Traditionally, pulley's are secured to a shaft via set screws or keyways. These would both be insufficient in a 3D printed pulley, since the torque needed to quickly accelerate a human and a board is very large. A set screw will not easily stay set in plastic, and will quickly strip the plastic when experiencing torque. A single keyway would locate all of the force between the shaft and the pulley on one spot. This is bad because it will cause very high stress concentrations when torque is applied to the system. Therefore a unique method of locking the shaft to the pulley rotationally was devised.
The solution was described in the v1 board documentation, and it essentially relied on the shape of the shaft, which had been cut to be flat on two opposite sides; a double D (DD) shaft. The design of the pulley hugged this shaft shape exactly. This meant that a spiral of plastic was essentially wound around the shaft hundreds of times due to the nature of 3D printing (as long as the pulley was printed such that its axis is vertical). Since the material is less likely to fail in complete compression, a solid metal ring was placed around the outside of the pulley, such that the shaft and metal ring were sandwiching the plastic, and the job of the plastic was simply to match the shape of the shaft. This was likely the area that would encounter the most stress. The second area would be the teeth. Again the horizontal layers provide strength in this direction, and the stress is distributed throughout all teeth in contact. This leaves surface wear in the teeth (which are continually abraised by the timing belt) as the first predicted cause of failure. Of course, confidence in this design is only achieved through testing, as well as essentially any design.
The lights are each a set of 8 PCB mounted RBGW individually addressable 5050 LED's in a row, controllable via an Arduino-type microcontroller. These lights were mounted in 4 places in 3D printed components. The front and rear main lights are mounted to the trucks via separate 3D prints, and require no hardware in addition to the hardware already on the board. This is because they are seated in between the nuts and the trucks. This idea seemed risky, so much attention to detail was given around the area of the 3D print that was to be sandwiched. With stress distribution/concentration in mind, I designed the major mounting point to be a ~1" surface on the truck. To use a surface as a method of mounting, I essentially used the bolts to pull the the print towards the upwards sloping surface, causing the only theoretical stresses in the areas of the holes to be axially compressive (depending on how tight the bolts are screwed in) and a small stress applied to the hole due to the tension of holding up the small weight of the LED row light. The goal of this design was to spread stress throughout as much surface area as possible in the 3D print. Again, testing will determine the success of this design.
All four lighting PCB's were wired to the 5V power and to a very small Arduino-type microcontroller. The wiring of these digitally addressable LED's only requires one data wire for all lights. While the code is, in practice, fairly more complex than this, the basic functionality of the code, which I wrote for this purpose, is as follows:
Make the forward facing lights full brightness white
Make the rearward facing lights do a fun initialization sequence.
Color the rearward facing lights red at 50% brightness
Read the input from the remote control
If controller commands the board to start braking:
Flash the lights red at 100% brightness (returning to 50% each time) 4 times, then...
Hold 100% red until brake is released
Return to 50% red
A test sequence is shown, where a blue dot travels across the data direction
The startup sequene is shown. The code for interpretting receiver input is tested.
A night/day mode testing configuration is shown (not used in final board: I decided that for safety and ease of use, the lights can stay in the same mode (and on) all the time)
For added safety, a flashing function was added to the braking sequence. When the brakes are applied, the lights flash full brightness 4 times before solid full brightness.
A front view of the CAD of the ESC Casing
This is the most complex 3D printed part on this board; I estimate that I have been designing this component for at least 80 hours. There are many details that I will omit here, for the sake of brevity (yes, this documentation is already many pages long already, LOL). I will include the most important aspects and details here. The part was printed with surface that contacts the deck facing downwards. This was due to size constraints and consideration of directional strength in 3D printed parts.
The outside was designed to look professional and simple. A glance at this casing as it sits on the board would reveal little detail as to what lies underneath. The functionality hidden below should be concealed in a sleek mystery, not unlike the hood of a car. While the geometry is in fact very complex, and was not simple to design, the form conforms to the natural flow of the deck, while also giving indication it is not part of the plain wooden deck. By itself, it appears simple, on the board, the smooth transition of form from the deck to the casing provides a sleek, aerodynamic feel. In addition, it was important to keep the lowest possible profile to maximize the clearance to the ground under the board.
Another visual design goal for the casings deal with the overall look of the board. The casings cannot simply be something that looks like it was thrown onto the board to hold some parts, the casing must appear to be a part of the board. As if the the whole board was designed as one. While a casing might look 'good' on its own, its the seamlessness that it adds that carries the design into the rest of the board. Many production boards fail to accomplish this aesthetic goal, and end up designing electric boards that look like longboard with a lunchbox strapped on it.
The bottom curve was designed to exactly fit the curvature of the deck. Also, the flexibility of the plastic and the mounting holes on either side can help conform the plastic to the shape of the deck. Neoprene closed cell foam was used between the casing and the deck to seal the ESC and other electronics from the outside, effectively keeping out water and dust.
Three oval openings in the top were included to open up the flat aluminum heatsink to the outside air to prevent heat buildup. Neoprene rubber was used here as well. It was cut to the shape of the oval holes and placed between the casing and the aluminum so that outside air, and possible water, can contact the aluminum where the three holes are, but cannot penetrate to the interior of the enclosure. The 6 mounting holes were designed to fit the 6 tapped inserts already on the $20 deal deck. I chose hardware that had large head surface area so that I could minimize stress concentrations from the mounting hardware. Since the layers are printed such that the bolts could easily separate layers, I decided to distribute these stresses to different layers, so that not all the force holding the casing to the deck would act on the same layer. This was accomplished in two way: 1) by sloping the hole and the surface in contact with the bolt head slightly off from horizontal, and 2) by varying the height of that area among each set of three holes in close proximity. This way, the bolt heads were acting on different layers. Additionally, the thickness of the plastic between the bolt head and the deck was maximized.
The notch seen in the side profile view below is positioned horizontally and about 45° rotated center. There is one on the other size that is mirrored. This notch was designed to perfectly fit the 8 LED lighting PCB described in the "lighting" section above. There are paths from this notch to the inside so that the LED's can be powered, be fed data, and feed data (since all the LED's should ideally be connected to the same data wire). The PCB on which the LED's are mounted is hidden behind a thin wall, so that only the LED's are peaking out to produce light. These lights are visible from the sides and rear of the board. This was an important design decision made for nighttime safety.
A side (profile) view of the CAD of the ESC casing
A top (bottom facing) view of the CAD of the ESC casing
A peek at the bottom of the casing
The Inside: As can be seen, space is a at a premium in this design. Space efficiency was a high priority, as well as optimizing for thin-ness
Since the inside will rarely be seen, less care was given to aesthetics. Structure and function define the interior design. The main component in this casings is the ESC. For ease of assembly, a snap fit for the ESC was designed. The snap fit secures the ESC's heatsink, to which the ESC PCB was mounted.
There is a main 'post' that supports the center casing by transmitting any hits from the casing to the deck (e.g. if the casing hits a curb).
A total of 9 wires/wire bundles must exit the rear of the casing to be connected to the motors (two are bundles of thin sensor wires for the brushless motor operation, and one bundle is for the headlight at the front of the board). Ports are placed in the rear so that the ESC casing can be completely removed with ease without removing or threading wires through holes in the casing. The 6 large holes exiting the rear are for the phase wires to the motors. They are designed to perfectly fit the female ends of 5.5mm bullet connectors, into which two sets of 3-phase motor power wires fit.
The rectangular holes in the center of the rear wall are for the sensor wires (and are rectangular to fit the JST connectors).
There is a cutout on the right side (in the view above) designed to fit the microcontroller that controls the lighting. There is a rectangular hole from this cutout to the main compartment to allow for re-programming via the microUSB port in the microcontroller without needing to remove the microcontroller.
There is a thru-hole at the rear that exits from the right angled face (in the view above). This hole is designed to exactly fit the pushbutton that turns the board on/off. Integrating this pushbutton was difficult due to the depth of the button as well as clearances with other aspects of the design. The lighting fixture is directly underneath the button. The button hole goes above the wire outlets, and fits perfectly on the exterior face. The hole was designed so that the threaded button could be screwed into the 3D printed fixture material.
The round half-dome-like area in the front is meant to leave room for the receiver, the main power port (an XT90), and any extra small electronics additions if necessary. The components in this area include the tile tracker, charging circuitry (for the tile), and telemetry circuit (to broadcast live sensor data to my phone like speed, range, and voltage).
A front view of the CAD of the battery casing.
Outer battery casing
Inner battery shell
The battery and ESC casings were designed to compliment each other. Similar outer shells add to the sleek, aerodynamic aesthetic and the mystery of the complexity that may lie beneath. The curved slopes serve a few visual tasks. One task is that from any angle viewed, the slope is always visible, which helps retain a consistent look from any view. Anther task is to seamlessly merge the casing with the deck itself. The casings cannot simply be something that looks like it was thrown onto the board to hold some parts, the casing must appear to be a part of the board.
The battery casing very likely experiences the most stress while riding. It both supports the weight of and dampens the vibrations of a 3.3 Lb battery pack that experiences very high shock loads. It houses the most dangerous component. Therefore it must be protected, and it must be protected well. If this is the case, then why did I use a 3D printed part for the job. There are two answers. The first answer is that I don't have the resources to do much else. The second answer is that I wanted the design freedom that comes with 3D printing. I have been designing parts for 3D printing for a while, and I have discovered many nuances that come with the task. One of my main goals with this new board was to make it look sleek... indistinguishable from a professionally designed production board.
To account for the directional strengths inherent in the grain direction of 3D printing, the battery casing consists of two parts, both printed with layers perpendicular to each other. These two are seen above. The outer shell was printed such that the part facing the deck was downwards on the print bed. This was for aesthetics (so that the top surface is clean when the print is done. This was also done considering the directional strength of 3D printed parts. The grains run across the width of the board so that they act as a sort of bridge between each side where the bolts are mounted to the deck. The same design considerations for the ESC casing were applied here in terms of the mounting surfaces how the bolt heads contact the 3D print. This casing experiences more load than the ESC casing, therefore more layers (thickness) was added in between the deck and the bolt heads.
The inner shell was designed to hold the battery at all times, even when it is taken outside of the outer shell (which happens while charging*). The inner shell was printed such that it 'wraps' around the batteries (i.e. printed such that the surface that contacts the deck is vertical). The notches in the outer and inner casings match up so that the battery and inner shell into the outer shell can be vertically slid into place. This also ensures that the battery is physically mounted to the deck at points where the plastic is thick and nearer to the mounting bolts. Otherwise, the battery would be resting on the thin upper surface, which has the potential to separate from the main body if too much force is applied here. The force is redirected elsewhere as described.
Within the inner shell, the batteries are padded with neoprene closed cell foam to absorb and dampen the energy from shock loads. There are two thin (1mm-2mm) layers of 3D printed plastic as well as fiberglass fabric material (the fireproofing) (and the foam) separating the battery from the ground. If the outer shell starts wearing from ground contact, I can simply print another one for a few dollars. The battery is well protected from punctures and abrasion.
*I did not allow charging while the battery was still in the board for safety reasons. The safest way to charge batteries that store this much energy is in a thick walled metal container and/or inside a Lipo-safe bag. For this reason, the outer shell must be removed, and the inner shell must be removed from the outer shell while charging. I place my batteries in a metal ammo box while charging at home, and inside a large Lipo-safe bag when charging in public.
Reflectors and Stickers
In an attempt to make the board resemble the first concept drawing (the Tron one) I wanted to incorporate EL tape/panelling into the wheels and casings. After a few weeks of research, I discovered it was possible and doable... but expensive. With modern EL tape and panelling available, I could reasonable integrate a tiny power and inverter package inside the wheel spokes and have them rotate with the tire to power the ring of EL light on the wheels. EL panel can be cut to shape for the casings. However, with the need for highly customized shapes, and the lack of flexibility of EL tape and panels, the exact shapes needed would need to be cut, rather than 'bent' into shape. There are many difficulties inherent with this process, and I decided that it was not worth the cost and time (since higher cost effectiveness was a major goal).
I decided to use a less impressive form of glow-lighting that resembles lighting from the Tron universe. Instead of EL lighting, it is possible to use retroreflective material. This doesn't nearly match the quality and aesthetic of the Tron universe, but for the inherent safety and passable resemblance, I decided reflective tape would be a fair option. Retroreflective lighting reflects light in the direction of the source. This means that under the light of car headlights, these tapes should be clearly visible. After some research, I quickly discovered that there are in fact different types/grades of reflective tape (in addition to the choice of color). I chose 'engineer grade' for the blue tape due to its flexibility, and since the blue color is mainly for aesthetics in this design. I chose 'high intensity prismatic (HIP)' grade reflective tape for the white and red tapes due to their higher visibility. These colors are for the side and rear for safety (and legality). I also added the white and blue tapes to my padding for visibility. See top of page for comparative photos. I replicated the photo with my camera flash turned on here for reference. These photos also clearly demonstrate the flexibility of the blue tape, which was a straight strip of tape before it was conformed to the perimeter of the plastic padded areas. The white tape, which is not flexible, clearly reflects more light. I was able to flex the blue tape at about .25" width around the perimeter of the wheel hub.
Under flash photos, the appearance is quite beautiful, as the pure blue turns into a bright teal. White was also added to the interior of the wheels to add dimension to the wheels while spinning. A single white reflector was added to each side of the board for legality. This reflector is seen in the image above on the battery casing, which is, in this image, not attached to the board. This location was very intentional. The white reflector is only visible from a few feet away, otherwise it is hidden from view. This means that people who would like to look at the board up close (from a standing height) may not notice the white reflector hidden under the deck, however cars at a distance from which I need to be made more visible will clearly see the white reflector. Additionally, from any angle that this side white reflector cannot be see, the added white reflectors to the interiors of the wheel hubs can be seen! (This was not coincidental). See the wheels with reflectors in motion in the image shown here as well!
This section attempts to assess the project in terms of my goals and its purpose. It is not called 'conclusion' or 'completed project' or 'the final result' because there is always room for improvement. Therefore, the 'current state' refers to the project as it stands or how I left it (for now).
At the top of this page, under "Goals" I wrote:
"The goal of this project was to build an upgrade my own v1 board. I wanted to optimize my new board based on my own user experience. I found that I wanted to change aspects of my old board as follows:Faster top speedSmoother rideSmoother accelerationBetter durabilitySafer (more stable, wider deck)More cost effective (and easier to assemble)Weather-proofSleek, aesthetic, futuristicThere were several things that must be sacrificed for these optimizations. Some of these sacrifices include weight, height, and carving ability."
The top speed turned out to be very high. As of owning and using a functional version of this board for about a year, I have yet to max out the throttle when approaching high speeds. Now, considering the ESC Parameters described previously, this means that the next increment of controller input I give will generate a much larger output. It is possible that this board may be able to carry me upwards of 42-45 mph. In practice, I typically let off just before 40 mph, with a top speed of 40.2 mph. Of course, this speed is only practical and safe on long straight smooth roads (which are not incredibly common in my area... by smooth, I mean like freshly paved or roller skating rink smooth). A trip to fremont in the evening allowed me to test my top (comfortable) speed in a ~100 yard dash on a newly paved unpainted residential street. The ESC parameters and sensored motor control allows for extermely smooth acceleration and deceleration. After my year of use, the amount of catestrophic mechanical breakdowns has decreased by ∞% (i.e. no mechanical breakdowns have occured). This is a signifiant accomplishment. The breakdowns that have occured only occured with the lighting system, in which the flex of the board caused wiring to malfunction. This happened on 3 seperate occations before I re-wired the entire lighting system. No failures have occured since the re-wiring.
The deck isn't as wide as i would have liked for safety and stability, however I have made various small variations to the trucks and bushings so that the response is stable under high speeds. This, in addition to constantly adapting my riding style helps me remain stable, even at speeds of around 40 mph. In one senario, I hit a large bump on a trail, and the beginnings of what felt like speed wobbles began for a split second. My body reacted naturally by slightly offloading pressure on the board from my feet. The board naturally corrected itself after about 8 ft, and the speed wobbles died out. The maximum amplitude of the lateral motion felt to be at most 2 ft.
Savings on deals and collecting parts for about a year allowed me to slowly gather parts I needed at lower prices. Overall, this board cost only a couple hunfred dollars more than my previous board, even with the big performance improvements.
The board was relatively easier to assemble, however modifications to the truck involved machining (hacksawing) the rear crossbar to fit the width of the wheel and wheel pulley's.
I have ridden in the rain and wet roads a numer of times. No electronics damage or short circuits has appear to date, however a recent inspection reveals some rusting bolts. These will be replaced with stainless steel bolts.
Many people (and even friends who have seen the board many times) don't realize that I built the board until I explicitly tell them that I did. They often don't believe me when I do. This validates one of my major goals of designing a sleek aesthetic board that could pass as a production board.
While the lighting effects weren't nearly as cool and futuristic as I prefered, the board still looks fairly futuristic under headlights or flash photography.
The purpose of this project also written at the top of this page:
"I starting this project to build a better product."
I believe this purose has been fulfilled. I am satisfyed with this board as a daily use machine. Of course, while there's always room for imporvement, this has been a good stopping point for the time being (the day is now Dec. 26, 2019).
Ideas for the Future
This section is under construction!
Air flow and cooling in Battery Casing
Battery Charging Upgrades