This is my first esk8 build, and my first actual foray into skateboarding since I was a kid. The goal is to make a reasonably sporty street board that hits my range target, hopefully around the lower end of the esk8 weight range. Additionally, something I need to work around is the fact that I have tendon problems in my arms which mean that using a remote would mess me up after a few minutes. So, I’m going to use strain gauges and an Arduino to try and make a hands-free control system that works by leaning forwards and back. I know that there are commercial boards out there that do this with weight sensors, but they all look like potato (in performance and foot ergonomics) and I would be concerned about the long term robustness of typical force sensors to vibration. I’ve been lurking here and accumulating parts for the past couple months, and started building the deck last week.
-Speed: 65kph/40mph (cuz rollerblading has desensitized me to speeds up to 50k)
-Acceleration/hill climb: non critical
-Weight: The dream is to be under 10kg, but it’s probably not going to happen. I at least want it to be ergonomically comfortable for my potato arms to pick up
-Deck: The TÜÜB special (1370mm/54in long because I prefer steady vehicles in general, I expect weight shift control will demand a wide stance for stability, and because I’m tall as heck at 200cm)
-Battery: 16s3p P42A, P-groups have been welded by the excellent Artem, and will be soldered by me. I’ll be splitting it into two 8s packs so that I can utilize my existing 500W balance charger setup
-Drivetrain: BN M1 helical 2.8, BN adjustable baseplates + Krank 93A barrels, BN220
-Trim scheme: Black on black on black on black. Possibly AC/DC stickers
I’m trying something new and documenting the build in a YouTube series, please do tell me how much it’s crap. I’m sure that composites work is old news to a lot of people on this forum, but since I’ve learned so much of what I know from people on the internet, I wanted to try and put my own knowledge back out there. Please enjoy my terrible smartphone-based AV setup.
That looks really cool, I haven’t seen much (any?) content about the process of forming carbon composite stuff before so now you’ve opened another can of worms for me. Very nice to see it all documented like this and I had a lot of fun with the concept
What’s the purpose of the braided red cable going down the middle by the way? Can’t tell if it’s just to bulk up on conductors or you’re using it for cell balancing or what
Thanks! Feel free to ask anything that you feel like I left out or unclear about composites work.
You totally got it, the red braid is just 22AWG silicone wire for balancing. One of the 8s packs is going to be under the front foot pad, but I want to put all of the charge ports in the same place for convenience, and so that I can hide them under a splash guard. So that balance lead is going to go inside the rear enclosure and get panel mounted from the inside. The white wires are shielded signal cables for the control system, and the black wires are a twisted pair of 8AWG silicone for the front pack main discharge.
weight isnt going to happen. Feel free to make it long enough for you to ‘luggage drag’ the board on its rear wheels.
Speed: 65kph controlled by arduino strain guage sensors is a death wish.
Are you going to write an advanced-ish communication protocol powered by arduino? (In case of: random reset, loss of power, loss of communication, bad values, loss of packets, interference, code bugs, etc)
the power ramping for weight distribution sensors usually needs to be really high ramping (very slow acceleration) so it doesn’t shoot away when you hit a bump, etc. You can’t stop safely or fast enough above 25-35kph.
a more realistic option would be to look into EUC’s as they are self balancing and completely controlled by lean. (with a gyro, and not weight sensors)
This can be a really fun project! But realistically, not practical at all for the type of board and rides you are looking for
I think you can set the ramp relatively high, if you set up PID control with a relatively long Integral time constant. It functions not quite as a delay, but that the response is proportional to how long the signal has been there as well as just how strong the current signal is. Sets a nice ramp up and avoids jumpiness
I’d agree with this but only in that it would be a death wish if I made it, not that it’s impossible for a competent person
I think this is risky, unless it’s just a temporary solution before you get everything in place. Overlapping and interlocking cables under high vibration are gonna rub off each other, and even though silicone is better it does eventually wear through and could short out the pack. Best practice is to keep balance wires
Non overlapping and flat
Insulated with fishpaper where they’re near each other
Retained in place so they stay that way
But yeah apologies if you know this and it’s just a temporary setup
I know, but it’s going to be close. The deck is going to be in the neighborhood of 2kg. And ergonomically, the big advantage of the narrow center section is that I can grab it with one hand and hoist it on to my shoulder, which is a great way for me to carry things.
Yep. I’m going to do everything I know how to eliminate failure modes, but this will still be more risky than a conventional remote and I’ll be wearing a full suit of armor for it.
-I’m going to be using old school 1-2ms PWM throttle signals, since that communication method has a very long track record for resisting interference in the RC world. Additionally, the failure mode for analog signals is gradual, compared to digital signals suddenly becoming gibberish.
-The plan for loss of signal is to make sure the ESC freewheels, then foot brake. And pull the loopkey if the throttle is stuck for some reason.
-The plan for loss of power is to figure out how to make a redundant power supply for the Arduino, but I’ll get to that after I know that it works and finish low-speed testing of the board.
-The control system requires 1 sensor set at each end, but I’m putting in groups of 3 so that a “voting” process can take place to reject a bad sensor value. This triple redundancy strategy is how aircraft control systems are set up.
-The plan for interference is to cover all components of the control system in grounded, shielded wires and aluminum foil Faraday cages. I will also be keeping the Arduino far away from as much of the power electronics as I can.
I will be running everything through a low pass filter to smooth out the influence of bumps, yes. And I will be operating on the ratio between the front and back sensor values instead of the difference between them, so that the throttle signal is independent of the changes in G loading that occur when you go over a bump.
However, that’s the only limiting factor on ramp time. My control logic will make it impossible for the board to generate enough thrust to tip me over in normal operation. And hopefully, my really wide stance will make that quite difficult in a fault condition.
Yeeeah might have been a mistake in hidsight. Routing the balance wires through that narrow space made them get crushed together anyway, so I figured I should at least keep it neat and uniform with a loose braid that prevented instances of a single wire crossing (they cross is a 3x3 grid, so its more distributed? Idk). I immobilized the braid every few centimeters with big globs of construction adhesive.
It’s a neat idea, but I won’t be using PID to control the board. Reason being that using a large I term like you suggest would create throttle inputs that persist after my weight distribution has changed, so for example it would allow a situation where the board is producing high thrust but my center of gravity is towards the rear. Not a good combo.
@mr.shiteside I opened up the tube core and redid the balance wires, what do you think? The original problem I was trying to solve with the braids was that a flat ribbon of disconnected wires will want to loosen up and get tangled as soon as it’s not strapped down to a flat surface. So my idea now was to embed the wires in glue, sandwich them between layers of paper to promote flatness and wrap them in fiberglass tape. To fit through tight spaces, I waited until the glue was halfway cured and still flexible, and just bent to the entire package into a U shape, so the wires still didn’t cross each other. The ribbon is still immobilized along its length with glue.