Show HN: I Designed and Built an eBike (endless-sphere.com)
This post shows and explains the design of the eBike I built myself. I decided to post it on this specific forum because this is where it all started, by stumbling on another post, as mentioned in my entry.
140 comments
[ 3.1 ms ] story [ 197 ms ] threadOtherwise looks good!
When I got a 3D printer it was amazing to be able to make my own designs, but I wish I had the capability to work with more durable materials.
Other easy things to get a good result is use nuts or inserts instead of threaded holes when you can, and try to keep printed plastic parts in compression, not tension.
I think not enough people use sheet metal. Doing as you described plus sheet metal to put some material on the outside of the cross section not only looks professional, it adds a lot of stiffness. And for aluminum sheet 0.5mm or 1.0mm thick you can literally cut it with a knife and straight edge, and use a couple of pieces of wood or even books to make surprisingly nice bends. For thicker or more complex parts, it’s pretty easy/cheap to make or buy a decent brake, and also pretty cheap to get parts made from sites like I described above.
I also think the carbon filaments are BS. I've seen several tests like this one: https://www.matterhackers.com/news/filament-strength-testing
The carbon fibre nylon is worse than the plain nylon in terms of strength. Which makes sense to me. If you grind up carbon and put that into your filament why would that make it stronger? Maybe for compression forces?
Thanks!
3d printing is here to stay, but load bearing parts are problematic. I had to change some of them to metal.
[1] https://www.schaeffer-ag.de/
(Sorry if that's answered in your linked post, I skimmed it perhaps a little too quickly...)
(I'd be able to make those, I would probably need to do it in a few passes though.)
I suppose a throttle is a much more direct connection between your mind and the motor control, unfortunately they are illegal in my country.
[1] https://hubsink.com/products/high-performance-domino-twist-g...
Mud and water could be problematic.
* A VESC based ESC instead of the HobbyWing one will give a lot more control and telemetry, including a phone app and detailed current use logging. I know from RC planes that HobbyWing ESCs are very high quality, but VESC is specifically written for drive systems and has a lot of features.
* LiIon or LiFePO4 round cells are safer and more energy dense than LiPo pouch cells, at the expense of current sink capacity you don't need. They're also cheaper.
The ECU selection was primarily budget driven. I need to have a look at those VESC.
Battery selection was again budget driven. The also hat to have certain dimensions to fit inside the kit.
I think we're about to see a lot of innovation in scooter- and bike-adjacent mobility things - I don't mean the motors, but the frames etc... eg we might see more cargo e-tricycles?
[1] https://www.nicolai-bicycles.com/
Any estimates of your expected range?
[1] https://www.schaeffer-ag.de/
I could cut these out on the plasma table in my garage for free, but that looks like at least 150 bucks worth of metal.
The motor sticks out uncomfortably close to the pedals and particularly on a MTB I could see a rider accidentally stepping on it...and it looks like an outrunner, so that would also be quite bad, as the foot would get launched backwards pretty violently.
This is a lot of effort compared to just bolting on a Baofang mid-drive motor. The components are also placed such that unless you have a pretty good front fender, they'll be covered in crap in no time.
There is also another detail, that you need to know, in order to understand how it all works.
Trial bikes (not trail bikes) usually have a special kind of drivetrain, where the freewheel is mounted on the crank and not the rear wheel. This allows for a very small rear wheel sprocket and extremely high gear ratios.
So to make this work, I used a trial bike crank [1] on which I'm mounting a special purpose freewheel [2]. The chainrings are mounted to the freewheel using this adapter [3].
[1] https://sickbikeparts.com/cranks-freewheel-isis-crank-set/
[2] https://sickbikeparts.com/front-freewheel-heavy-duty/
[3] https://sickbikeparts.com/freewheel-spider-4-arm-104-mm/
I'm extremely surprised to hear that soldering the XT90s was hard. Even a $20 Pinecil (well, maybe I shouldn't say "even", it's a fantastic iron) can handle those no problem.
Also, what cost 1200 EUR? That motor costs 30 EUR, an ESC another 40, and the batteries maybe 80 EUR for Li-Ions (less weight). Were the CNCed plates that expensive?
EDIT: Added the praise I meant but realized I forgot to actually write.
Once the wire goes in, you remove the iron, hold the wire there for a second or two until the solder solidifies, and you're done. I used 12 AWG but it should be pretty close.
EDIT: I just did it again, make sure you hold both the wire and the XT90 firmly with helping hands or something similar.
You can hack it by just turning the temperature way up, but there are a number of drawbacks to doing so. It's better to use the right tip.
But if you take that perfectly good soldering iron and apply it to AWG 10 wire, you're going to have a bad time - because the wire will conduct heat away almost as fast as you can apply it.
Personally own the TS100 and absolutely adore it for mostly building keyboards.
Yes you are absolutely spoiled by both of these. Before them were super terrible irons or the expensive brick sized ones with minimal optionality. Having come from the 'dark ages' it's night and day.
The debug board, which I don’t have is a great example of using usbc for development and debug (as opposed to test contacts inside the device).
It does need a version 2 though as the tip holders are not great.
ps. thanks for the tinning walkthru below
It's great to see good builds like this - I keep saying "one of these days". In my neck of the woods we have a pretty resource in the form of GRIN[0]. I believe Justin is a longtime participant on endless-sphere.
My latest build (on a Walmart special) was a BBS02 that probably rings in around the same cost but without the openness of this build.
[0] https://ebikes.ca/
> Being on a tight budget, I was able to procure/purchase all parts month by month and finish the build within a year. This includes all parts for the kit, as well as building the bike itself, which I did not have prior to starting the project. You can see the final build below.
Mine will pull almost 750W going uphill on a freshly charged battery.
It has a very prominent "200W" label on the outside of the battery holder.
If I rode it like an asshole, doing 45kmh on sidewalks, I doubt my "200W" label would do me any good. But I've had it eyeballed by police when they're doing their semi-regular "fine everybody without a helmet or a working bell" crackdowns on CBD commuters and have stopped me to check my bell works, and never had the power questioned.
The high-end E-Bike I rode briefly did a marvellous job of detecting road conditions and maintaining semi-constant speed. The ones I use on the street are more crude, they use pedal effort as a primary guide to intent but back off aggressively, I guess to limit risk to pedestrians. I am pretty sure off-road wouldn't want that. (not a big e-Bike rider, just fascinated)
If it walks and talks like a bike, and only goes as fast as the average cyclist, it is probably a bike. But if it has a throttle and goes faster than your average cyclist, it's probably closer to a moped at which point you need a license, different kind of helmet, turn signals, registration, so on and so forth.
Older bikes used the circumference of the wheel and a count of rotations to calculate the speed, so people could de-limit them by changing the stored circumference constant in the eeprom. Modern e-bikes use a combination of sensors at various locations to detect motion, acceleration, pedal force and tilt angles to decide how to apply the power but I think even with those added sensors, the speed is still detected via the wheel rotation sensor.
Seems some people don't like the regulations as the parent comment got downvoted, rest assured I did not single handedly write the global regulations, and I don't like the speed limiter either.
I also love that it's internally geared, I've always felt this made way more sense for mountain bikes anyway given the weather protection. Good stuff!
It appears this e-bike only has a single-speed chain drive, and only a throttle for adjusting motor speed.
Very useful site with more similar examples, but just in case you want to save them some bandwidth in load ( it contains multiple images of the bike):
https://archive.is/QMx56
>Hey, congratulations. Traffic is flooding in from ycombinator to this thread and we're getting slashdotted :shock: I just now had to upgrade the server to a huge one temporarily.
Are you important on there or something?
A 2400W motor is very powerful for a bike. For Americans, 750W is one horsepower so that's over 3 HP which should get the bike to ~40mph or 70km/h.
Before I got the bike I wanted the most powerful one available but I realized after getting one you don't really need that. And it's actually kinda dangerous, even for a experienced cyclist who's used to speed.
Even the standard tier batteries are more than enough for any urban biking in terms of distance. I rarely use even 20% at 48V, 10.5A.
Cargo e-bikes tend to be at or less than around 1kW. 750W is still considered fairly powerful, and 500W is much more typical and still plenty enough for even a cargo bike to get up a hill, with appropriate gearing. 250W is about the standard for low-end e-bikes.
2.4KW is gross overkill, especially if you're not doing direct drive. More watts just means you can go up the hill faster. This assist motor is driving the wheel through the rear derailleur so it potentially has a pretty wide range of gearing.
Tour de France cyclists racing up mountains generally sustain around 400W, which is enough to go 25+km/h for hills that aren't too steep.
I have a 2500W Luna Cycles fat bike (52V BBSHD mid drive motor and 12 speed drivetrain). It climbs hills effortlessly. It powers through sand and snow. It's been a lifesaver on uphill technical singletrack when I've had to stop and a blip of the throttle gets me going when I wouldn't have been able to just with pedal power.
I have no idea what the top speed of the bike is and don't care. 90% of the time I'm doing under 20 mph and almost all of the remaining time I'm doing 20-30 mph. I've ridden 750W hub drive bikes and they're not even remotely comparable. This is what people who cling to the antiquated eBike class ratings and watt limits don't understand. It isn't about top speed (I think doing >35 mph on a bicycle is crazy) it's about torque application in circumstances that demand it. I'm considering upgrading it to a 72V battery and 4000W controller.
So far the simplest assist I can think of would be a front wheel hub motor and a switch that turns on a constant amount of torque. This would still let me control my speed by how hard I pedal. Of course I'm scratching my head over how I would control it, since it would probably not be an off the shelf controller. But I'm also fascinated by the electronics, and having to design my own controller would be a benefit, not a liability. I also have an obsession with knowing how things work, that I have to contend with.
Of course all of this is until I think of something better.
On the controls side of thing, I have the ECU and an Arduino Nano. The Arduino uses the BEC port of the ECU as a voltage supply to run a simple routine translating the analog output (potentiometer) from the throttle handle to a PWM signal that the ECU can understand.
The way that the assistance work is that while you are pedaling, you can turn the throttle and adjust until you get the assistance you want - pretty simple. After a while it becomes natural.
The simplest: sure. However if you're used to cycling chances are you're not going to like that front wheel hub all too much. It feels weird overall (might depend on geomerty though) and has the tendency to slip on upwards sloped unless you do an effort of getting weight over the front. For my personal taste, hub in the backwheel feels the most natural.
noise: quite a bit
also bear in mind that just 100w added to someone which is already sportive is quite sufficient for commuting. going harder would put excessive wear on bike chains that for me are already excessively thin.
now, tires and brakes. did you use bicyle grade tires that are way too soft or some ebike rated tire that are much harder ? also brake, olease check your brake, you don’t need to be electric for make the brake suffer, so that you need bigger discs, ceramic pads…
After building this, I'm dreaming of building something very elegant - a 50W to 100W torque generator that is tiny in size and very lightweight, and it fits within the hub of the wheel. Maybe not using batteries at all, but supercapacitors the are charged when you brake and discharged when you accelerate.
Given the KV rating of the motor and your max voltage you can calculate the max rotational speed of the motor = KV*V. From there it is just a matter of multiplying by all gear ratios along the drive train to calculate the rotational speed of the rear wheel. Knowing the circumference of the 26" wheel gives you the max speed.