July 5, 2012

All-Terrain Electric Scooter

As seems to be the case with many of my projects, for this one, I acquired the parts first, and then designed a project to fit the parts I had. Here is the back story of how I got the parts, and what drove me to use them to build the vehicular monstrosity this post is about.

But first, a brief overview of the scooter's specs:

Motors: 3x CIM motors
Batteries: 8x Turnigy 5000 mAh 4s LiPo packs, 16s2p configuration (59.2V, 10 Ah)
Controller: Kelly KDS72200E, 72V, 120 A continuous, 200A peak
Wheels: 12.5" with knobby pneumatic tires
Deck: Hand laminated carbon fiber with polycarbonate top
Frame: Royce Union Transit kick scooter, with about 200% more aluminum added
Brake: Pedal actuated rear disk (sprocket) brake

Videos are at the bottom of the post.


At the beginning of this past school year, my school's robotics team decided to finally dismantle some old FRC robots that had been collecting dust for six or seven years. Since we now compete in VEX, not FIRST, we had no use for all the parts we stripped off the robots. Therefore, I generously relieved the team of some of these parts, including six Victor 884's, six CIM motors (4x 2.5", 2x 3"), 2 AndyMark Toughbox gearboxes, some #35 roller chain, and assortment of sprockets, some steel shafts and shaft collars, a 1/4" by 4" by 3' aluminum plate, 1" aluminum U channel, 1/8" polycarbonate sheet.... Anyways, while looking up the CIM motor curves, and trying to figure out by how much I could over-volt them, I ran into this, which at one point used a pair of CIM motors. That vehicle then led me to these. By an interesting coincidence, only a few weeks after I discovered those vehicles, I went to the Atlanta Mini Maker Faire, and guess what I saw there. After seeing the vehicles in action, I had to have one of my own, and thus my scooter was born.

I started on the actual construction of project part way through my second term senior year, and have continued it into the summer. It is still a work in progress, and updates will be posted HERE. Also, I plan on taking it up to Boston with me for school next year, where it will be my first (and probably not last) entry to the MIT branch of the "Collegiate Silly Vehicle League."

The scooter started out as my old kick scooter, a Royce Union Transit, which was a small step above the classic Razor scooters in terms of quality - for instance, it has front and rear wheel suspension that uses actual springs and shock absorbers rather than rubber blocks, and it has a more elegant folding mechanism.


Front suspension. The actual spring and shock absorber are hidden withing the steering column:

Rear suspension:

For use on softer and less regular terrains than asphalt, the original wheels needed to go. I got new ones from Norther Tool. They sell a version with a 60t sprocket included, but I managed to scavenge a 60t sprocket with the same bolt hole pattern from school. The best part of these wheels is that I can later upgrade the tires to these epic snowblower tires if I want to use the scooter in snow:

To accommodate the larger wheels, I built a completely new fork and rear suspension assembly out of aluminum. I found some cheap mountain bike shock absorbers on ebay, and bought two with 1500 lb/in springs. Here they are next to the scooter's original rear shock absorber:

I made the rear swingarm of 1/4" x 4", 1/4" by 2", 1" angle, and 1" U channel aluminum, plus an assortment of nuts and bolts:

Yes, I know one side is longer than the other one.

For the new fork, I started out by tearing apart the fork from an old mountain bike. Fortunately, the scooter used a 1" threaded headset, which I was easily able to replace with the mountain bike's 1" threadless headset.

Using the MTB fork's springs, damper, some steel tube and solid rod, and U aluminum, I built two new shock absorbers:

These were then bolted to some more aluminum and the fork crown, to make a leading link suspension arm:

I cut dropouts in the aluminum, so the wheel could be removed easily. I cut horizontal dropouts in the rear suspension arm:

I sued a 1/2" threaded rod for the axle, but the I.D. of the wheel's bearings was 5/8". To make up for the difference, I filed down some 1/2" nuts, so the fit inside the wheel bearings:

To test handling, I pretended it was still a normal kick scooter, and pushed it around my driveway a bit:

It handled okay, but not great, due to the forward offset of the front wheel. This made it impossible to lean while turning, because if you lean into a turn, the wheel naturally wants to swivel away from the turning direction. Hardly ideal. Additionally, at higher speeds I realized that there would be the risk of the front wheel "castering" - trying to swivel around 180 degrees.

In the above pictures, you may notice that one side of the rear suspension arm has been bent to accommodate the sprocket on the rear wheel. I managed to bend the massive piece of aluminum by heating both sides of the bend lines simultaneously with two propane torches, while the piece was clamped in a vice, and applying force to the metal via clamps clamped to it.

To improve the steering geometry, I rebuilt a new fork out of the metal used in the old one. It does raise the height of the front of the scooter, so the platform is no longer parallel to the ground. Notice how if you draw a line down the center of the steering column, it intersects the ground where the wheel touches the ground. This means that the wheel will not have a tendency to either aim straight ahead or be flipped around:

This scooter is powered by 3 CIM motors. To interface all three of these motors with a single drive sprocket and reduce the speed of the drive sprocket, I built a new gearbox out of the two Toughbox's I scavenged. I removed one stage from each gearbox, so that the input:output ratio was 14:50, rather than (14:50)^2. I chopped up the gearboxes and aluminum-zinc brazed the pieces back together, to get a 3 motor 1 stage gearbox:

I had to order the third pinion gear (not pictured) from AndyMark, the manufacturer of these particular gearboxes. Unfortunately, the newer version of these gears do not have a set screw, so I had to lock it to the shaft with a layer of CA glue to keep it from sliding off while riding.

I built a structure to attach the gearbox to the frame out of some angle aluminum and 1/4" scrap:

Because of the shape of the rear swingarm, there is not a direct line from the sprocket on the gearbox to the sprocket on the wheel. To address this, I added an idler attached to the swingarm. Since the sprocket on the gearbox is not located at the pivot point of the rear swingarm, as the suspension is compressed, the effective chain length increases. To take up the slack in the chain, I added a chain tensioner:

Motors and drivechains are nice, but being able to stop is nice too. Since the rotors of disk brakes are simply large rotating metal disks bolted to a wheel, I decided to simply use the wheel's drive sprocket as the rotor for a disk brake. I built a caliper to grip the sprocket out of an aluminum block that used to support the scooter's original suspension, some aluminum U channel, two road bike brake pads, a spring, and some bolts.

The right brake pad is fixed to the rod that passes through the caliper, springs, and aluminum frame of the suspension, and the pad and rod can slide through all those bits. Since the spring is split in the middle, the brake is applied, and the brake cable pulls the two halves of the caliper towards eachother, both halves move equally in opposite directions, so that they both move towards the sprocket.

Normally on kick scooters, you brake by pushing down a lever above the back wheel with your foot. THe lever rubs against the wheel, slowing the scooter down. I wanted the disk brake to be actuated the same way, so I built a brake pedal out of some 1" U and angle aluminum. Two pivoting segments are attached to the brake pedal, which, when fixed to the scooter frame, form a kite shape with hinged joints. When pressure is applied to the pedal, the kite is deformed, which lengthens its long diagonal. Since the brake cable is attached across the long diagonal, the cable's housing is pushed up the cable when the pedal is actuated:

The pedal fits between the two shock absorbers:

Here's what the mechanism looks like with the brake lever released:

And pushed:

To control the motors, I originally planned to use the Victor 884's I got from school, which are rated for 40 amps continuously and about 16V. I decided, however, that I wanted to try and get some more power out of the motors, and run them at around 20V, rather than the 12V they were rated for. At that voltage, and the extra current that would result, the 884's would probably have quickly released their magic smoke, so I bought a proper vehicle oriented motor controller. So I could use a lower current controller, I wired all three motors in series, for use with a 60 volt battery pack. It's easier to get a high voltage low current controller than a low voltage high current controller. I got a Kelly KDS72200E controller, which is rated for 120A continuous and 200A peak at up to 72V.

Here's one of the Victors:

And the Kelly, next to a DVD for size reference:

Along with the controller, I got a hall effect twist grip throttle, which fit perfectly onto the mountain bike handlebars I fitted to the steering column.

Getting the mountain bike stem to fit the scooter's steering column involved a lot of shimming with aluminum flashing:

One problem with the original frame of the scooter is that its tiny deck is impractical for anything other than people with very small feet riding on very smooth surfaces. To fix this, I built a brand new deck, that is about three times the width of the original. I could have just ordered some 1/8" aluminum from McMaster, and with a little jigsawing and filing it would have made a great deck. I didn't think that approach was very interesting, though, so I took a much longer and more tedious route. Using the carbon fiber tow and resin left over from my bamboo bicycle, I made a sheet of carbon fiber by wrapping the tow around a wooden frame, soaking it in resin, and compressing it between two pieces of plywood. Six times. Since carbon fiber is strong, but not particularly impact or abrasion resistant, I bonded the carbon fiber deck to a layer of 1/8" polycarbonate.

Here's the roughly made sheet of carbon fiber, along with the roll of tow and some resin. I still probably have a few kilometers of tow left, of the 5 km I originally got:

Cut to shape:

Plus the polycarbonate. The cutout is for the folding mechanism:

The new deck was screwed to the original one with some countersunk stainless steel bolts:

Here is the entire deck pictured, along with the Hella master power switch that I used for, well, the master power switch.

To maximize battery space, I mounted the motor controller to the gearbox, via some aluminum angle:

I wired the motors in series using some large bullet connectors:

The main power terminals were connected to the controller parallel to its top, to conserve space:

I attached a power fuse to the deck with some more aluminum:

While testing the suspension, I noticed that the 8mm stainless rod that I used as the pivot point for the rear suspension arm flexed significantly under load (e.g. me jumping up and down on it). To fix this, I drilled out all the 8mm holes to 1/2", and replaced the rod with a 1/2" stainless rod scavenged from an old scanner:

For batteries, I ordered 9 4S 5 Ah hardcase LiPo packs from HobbyKing, which happen to be the best value in terms of Wh/$ you can find on the site. The 9th battery was an extra, since HK's quality control is not highly regarded. I used 8 of them to make a 16S2P pack, at a nominal 59.2V 10 Ah pack. That's around 600 watt hours. My poor 50 watt charger will need to be upgraded soon:

I paired off the battery packs, and wired the pairs in parallel, including the balance connectors:

I made a battery case out of some more aluminum angle, which I brazed together:

The front cover of the case can be removed by thumbscrews, for easy access to the charging connectors:

Not shown is a polycarbonate bottom to the case, which holds the batteries up. Eventually, I'll plate the entire case in some sort of plastic.


When riding the scooter over bumps, I found that the chain frequently came off the drive sprocket. To fix this, I made a chain guard around the drive sprocket out of some angle aluminum. I cut out triangular wedges of the metal, so that I could bend it, and then brazed over the seams:

The chain guard worked great, until I broke it. I went over a massive bump going down a trail, and the chain pushed up against it, breaking a brazed joint:

I've since re-brazed the joints, and will be securing the chain guard at the other end as well.

Here are some final pictures of the scooter:

To make the scooter more convenient to charge, I made some modifications to my charger. I put the charger, along with the 12V power supply from a 2006 iMac desktop, into the shell of an old ATX power supply, so that I would not have to carry aroudn a separate pwer supply along with the charger. Also, I made a charging harness that will let me charge all the battery packs in parallel, for convenience.

The charger is held down by some clips that were meant to secure wires inside a PC case.

Finally, here are some videos of riding the scooter:

This first one is from before I added the chain guard. I went pretty light on the throttle for most of this video, and the only point at which I maxed it out is right at the end.

Here's a video taken with a camera attached just below the handlebars. This is where I broke the chain guard, and later on the master link of the chain. Average speed for the run was 15 mph, with a top speed of 24 mph. The scooter should be able to hit a little bit over that, give a long flat:

Test with the camera attached to my helmet.  The video is much smoother than the previous one as a result, even though it covers the same course.  Also, in this run I broke even more things.  I toasted two of the CIM motors.  Honestly, I kind of deserved it for trying to go full throttle up a long steep hill... On taking them apart, one has some blackened windings, and while it spins, it does so slowly and unevenly.  The other looks okay, but exhibits the same behavior.  Also, I'm redoing the idler arrangement to keep the chain on track:


  1. This is awesome. Bring it to the Invention Studio this weekend! I & the GT scooter guys will be around!

    1. Thanks! I wish I'd been able to.

    2. Are you starting at MIT this fall?

    3. Oh snap, that means we'll be travelling in opposite directions by the time Dragon*Con and Robot Battles roll around. I was faced with this same situation my freshman year in 2007. When will you be on campus, approximately? Maybe we can meet up beforehand.

    4. I get to campus on the 21st. I'm going to have to ship the scooter up, so I'l probably get it a week or so later.

  2. Charles wants to suck you into a world of mikucopters and such.

  3. Really nice build, congrats!

  4. Absolutely amazing! Your post inspired me so much!

  5. Really cool! By the way, the motors connected together directly with the gears. There is not so much loss because of the shaft speed difference? I have built a larger 3 wheel car powered by 8 cordless drills, and it was a huge problem for me. I used HF1616 one-way bearings, so the torque distributed between the motors better. Pics about my work: (http://team.giganet.hu/)

    1. I don't have any current draw figures, so I don't actually know if the speed differences in the motors are causing any problems. However, these motors are routinely used in two-motor-per-gearbox drivetrains (such as in every FIRST robot ever), so I don't think there is too much loss.

  6. would wiring the balance connectors in parallel not be detrimental to the life of the batteries?
    otherwise awesome build! cant wait to do one myself one day.. what sort of mileage do you get from a full charge?

    1. It shouldn't hurt the batteries. What wiring the balance connectors does is connect the corresponding cells in each of the two packs in parallel with each other, to make the packs easier to balance. During normal use, though, there won't be any current flowing through the combined balance connector, because the voltage difference between corresponding cells in each of the packs is zero.

      I don't know what the range is, and I don't have a wattmeter to measure it's power usage. I'd estimate a range of around 10 miles, give or take a few, although it would vary significantly based on terrain and riding style.

  7. Nice! Excellent work.

  8. damn that's awesome

  9. Guy, very nice job! Recently I saw your project in the hack a day site. Keep working. Your friend from Brazil.

    "When the server is ready, the service appears." (Emmanuel)

  10. When fully charged, what voltage did a single battery pack measure?

    1. I should have figured it out just by the 4.2v lipo * 4 in series, but thanks for the confirmation.

  11. Hope you dont mind me asking , as this seems bit of an odd question, but what would these CIM motors be like in a freewheeling capacity if you wanted to have the option of using your scooter manually in addition to the motor as well , would there be any long or short term drawbacks ?

    thanks - wonderful build


  12. Hey, I have a quick question about your controller. Did you need a DC-DC converter to drop the voltage down to run the internals? It says 8-30v input through the pins, and I don't want to hook mine up to 72v and explode the thing. I see you're wired up for around 60v, did you run 60v to the pin to turn it on?

    1. The Kelly KDS72200E I used is fine with up to 72 volt input, including for the controller logic. It works fine without a step down converter.

  13. So impressed ! After watching yr video, I want to build ev scooter myself .
    Can You tell me how powerful are this CMI motors ?

  14. Looks really nice, and I can feel the effort done. I am trying to make my own electric scooter, I made one couple of years back, started with a stand up scooter 24 Volt 500 watt motor with no suspension, then added a seat and front suspension, but now I am making some sketches and designs for a new 2000 Watt scooter. It is great to see people around sharing the same passion.. Good luck!