Do-it-yourself Scooter Project:

I was prodded to build version 2 when Chris Johnson brought over his two Segways and we went for a ride. (I had never ridden one any distance before.) The Segway was faster and smoother than my machine. The gauntlet was thrown! I had to surpass it.

As usual, I followed my personal rule for building projects: only use parts that can be ordered over the internet without talking to anyone.

Wheels

Some punk riding it

The first change was larger, lighter, smoother wheels. The original wheels were 14" diameter foam-filled trailer tires. They had fairly high rolling resistance, and one of them had started to make a skrupp-skrupp sound every time it went around where the foam was peeling away from the rubber inside.

I decided to change to bicycle tires. They are lighter and have less rolling resistance than the trailer tires, and the narrower width helps it fit through doorways. To do this, I bought a pair of 20 inch diameter bicycle wheels (rim, hub and spokes) and machined a new hub to bolt onto the output shafts of my motors. I started with a 3.5" diameter by 3" long cylinder of steel, machined the mounting faces, removed 80% of the material to reduce weight, and then drilled 36 holes around the rim to fit the spokes. The diameter of the hub is larger than the original hub to get maximum torsional stiffness of the wheel. After the tedious job of stringing, tightening, and adjusting the spokes for even tension and no wheel wobble, I put on 100 PSI snake belly tires.
The 20" wheels propel the scooter 43% faster for a given motor speed than the 14" wheels and give 3" more ground clearance. They also look much better, especially with the all-black smooth tires. The narrower wheels reduce the width by about an inch, making it much easier to get through doorways.

Segway Polo

I may make a second set of wheels and outfit them with knobby mountain bike tires, for playing Segway Polo on grass.

Electronics

I also replaced all the electronics. Version 1's electronics were chosen for maximum convenience. Version 2's electronics are designed for performance. I replaced the dual-channel RoboteQ motor controller with two OSMC controllers. While the RoboteQ is fine for many applications, its use of a 9600 baud serial link for control added too much delay to the feedback loop for good performance. The OSMC controllers take a PWM signal directly from the microcontroller. I also get more precision in the PWM control: 9 bits instead of 7.

Most important of all is how the OSMC lets me precisely control motor voltage, regardless of the condition of the batteries. In the RoboteQ system I couldn't measure battery voltage fast enough to include in the feedback loop, so the gain in the balance feedback loop depended on the resistance in the batteries, which increases as they run down. Fresh off the charger the gain was so high it would start oscillating if you didn't hold the handlebar firmly, and after 3 miles it would start feeling mushy and unstable at high speeds.

In the new system an analog-digital converter in the microcontroller measures battery voltage 2000 times per second, so I can adjust the PWM controller to get a desired motor voltage. It turns out that most of the "clunk" the old one produced as the motors changed direction was due to the electronics, not the gearboxes. The new version feels perfectly smooth.

I also changed the gyro and accelerometer. I had been using a Tokin gyro that came on a board from RotoMotion. It had some performance problems: it would occasionally glitch (causing the scooter to jump), and it was susceptible to vibration (causing the tilt angle to wander at some speeds.) I changed to the CRS03-02 gyro from Silicon Sensing Systems. Noise is lower and it seems completely immune to vibration. I'm also using the ADXL105 accelerometer instead of the ADXL102. The difference is that it has a higher saturation threshold (5 Gs of acceleration instead of 2) so it's less likely to saturate on bumpy roads.

The new gyro and electronics have a much faster response, allowing tighter control of balance. While version 1 required a firm hand on the handlebar at all times, the new version can be controlled entirely with the feet, even at high speed. You can lean it up against a wall and it will remain almost motionless. (The old one had a tendency to start whacking the wall.) The new gyro also has less drift, so the handle angle of the scooter doesn't wander as much.

Batteries

I also installed better batteries. Version 1 started with 120 cells worth of cheap off-brand NiMH batteries. One bank of 30 caught fire because the fragile plastic shell wore through, and one seems to have a dead cell that won't accept a charge. Version 2 uses 60 HHR-6500 D cells from Panasonic. Digikey part P019-ND. These are high-quality cells with low internal resistance (typically 2 milliohms -- theoretical short circuit current = 600 amps!) I wired them in two parallel strings of 30 for a nominal voltage of 36 volts, able to deliver 200 amps at 30 volts. This works out to 8 horsepower peak. Those pedestrians better get out of my way!

In the original scooter I used a diode bridge to parallel the batteries while allowing some voltage difference between them. In the new version I use a relay to connect both batteries simultaneously when the power is on. This reduces power supply impedance and gives smoother control. Also, the relay disconnects the batteries from each other during charging so I can charge the packs separately. There are two charging jacks on the console.

Remote Control

The scooter is now Bluetooth-equipped. Using a Bluetooth wireless connection, I can can change parameters, download logs, and even drive it using my laptop. I got a pair of serial port extenders from Free 2 Move which look like DB-9 connectors with no wire coming out of them. Within 100 meters range, they provide a transparent serial port connection. Then I wrote a GTK application that lets you drive it around without a rider. I can now leave my scooter parked somewhere and use my laptop to have it come and get me, thus saving valuable steps. All I need now is to make it run on one of those Linux wristwatches, and I'd have the completeJames Bond remote control system. (Or at least a yuppie techno-geek version of it.)

Battery Monitoring

The old version would let you know that the batteries were getting low when the balance feedback would get mushy. The new version measures and compensates for battery voltage and resistance, which is great except that the rider has no way of knowing when the batteries might let him down. So version 2 monitors battery state by doing a least-squares fit between battery voltage and current draw. It calculates the maximum speed at which it would have enough torque to balance properly, and limits the rider to that speed by tilting back. It also has a beeper which sounds when the motor drives are nearly maxed out, or the battery voltage drops below a threshold, or the battery current exceeds the battery fuse rating, or a few other exceptional conditions.

Steering

The Segway uses a twist grip on the left handlebar to control steering. My version 0 used a potentiometer conveninently located where my pants would brush against it and send me into a spin. Version 1 used a pressure sensitive touch pad with left and right sides. This was all right, but required careful finger positioning and wasn't very intuitive for others to learn. Version 2 uses handlebars which you twist. The handlebars don't actually move when you twist them; they sense the torque applied using four strain gauges which measure the slight change in resistance of a thin wire bonded to a piece of metal when the metal bends.

Just to make sure it can outperform Segways in every way, I increased the steering control so it can spin in place much faster than a Segway. It goes around 1 revolution per second. It's pretty terrifying to be on, actually. I'm going to have to add a mode switch: normal (for me), beginner (for letting other people try it) and yee-haw (for proving that it can beat Segways.)

Chassis

To protect the batteries against damage, I added a stainless steel plate around the bottom. It also looks better than the mishmash of wires, plastic sheet, and fiber tape that held the old undercarriage together. The steering column, formerly 1.5" square aluminum extrusion, is now 2" square tubing which is lighter and stiffer, and looks cleaner. Also, all the wires and switches can be mounted internally sticking out the side.

To improve chassis stiffness, some metal bars are bolted onto the front and back of the main plate. It still isn't a very well engineered chassis, but it handle the usual urban terrain (jumping off curbs) pretty well.

Controls

Styling

Segway owners report being yelled at by people in pickup trucks (and I've gotten this a number of times,) usually something to the effect of "Too lazy to walk, you f***ing homo?" Objectively, it's less lazy to ride the scooter than drive a car. It also uses less fuel, and pollutes the air less. The people yelling from inside their truck aren't walking either. At least I'm standing up. But, like, whatever. I dismiss this part of their criticism.

However, I think something can be done about the "homo" part (not that there's anything wrong with that.) It reflects the not-so-masculine soft plastic styling of the Segway. I'd like to see if I can tweak the styling of mine so I don't get yelled at as much. At least mine is metal, but it still looks spindly. I'd like to give it a muscular look with lots of chrome tubing and polished metal surfaces.

The new wheels are a step in the right direction. After I put them on, the first comment I got on the street was from a hip-hop-looking guy in a Mustang who pulled up and told me, "Dude! That's tight!" Later at the office, I checked Urban Dictionary and found that "tight" means "stylish, cool, having everything together," "dude" means "male person," and "that's" means "that is." So he liked it!

By the way, it's not just any new form of motorized transportation that attracts hostility. I often ride my electric unicycle around, and I've never gotten a negative comment.

Futures

While the new chassis is better, it still isn't great. I'd like to propose an offer to someone with good metal fabrication skills. Make two very cool looking frames, one for me and one for you, and I'll provide the electronics for both. We'll each pay for our own motors and batteries. Or, if the producers of Monster Garage are reading this, I think we could put together a pretty amusing monster-Segway project. I'd picture a big beast capable of 30 MPH, with lots of chrome and bad-ass styling. Maybe some of those spikes that pop out of the wheel hubs, like in Ben Hur. If anyone insults Jesse James's masculinity while he's riding it, he gets to send it off a cliff. (I think it should be able to stay upright all the way down. We could mount a wireless camera right on it.)

How fast can a scooter like this go? There's no fundamental limit. A bigger, heavier one could go highway speed. The danger, of course, is that if something fails it'd be a serious accident. Under 10 mph the rider can probably land on his feet if it falls over, but at higher speed you'd want more protection. Also, the differential steering might be iffy at high speed, and hitting a pothole without any suspension might be hazardous. Someone else with no fear of death should experiment with this.