NetBurner Community Forum

https://www.youtube.com/watch?v=CLjzHqDZH6k

https://www.youtube.com/watch?v=yLZIRqlFp3Y

I was originally going to make a quadrotor drone - but then the FAA decided to require registration, I have a commercial pilots license, and so I didn't want to risk a problem with the feds and put that license risk if the drone broke a window. So I decided that a self balancing scooter would be the way to go. The gyro is an Invensense MPU-6500 available for cheap on Ebay and the processor is a Netburner MOD5213 module.

The processor controls off the shelf Mystery 30A ESCs and RHD 2212 brushless quadcopter motors driving 10 x 4.5 props. Right now I have it connected to 12V power supply – but it will use a lithium battery in practice. The rotor mounts tilt on bearings and, after I mount a 1/4 scale servo, it will capable of "tiltrotor, forward and reverse propulsion. The idea is to eventually get to a man carrying, recumbent, highly aerodynamic (think competition glider fuselage) that will travel 120 miles on a 5KWH battery. I figure that people will need to retain their large gasoline fueled vehicles for suitable missions so a second vehicle for efficient individual commuting needs to be small and cheap. It can't be cheap if it is a full size car with a $50,000 battery. My target is 746 watts, or one HP, at 60MPH using 2 x 2000 watt BLDC motors for balance (especially when stopped) and propulsion. An athlete can produce about 500 watts and a recumbent bicycle with a fairing recently set an 85MPH speed record. All I want is 60 MPH.

If you're pursuing this project as an exploration of what is possible, ignore the following post and have fun, because this does sound fantastic!

A short bit of commentary regarding balance, propulsion efficiencies, and road travel. If the vehicle you're envisioning is going to be ground based, then I really don't see why you would choose to use a fluid thrust mechanism (aka propellors) for propulsion. They're much less efficient than a direct motor coupling to a drive wheel on the pavement. On top of that, you can use regenerative braking with a drive wheel, whereas the efficiency precludes much effect using propellors.

Second, if the vehicle is to be manned, then there's not really much need for self-balancing. Perhaps it might be nice for assistance at low speed, but the required force to counteract the weight of a person and machine capable of transporting them gets tricky in a hurry.

Finally, the big one: road travel. I will preface this by saying, I'm a motorcyclist. I've been riding for close to five years now. There is a difference between whether a vehicle is capable of a certain speed and whether it is safe to travel with traffic at that speed. Let's go with a generous assumption that a 0-60 time of 15 seconds is acceptable. Some quick math puts the time average force required for a combined weight of 250 lbs (150lb rider, 50lbs of batteries, 50 lbs of everything else) at 202.7 Newtons. With a 9 speed transmission, the best timing I could come up with for a pair of 2kW motors is 33 seconds*. This is far too slow for comfort. The little Honda Grom, with it's 125cc engine putting out 9 HP is just barely adequate for some people to get on two lane highways. Yes, air resistance plays a role in this, but so does acceleration.

The other thing about weight is handling in the wind. The lighter the vehicle the more it tends to blow around. With only two wheels, this effect gets much more pronounced, as the wind is applying a torque to roll the vehicle and induce a turn on top of the normal direct lateral force. Having purposefully ridden in a major windstorm, I can attest to being blown across a full lane (line-to-line) before stabilizing by a 50mph crosswind gust amidst steady 20mph crosswinds. And that was on a 400lb KLR650.

At any rate, good luck, and thanks for Nerd Sniping me on a Saturday. I just spent the last several hours going through brushless motors, trig functions, and other stuff.

-Dan

============

*I made some assumptions here; I wrote a small python script to numerically integrate the torque curve of http://www.cpmotion.com/download/telech ... rev1_1.pdf to run my calculations. I extracted a few datapoints and then ran a exponential decay fit against them. Being generous, I did not limit the stall torque to 10.5Nm, but rather used the extrapolated torque directly from the regression (such that torques reported below 2000RPM are Higher than the actual available torque).
I then set a maximum speed of 65 MPH to allow for variation without blowing the engine (going downhill, etc.) and determined the final effective drive radius. From that I simply selected a set of gear ratios and ran the integration for various total weights.

Excellent points. So I have a new idea. Ebay treadmill motor and Newtonian mechanics applied to a bicycle training device. Treadmill motors are sold cheap (I just bought one) and are characterized by large rotating mass and a flywheel. Imagine a bicycle with the treadmill motor and flywheel slung below and attached to the bar running from the steering bearing to the seat support – then connect the accelerometer/gyro microprocessor control and H bridge. You sit on the bicycle straight up at a standstill and turn the system on. You put your feet on the pedals. If the bike leans slightly in one direction the motor torques such that its Newtonian opposite reaction, the flywheel provides a "quasi rigid" mass in space to react against, turns the bicycle such that the tire contact restores it to the upright position – the motor/flywheel, controlled by the accelerometer, gyros, and MOD5213 processor rapidly operates forward and reverse to maintain balance. As you start pedaling forward the speed is sensed and the gain is reduced. Besides novelty it replaces training wheels in teaching kids to ride bikes. The kid is confident because the bike stays up at a standstill and smoothly rolls off. The parent gradually reduces the gain until the kid can ride. Note that a treadmill motor will produce 5 – 10 foot pounds of stall torque at 24V – more then enough to reactively kick a bicycle over a couple of inches. I have a massive H bridge to drive it and it will use RC style lithium batteries. More info when the motor gets here. You can prove the concept - if you have ever run a substantial DC motor on a bench that is not bolted down - notice that even without a flywheel it tends to flip over in the opposite direction when started. I intend to harness this flipping tendency.

And I believe propellers run about 85% efficient, mechanical drives do have losses, and the mechanical drive requires more weight and complexity.

But you are right that it would require special bicycle lanes to accommodate the slow acceleration and so on. So it was a good experiment but I am probably not going to build anything full size.

Huh. You know what, I opened my mouth on a subject that I don't actually know much about. Propellers are much more efficient than I thought.

On gyro stabilization, that'll work. I know that there's already a gyro wheel for children's bicycle. Turns out gyros/reaction wheels are used in a lot of things.

https://www.youtube.com/watch?v=ngdOgTF2FPo

While the theory sounded good the balancing wheel has problems of it's own. There is a vibration feedback, it vibrates really hard, and this is like too much mike gain on an auditorium PA system. Unwanted oscillation. If you turn the gain down (change software multiplier) and increase the exponential filtering the vibration stops but then you don't have enough torque fast enough to roll the scooter. That is illustrated in the video. Had the flu and have not done much but this is going to require some sort of vibration isolation and digital filtering to work.

In addition - the motor is way too heavy. I need a large RC BLDC with a steel disk instead of a propeller so I am going to work on that.

Unlike the propellers - this was not an instant success.

Hmm... Perhaps Paul can weigh in with his experience from his hovering rocket. I know he had to do a lot for the balancing.