I don't mean to imply any stupidity in your question, but are you absolutely sure you need a motor taking up to more than 7kW and outputting roughly 3kW (3.8HP)?
If it's for your unicycle project, it's gross overkill (and I mean really-really gross).
The only thing I'm absolutely sure about is that I'm not absolutely sure about anything.
Yes, it's for the treadmill unicycle project we were talking about
here. There is a lot of stupidity coming from me. I wish I already had the necessary motor knowledge. I'm a software engineer by day.
Here's some of the logic (good or bad) that led me to that motor:
VegaObscura helped me with how to determine the torque I need.
I put that in a spreadsheet, here it is:
mass (lbs) 160
acceleration (ft/s^2) 11
wheel diameter (ft) 0.66666666666667 (previously I thought it was 0.5 ft)
mass (kg) 72.5747792
acceleration (m/s^2) 3.3528
force (n) 243.32871970176
wheel radius (cm) 10.16
torque (n-cm) 2472.21979216988
torque (lb-ft) 18.234157434005
double torque (lb-ft) 36.4683148680099
For the wheel radius I am measuring the existing
pulley on my treadmill's drive roller. It has an 8" diameter. Maybe it's the roller itself that I should measure here?
This value effects the results significantly. PLEASE let me know if this or any other values don't make sense. It's confusing to me.
I saw
a self balancing unicycle project online. But then, I'm not sure how fast his uni accelerates or his top speed. He used
this A28-150 motor.
The specs for this motor from that link are:
24V (can be run higher)
3 horsepower
1970 oz-in Torque
Max current 285 Amps
82% Efficiency
6000 rpm
To convert 1970 oz-in to lb-ft I divide by 16 to get lb-in, then divide by 12 to get lb-ft.
1970/12/16 = 10.26 lb-ft
Hmm. Less than the 36.5 lb-ft I'm after?
But I learned that Torque in lb-ft = (horsepower / RPM ) * 63025
Using the specs: T = (3 / 6000) * 63025 = 31.5 lb-ft
Hmm. Completely different than the 10.26 lb-ft spec, and still less than the 36.5 lb-ft I'm after.
I certainly feel stupid here.
But maybe it's close enough? I don't have any past experience to make a judgment.
What happens if I gain weight or my 200 lb friend tries it?
Then my spreadsheet says I need 45.6 lb-ft.
Looking at
the A28-150 motor, it costs $300 and I'll need the $30
motor mounts for a total of $330.
Looking further I found some motors at
robotmarketplace that had more torque but cost more, and some the would require some kind of L-bracket for mounting. Then I saw
the NPC Black Max motor that has the mount I'd like, is cheaper than the A28-150 and has way more torque.
I figured it was a good value and would remove any doubt I had about power.
It sounds like it's way overkill.
But again I ask, is there an alternative that is going to be better?
I haven't found a source of motors that ends up cheaper for me.
In the back of my mind I figure I might be happy to have such a powerful motor for some future application. I think of it as an investment.
But at the same time I'm worried a bigger motor will require a more expensive controller. But I have no frame of reference. Yes, I'm stupid. If the motor is going to be more trouble than it's worth, despite what seems like a good value, please let me know.
If you insist that you either go super fast or have an extreme acceleration/deceleration (or both) warranting such a powerhouse, I'll pretend to believe 
you
I generally ride a max speed of 10 mph (16kph) and average about 6mph (10kph), but I own a geared hub that I'm not yet good with, but I expect to go up to 15mph (24kph). But riders have gone 24mph+ (39kph) with that hub and my current wheel size (24"). Unicycle wheels are made as big as 36", and I've read of people riding near 30mph (48kph). I may own such a ride in the future.
I'd currently be happy to ride about 15mph (24kph).
What's my acceleration? I admit I haven't really measured that... just imagined. I can get up to normal speed (6mph (10kph) ) in a few seconds.
What might be more important is deceleration. If I get tripped up and have an unplanned dismount, ideally the treadmill will come to an instant complete stop. (Failing to detect a dismount and false detection will both be dangerous... but that's a different problem to solve)
3 lead-acid "marine" batteries connected in series - fed from a charger that can make up for the average demand (and a little more) over a day (or whatever your timeframe is) and things should even out without making the supply hideously expensive.
I've grown to think of batteries (and unrelated to this conversation, combustion engines) as needy dependents that I'd prefer to avoid around the house. I figured some kind of AC solution would be available, and since my treadmill isn't mobile, I hoped to take advantage of AC.
It must be possible.
Is it not practical? I'm stupid here too.
The (DC) motor controller needs to be able to handle the stall current and will be very expensive, but shop around - the same capacity may cost more than double the "cheapest".
(Or build it yourself).
Honestly, I'm too stupid to build it myself. I'm sure I have the brain capacity and dexterity, just not the knowledge and time. I'll have to buy one.
But of course, I need the motor choice before I can choose this.
I'd also like to know how I could power the motor controller with AC. Aren't there AC powered DC motor controllers? Or, aren't there DC motor controllers that can be powered with a AC-to-DC power supply?
AC-to-DC-to-battery-to-controller is one way... but why can't we remove the battery part?
The 250W motors alone should be able to take the average two-wheel rider to 50..60km/h (don't remember the exact speed) without pedaling and, judging from some avi's I've seen, with a fairly decent acceleration.
Can you point me to one of these motors? Are you suggesting it would be good for my application?
I have a hard time imagining a one gear, one wheeler touching this (and the rider living to tell).
Don't forget, in my case the rider is controlling the speed. Unless there is a problem with my system, the treadmill will only move to match the rider's speed. Still... it will be dangerous until all the inevitable bugs are worked out, but even then it will never be completely without risk.
