Thanks for posting Jason. Can you tell me where you bought your big belt pulley? Would you mind adding the belt and pulleys to the components list or telling me the specs? 

I love to see projects with old scooter parts. I have built ebikes and other cool things from those 24v motors.

Another quick question, with your chain driven version do you remember what the gear ratio was? I currently plan to use 11:80 which I think should give me a similar top speed of 10mph. I actually plan on building this as a means to commute to/from work, and to/from the local climbing gym (should be a good test of it's range) :)

Yes, the chain version was 11:80, and the new belt version is 13:95.   Oddly the belt version feels faster.

Awesome build. How have you attached the sprocket onto the wheel?

Thanks!  The wheels and sprockets both use a standard thread.  The sprockets / wheels aren't meant to be driven in reverse though.  When I reverse drive them quickly, sometimes the torque is enough to loosen the sprocket, so I use JB Weld to hold them on.

Thanks, I hadn't thought about that issue before. The thing I was confused about was that the sprockets themselves don't have a thread, just 4 screw holes like in this: http://www.ebay.co.uk/itm/Electric-E-Scooter-55T-Teeth-6mm-Drive-Wheel-Back-Hub-SPROCKET-Electricscooter-/191671815227?hash=item2ca087cc3b. I've figured out that something like this is required which then screws onto the wheel http://www.ebay.co.uk/itm/Electric-E-Scooter-Bike-FREE-WHEEL-BEARING-Hub-Roll-No-Pedal-24v-36v-48v-Ride-On-/171916176430?hash=item280700ac2e. Only issue is that is free-wheeling when rotated backwards. Is is possible to buy non-freewheeling versions, and do you know what search term to use, or do you just weld it up? Thanks again.

Ahh, I was thinking the new ones, which are belt sprockets, and those have a built-in thread.  Yes, the old ones did use a freewheel adapter that I welded to fuse it.  You could probably also drill a hole into it and fill it with epoxy, which might look cleaner.

Where did you find those wonderful wheels and tires?

Search for electric scooter wheels - there's a bunch out there, ranging from small Razor-size, to larger ride-on style.

There are a bunch of places that sell them, like this one:

http://www.partsforscooters.com/All-Scooter-Parts/Rear-Wheels

Thank you. My Google searches didn't go any where because I didn't know what keywords to work with.

Great site too.

Yeah sometimes that's the hardest part.  I looked for large robot tires, lawnmower wheels, bicycle wheels, scooter wheels, go-cart wheels, wheelbarrow wheels, ...  Took a while to find good ones.  There's another site called ElectricScooterParts.com (I think) that sells stuff too - motors, belts, chains, sprockets...  The scooter stuff is a gold mine for larger robots.

Hehe. I recognize a lot of those same searches.

I like it. Love overhead on the motor controller. What would you change, if anything, on the next one?

I have a bunch of changes - wider stance for the wheels to reduce side-to-side wobble, shorter board (the length isn't necessary, and adds weight), aluminum & wood frame instead of steel (less weight), belt drives (less noise, easier to tension).

Ultimately I'd like to try to make it steer like a snowboard - shift your weight toward one edge and pressure sensors notice the difference and steer accordingly.  I'd also like to make the steering speed sensitive.  As it is right now, it's too easy to oversteer when you're at speed, and it'll throw you.

Given that this is a first attempt though, I'm really happy with how well it turned out.

I'd also love to make it brushless, but it's relatively hard to find a sensored motor controller that will handle instant reverse.  They exist, but the ones I've found so far have complicated wiring harnesses and are primarily for E-bikes or scooters.  Tempted to try to make one, but I know I'll burn a lot of FETs and probably a few motors before I get something usable.  :)

maybe not, there are a shit-ton of brushless motor driver ICs out there, there might be one good for just what you are trying to do, if you find the right chip I can help design the pcb and pick the fets, it could even be a parallax compatible, because I could use one too - wanna convert a unicycle to be powered... needs the same kind of reversibility.

This one will handle high voltages, high amperages (up to 150a), includes a charge pump, and takes a PWM input.

http://www.allegromicro.com/en/Products/Motor-Driver-And-Interface-ICs/Brushless-DC-Motor-Drivers/A4915.aspx

This looks interesting, too, as it'll apparently do sensorless operation with absolute phase determination:

http://www.melexis.com/Assets/MLX81205071015-Product-Abstract-6114.aspx

The first one, taking PWM and direction as input is pretty cool.  Presumably I'd have to manage braking though, which is something I'm not sure how best to handle.  My ideal would be a motor controller where you gave it a value, positive or negative, and it applied that "amount" of throttle in that direction.  I suspect any of these chips is going to need a higher-level piece to manage that bit though.

I'm liking the allegro chip, what kind of voltage do you need/want?

I'm at 24v right now, and I think that'd be plenty, but it really depends on the motors I can find.  What ranges are easier to work with on the electronics side?  I realize lower total wattage is simplest, but given a choice, is it easier to work with lower voltage or lower amperage? or are they just different but equal problems?

If I end up working with R/C plane components, 12v to 24v is kind of their normal range.  Scooter components tend to be 24v/36v/48v, but I don't need a ton of power for this thing, so I think 24v with a little headroom would be fine.

(Also, I thought I replied to this a week ago.  Work has been nuts)

To me, problem areas for motor controllers are when voltages are high, above 90v. High amperages are typically easy, just use more fets! In the 12-60 v ranges, you have a ridiculous selection of good components to choose from.

Work has been nuts here too! 60 hour workweeks are fun. You know why engineers have mistresses? While they work, the wife thinks he is with the mistress and the mistress thinks he is with the wife.

Do you have suggestions for what FETs you'd use?  Know of any suitable ones that have TTL input (3.3v) instead of needing a charge pump?  Work is starting to slow down, so I'm looking at doing a board layout for the next iteration of the Sideway, and it'd be cool to eventually include the motor controller.

Sorry for the wait, I looked for a little bit this morning,.  To educate yourself, read the last tip on this page for 3.3v logic level mosfet finding advice: http://ww1.microchip.com/downloads/en/DeviceDoc/chapter 8.pdf

here's a search I narrowed down on mouser (sorted by price right now): http://goo.gl/hhb9Lk

My next choice depends on how to build the PCB, and what I might use these fets for the in the future.  PCBs are a PITA, so we'll often try to add as much functionality as possible into one. for example, why not use much higher voltage rated fets? well, above a certain voltage you're going to need fet drivers. And power dissipation becomes an issue above a certain amount of amps. 

My goal would be to use lower rated amp fets and double them up so we can let the packages do the heat sinking instead of actually having to heatsink the fets, and to use ... 50-60v rated fets since it isn't uncommon to see 48v systems.  I'd also give it 4 channels - which you probably need anyhow, since you are going to essentially need to make two H-bridges with fets.

Your goal might be different, a motor controller of this power, could be tiny! I'd be torn between the two. But one thing is certain, design one, the second will be much easier as you'll simply change out some parts and redesign the PCB accordingly.

here's my suggested fets, which ones are chosen will determine your gate drivers (mainly based on gate charge):

15A, 60V, 1.5v turn on, 49nC gate charge, 4.5mOhms on resistance:  http://goo.gl/EsXODv

40A, 60V, 2.8V turn on, 27nC charge, 4.2mOhm Rds On: http://goo.gl/AFKYxQ

50A, 30V, 2.5V, 44nC, 4.2mOhm: http://goo.gl/Y9JIN7

40A, 25V, 1.7V, 31nC, 2.8mOhm: http://goo.gl/aZQz1v

16A, 30V, 2.4V, 7.2nC/45nC, 1.3mOhm: http://goo.gl/A4iYiR

35A, 30V, 1.3V, 78nC, 1.23mOhm: http://goo.gl/fRQwBz

50A, 40V, 2V, 23.5nC, 3.5mOhm: http://goo.gl/Q0ooyq

So, package power dissipation is important, it relates to the Rds On resistance like this:  

Power loss = I^2 * RDS, I=average current and RDS is the mOhm values from above.  and every package of every fet is unique, look at the graphs typically shown for this, on the last mosfet listed: http://www.mouser.com/ds/2/149/FDD9410_F085-533623.pdf

on page 3, figure 1 and 2. On figure 1, you'll see how you need to de-rate the power handling based on how hot the outside of the case becomes. If we assume you used this fet, assume 300 watt averages, at 11V (I like to use the sagging voltage of a heavily loaded lithium, so I calculate a higher current, which gives me a higher estimated temp, for safety) then you'll pull Amps = power/volts = 300/11=27.27.27= 28, or 30 for more safety! Now, knowing the current, calculate the power loss, 30^2*0.0035Ohm=3.5watts!  Now this is not bad, for a package of 4.5mmx5.2mm, this might be heatsinkable on the PCB alone. However you're going to have four of these, and being a balancer, it'll split the time between the fets unless you are traveling forward constantly, basically you'll rarely see this kind of constant heat. but you might see at most, 3.5watts x 4, or 14 watts if you simply swung back and forth at a standstill (assuming it doesn't take even more power when reversing direction - which it probably does), depending on the PCB size, airflow, etc, 14 watts could be too much.

Next, I'll look at drivers, but I've already spent enough time on this, time for another cup of coffee!

On a side note, I like to personally spend more money, a $1 or two more per fet goes a a LONG way, just like your beers. check these beasts out: http://goo.gl/cAoFk5

I haven't yet, but have always wanted to try some of the new tech with GaN and SiC internals:
http://goo.gl/R0w4tX

http://goo.gl/cvSM3k
http://goo.gl/lpt6hU
as well has the hexfet nexfet designs coming out lately:
http://goo.gl/el2xhw

So I've started playing with building my own boards, and I'd like to try to give this a go.  I've ordered a few of the Allegra driver chip, and a bunch of the FDD9410_F085 (50A, 40V, 2V, 23.5nC, 3.5mOhm) FETs.  The Allegra has the gate drivers built in, can be externally PWM'd, and has brake and coast, so it looks like a solid choice.  The datasheet talks about selecting appropriate value bootstrap and vreg capacitors - would you be willing to have a look and provide guidance there?  At present, I'm still assuming 24v, though I might drop down to 12v and go higher amperage so I can run the motors at a lower speed.  Haven't made that decision yet, and I'm not sure how heavily that affects the bootstrap and vreg decision.