Search all of my sites with Google

Friday, June 13, 2008

Roller Skates & Fan Motors: Friction Drive 2.0

What would you make with an old rackmount plate, a pair of roller skates, a couple of car radiator fan motors, some aluminum bracket, and a handful of nuts, bolts, blind rivets and a popriveter?

That's what I used to make the 2.0 version of this summer's friction drive. It's no beauty, but it works even better than 1.0 by a significant amount, though there are some major hurdles I have yet to overcome. One main advantage it has over 1.0 is it's location--it's out of the way of the baskets, leaving the top of it clear to be used as a cargo rack again. (I was missing that quite a bit, in the few days I used it with 1.0). Another is center-of-gravity: it's all a lot lower than it was before; bracket and motor mass is now not as likely to tip me over (still have the overpoweringly heavy batteries, but at least this stuff isn't up top now). It's also now clear of the rear wheel for service access--I could not have flipped the bike over to work on it with 1.0, but 2.0 is designed to not interfere with doing that.

The ex-rackmount-panel mounting plate is bolted to a pair of mounting brackets, which are themselves riveted to the bike frame itself on the rear triangle. It was much more secure than trying to clamp it, and easier and faster to do. The panel already had a series of holes in it that probably held switches and/or lamps for a mainframe computer system (that was likely all gone when I got the panel, so long ago I can't actually remember). It's nearly 1/4" thick, unlikely to bend under any stresses I might subject it to in this design.

It's mounted as close to the frame as I could get it, and still allow space for the motors to be between it and the rearwheel/tire of the bike, with clearance for tire movement/etc. That just barely leaves clearance for the pedal cranks (more later).

Closer pics of the panel and motors above. The friction wheels aren't installed in these pics. It might be a little hard to see, but the mounting panel is cut out around the motors and motor mounting points in two of three places to allow them each to pivot around the third mounting point, so tension against the wheel can be adjusted. If you've ever adjusted tension of fanbelts on alternators/etc in a car, you'll be familiar with the way this pivoting works.

Initially, I was going to have the entire mounting panel just slide back and forth a bit to pull the wheels against or away from the rearwheel tire, but I realized that would probably make tension on each motor different, since I'm not doing this in a very precise way right now, as a test version. So I used the pivot method of moving the *motors* instead. It slightly complicates the method of controlling that tension, in that now instead of just one spring to pull the motors away from the wheel, and one cable to pull them against the wheel, there now must be two of each. It is possible to combine the two cables into one, and I may do that, but still needs two springs, to pull the motors away from the wheel when not in use--that keeps me from wasting my pedalling energy on the motors, instead of only the rearwheel. If I use two cables, it also allows me to use one motor or two, as the need arises. That will likely increase my battery life, by only using up current on one motor unless I really really need the power of two (which will happen often enough to put two on here--they're really not very powerful motors--at 12v, I think locked-rotor current was around 5amps? And I'm absolutely sure they wouldn't sustain that current very long. I think they drew about 1.2amps while spinning their fans at max RPM in their original use, at 12v).

The top mounting bracket for the panel, along the outside of the top bar of the rear triangle:

The mounting bracket is pop riveted to the frame; you can see one of the rivets at the lower right. The other is hidden, under the big plate, between the two nut/washer/bolt stacks that secure the plate to the bracket. Those aluminum brackets were already shaped just about exactly like I needed them (or rather, I built it this way because the brackets existed!). The bolts are secured to the brackets with lockwashers on each side of the bracket, and a nut to clamp it to the bracket. That way I can easily secure/undo the nuts holding the plate to the bracket without having to have something to hold the bolthead from turning. The nut between the plate and bracket is also a spacer to keep the motor cases clear of the bike's rearwheel/tire.

These are the motors, with ex-rollerskate friction drive wheels installed:

The top one is on the left, and the botton one on the right. There's more clearance than it looks like for the top one--the pic is just at a bit of an angle from the front--the straight-on pic always blurs due to the autofocus light hitting something wrong when I do it that way. These are the pancake-style radiator-fan motors I got at the junkyard at the same time as the ones I used in the 1.0 version of this friction drive setup. They're a bit over an inch thick, instead of the maybe 4" thick the others are. The important part though, is that they are larger diameter, and apparently have more torque because of it, as they draw about the same power as the other ones do under the same load. I would have used these in 1.0, except there was no way to use them with any smaller hub than the fan hubs that came with them, which were such large diameter (~5"?) they would have been horrible drive wheels--much much worse than the 3" hubs for the other motors.

These are the roller skate wheels I mentioned getting via the thrift store in the previous post; they've become the actual friction drive roller wheels in this version, since they're much smaller than the fan-blade-hubs I was using before.

The bearings from inside them are a bonus--I didnt' expect it to be that easy to take them out, but they just pushed out with a little tapping. The way they're made, I should be able to use them for a few different things, including one of the ideas I'd had for a trailer-hitch bearing a few posts back. Don't need them for this friction-drive setup, though, so they get put in the "hmmm" box for now.

This is one of the wheels after hollowing out some of it's rubber to make room for the motor's aluminum fan hub core. I have to use those cores because they are already D-cutout for the motor axle, and I don't yet have a lathe or similar to make one myself. Might need to build a lathe, though, because there are more and more things I need one for.

Now, I didn't have any good way of putting the core really deep into the wheel, as there is a *very* hard plastic core to the wheel; perhaps nylon? I couldn't easily cut into it with hand tools, including my titanium-edged utility-knife blades, which I can whittle hard aluminum with! So I cut as deeply and symmetrically as I could, and then JBWelded the two together, ensuring the best balance I could get, which turned out pretty close to perfect (no wobble or vibration to the motor because of any off-centeredness of the mounting job). I let it cure for almost two days before trying it out, but in the end it didn't make much of a difference.

With the mounting plates/etc., the only problem I was seriously worried about was if I could make the pedal cranks clear all the fasteners, mounting plate, etc, and still have the motor casings far enough from the rearwheel treads to not have to worry about anything during rearwheel side-to-side flexings, such as during turns, etc (it doesn't move much, but it does move a little bit, especially with that heavy battery load in the rear baskets).

As you can see, it's a very tight clearance--maybe 1/8" max on the pedal end of the crank to that rear nut. It actually didn't clear at all until I cratered-out the area around the mounting hole there, with the angle-grinder. Since the rackmount plate I used for the mounting plate is nearly 1/4" thick, grinding down some of it here and there is not going to make a significant difference to it's strength in the directions it needs to be strong in. I also had to cut out a curved area near the shaft end of the crank, so it would clear the curving-inward portion of the crank at the mounting plate's leading edge.

I was in a hurry to test it out today, and had to go to work, thus shortening my time to work on it too much to do a normal fix for the controller that had burned out, so I looked for any switches I had that could handle the current I needed, and were big enough to easily toggle with gloved fingers.

Only thing I had easily accessible were some spare lightswitches in a wallplate in the utility-room junkbox, so I used one for each motor, both to limit the current flowing thru any one switch, and to give me some semblance of speed control (one motor on and one off means the first motor pushes against the unpowered resistance of the second motor, so it doesn't go as fast as it could, then both motors on means it goes perhaps three times as fast as only one motor, since neither motor is resisting the other). This was only a temporary first-run test; I won't be using it later on--but it did work, and fairly well--just hard to deal with being so big and on the center of the handlebars instead of easily reachable without taking a hand off of the grips.

There was a failure (expected, just hoped against) of the JBWeld holding the motor hubs to the skate wheels; even though they were embedded in the wheel a little, it obviously wasn't enough.

Both of them came off within seconds of each other, because I hit a pothole, which probably forced the tire to expand more against the wheels, forcing the wheels even harder laterally across the axis off of the hubs. The JBWeld couldn't stretch any more and broke. Only bad part is that one of the wheels stuck between my tire and seatpost, forcing me to a rapid stop, and wearing a chunk out of the skate wheel. I've got six more, so no problem there, but it was unexpected for it to jam against the tire like that, and the sudden braking was potentially dangerous.

So I need to make a much better way to fasten the skate wheels to the motor hubs (or directly to the motor axles) before I test it out any further. I have several ideas, but some of them require parts I don't have, and would be difficult to make without a lathe or other power tools I don't yet have. I can build a lathe from one of my drills, and that will probably let me make at least one of the ideas I have, but that's more time to do that, too. Well, better to spend extra time than to have it break yet again due to poor engineering and implementation. :-)

However, before the failure, I had done some tests on a clear-of-all-traffic paved path, one with inlets only at the ends, and no chance of anyone crossing my path while I was at speed (in case anything went wrong). Since these motors still have almost no torque at low speed, I pedalled up to about 10MPH, then cut the motors in and stopped pedalling. Unlike with 1.0, where I would have about sustained the speed and maybe gained one or two MPH before falling off and slowing down, I accelerated quite a bit, and pretty quickly, up to about 32MPH on motor power alone! I only let it go for maybe 10 or 15 seconds, because I felt too shy of letting it go any longer or faster (I don't think it would go any faster, but the bike felt skittish and I didn't wanna risk it). I also don't know how long it would last at that power level, with no input power from me, as the motors weren't designed to run at these power levels at all--I was just testing them with the 36v straight from the batteries thru the above switches, rather than thru any current-limiting controller. I doubt the batteries themselves would sustain that kind of power for long, either, as they are (as I've said before) old and not great to start with, which is why I am using 36v in the first place--making up in voltage what I am not getting in current.

It's also, AFAIK, not legal to run it at over 20MPH on public roads with the motors engaged, so there's another good reason not to work on it for speed. :-) Realistically, I don't want it to be fast, anyway, as I need distance, not speed. Just something to help me not wear *myself* out getting places during the summer, and to be able to go farther than I can on my own anytime/season.

I have a few ideas for switching in the third battery only when the first two drop below a certain level under load, and things like that, but haven't even drawn up a sketch of the circuit yet. Might just do it manually with switches/relays first.


  1. That is astonishing. The idea of using roller skate wheels as a drive wheel is pretty inventive. Are they substantially different from skate board wheels? I know they make lots of variations of those.

    I loved the detailed photos and descriptions of every step in the process of 2.0. It seems like a huge step forward from the previous model.

    Your ability to make useful and functional end designs out of a mish-mash of whatever pieces you have been able to gather will never cease to amaze me. I have to say, should my starship ever crash land on an inhospitable world, or my time machine bust down and leave me stranded in the outskirts of Knossos circa 1700 BC, there isn't anyone else I'd prefer to be with to bail me out.

    Well, except maybe Jessica Alba, but that's assuming another set of circumstances entirely.

  2. Astonishing? Doesn't seem like it to me, but I guess I'm used to the whole idea of what I do. :)

    Well, I borrowed the idea of the skate wheels from Jim Rudholm's friction drive setup (along with the basic version of much of 1.0), except that he used skateboard wheels. The only reason I used skate wheels instead of skateboard wheels is because it was $10 for a skateboard (with 4 wheels) at the thrift store, but only a couple bucks for a pair of skates (with 8 wheels).

    I didn't examine either one closely enough before buying the skates to notice any significant differences (besides color); they both looked about the same diameter and width at the glance I gave them looking for wear and tear. I figured they'd both be constructed of some sort of hard core with rubber coating of some type, and bearings inside. I wasn't sure if I'd get removable bearings or not, and figured on having to destructively remove those (but didn't have to, fortunately, as I can actually use more bearings like this).

    I'm sure there are better skate/board wheels I could use for this, but I didn't look around for anything other than what was in the thrift store before using these, as it was really more of a "lets just DO something" kind of day. :-)

    2.0 is definitely a huge step forward from 1.0, at least in the idea of it, and how it has performed so far in it's rudimentary incomplete state. Once it's on a PWM controller (as 1.0 was before it smoked from the midday heat while in use), and I complete the tensioner portion that lets me adjust the tension of it on-the-fly, and disengage it from the tire entirely when not needed, it will definitely be better.

    I'm not sure what I could do about a starship or a time machine, but hey, you never know until you try. :) I sure wouldn't have imagined motorizing my bike 20 years ago (which kind of surprises me, in that I've been a gadgeteer for as long as I can remember). I *might* have imagined putting lights all over it, though I never did that (I did design a totally electronic dashboard for a 1968 Mustang I was going to rebuild, but never built it, as I gave the car to my mom when hers died quite suddenly, leaving her with no way to get to her many-miles-away daily job (whereas mine was close enough to bike/bus to)).

    Now, Jessica in one of her characters' personas might be able to help you more than I, but I definitely get your meaning. :-P

  3. What did you use for the wiring and controlability of the motors? And could you add some pictures of that part of the build/materials ect. It would be most helpful thanks awesome setup I'm wanting to use fan motors on mine I was trying to see more details on electrics on it so I can come up with my own version of it.

  4. What did you use for the wiring and controlability of the motors? And could you add some pictures of that part of the build/materials ect. It would be most helpful thanks awesome setup I'm wanting to use fan motors on mine I was trying to see more details on electrics on it so I can come up with my own version of it.

  5. I would've sworn I had a post all about the ex-scootngo controller repairs, including the replacement of the blown output FETs with external ones, and then later going up to 100V 123A FETs on an external heatsink that was then bolted to the rackmount plate on the side along with the motors, but either the interwebs have eaten the post (possible) or I never got around to posting it (more likely). :-(

    Anyway, I don't have any other pics I know of of this old drivetrain, but some of the best pics are gathered here, on Endless Sphere ebike forum, in the DayGlo Avenger MkII thread.

    Because of a motor shaft breaking (documented in later posts) I had to abandon the friction drive, and it's now using a hub motor, resurected after quite some time of using a power chair motor chain drive on a custom-built bike, CrazyBike2, that replaced DayBlo Avenger.

    Even CrazyBike2 has changed into a hubmotor drive, too, now, after chainline problems broke too many things, but a new custom bike is being built, covered over on Endless Sphere, just like the others were here. I do not really update this blog very often anymore, becuase I am posting and helping out over on Endless Sphere instead.

  6. Oh, and what I meant to add was that you should join up over at Endless Sphere, and post up a build thread, where we can all help you figure out what to do. :-)


Alternate suggestions or improvements to anything that's been posted is very welcome, and extreme detail is preferred to brevity.

Keep in mind that unless you leave an email address in your comment, I haven't any way to reply to you except to reply to your comment here. That means if you want a reply, you'll have to come back to *this* blog entry and it's comments to see my reply to you, unless you leave some method of contact within your comment.