Page 1 of 1

Friction Drive Project

Posted: Jul 27 2014 2:01pm
by Solcar
Friction drive project 3

The drive unit should be buildable by DIYers like me who have access only to basic hand tools and who aren't all that skilled at working with materials. This roller is built onto 1/4" threaded rod. That should offer many advantages. I was also working on one that used a 3/8 threaded rod, but I set that aside when I found, at the local hardware store, bearing pulleys that fit onto 1/4" shafts. I plan to bend thick wire around these pulleys, which will then be the mount for the drive assembly. Not blingy at all, but it is hard enough for me to just get a functional design I like, let alone the appearance. I also found that I had some 2 1/2" polyurethane wheels on hand. Those seem to have good potential as actual friction drive rollers. They have preformed 1/4" holes. I have noticed that if I try to drill a hole through anything with almost any thickness, the result ends up pretty unusable because of inconsistent circumference through the material, and worse, because of the hole not being perpendicular to the surface.

I feel a bit more confident in this newer roller assembly made on the 1/4" threaded rod than an earlier attempt that used the 3/8" rod because the polyurethane wheel seems to grip better on the tire than the earlier one's polypropylene roller (I'm hoping to avoid having to apply grip tape to the roller),

The sprocket on the friction drive roller assembly is the standard freewheeling sprague type often mounted on the shafts of motors powering electric scooters. On this drive roller assembly it is kept in place and locked onto the shaft with claws instead of pins because drilling a straight hole through a rod is almost impossible without a good drill press, and the 1/4" rod doesn't have a wide enough diameter to support a hole for pinning the sprocket, especially since the threads on the rod sacrifice the strength diameter-wise, but the benefit of being able to secure the items onto the rod with nuts way more than makes up for that.


The friction drive unit looks almost ready to work. I replaced the bearings with better ones, removing the sliding door rollers. That enabled me to use pipe hanging strapping to construct the pivot arms from.

On the bottom side of the arms are clothespin sticks that were drilled with holes before the screws were inserted that hold the pivot arms together. I decided to use them instead of pine plywood carpet tacking strips because they are convenient and because they are harder wood or maybe even bamboo (for the outside ones). My guess is that the inner ones arbe poplar. They greatly improve the stiffness of the pivot arms.

The blue roller is polyurethane, I believe. It seems to get a decent grip on the tire rubber. The diameter is over twice what a roller might tend to be at 2.5 inches. That should lower rolling resistance and also improve the grip by enabling easier broader contact with the tire.

The jackshaft assembly will hopefully stay put fine with just the aid of something like tape being placed on the threaded rod and the bearing pressed over it to hold tight.


One feature of this friction drive is that the pivots for the pivot arms are nuts that rotate on the threaded rods. They hold the pivots in place while allowing rotation.

Presently, the reduction gearing ratio is about 6:1. The friction roller rotates once for 6 turns of the motor shaft. At first thought that might seem too low a reduction for such a small motor. However, the drive being applied to the perifery of the wheel gives a large mechanical advantage. That is why friction drives can be relatively small and lightweight. This drive assembly weighs just 3 lbs, or slightly more. The motor weighs about 2.2 lbs.

That the pivot arms are made out of pipe hanging strapping makes it almost automatic that the arms will have the same length. That was a great advantage over the previously planned method of using steel wire. The pipe hanging strapping is also easier to work with.

The spring that keeps the roller on the tire was made from MIG welding wire. It isn't bad spring steel. It works but just barely has enough springiness to hold down the roller if the rim is out-of-round. I needed a bit more force than originally anticipated because the pivot needed to be moved in closer to the tire because the roller was slipping during a static test. With the pivot closer, the beginning angle (at light engagement) is steeper, causing the roller to tend to glance off of the tire. Adding grip material would help, but I hope to not need it.

People with very good memories might recognize that freewheeling jackshaft that is held together by claws made of 14 gauge wire. I removed that from the input section of the now retired through-the-gears project that was my last e-bicycle.

Re: Friction Drive Project

Posted: Sep 20 2015 5:10pm
by Solcar
Okay, i may as well bump my friction drive project thread. I don't have further to say on the motor drive assembly this year.

I've been working on the motor controller / charger. It occurred to me lately that the charger is more important than the motor controller because without the charger, the battery can only be discharged once and then if left like that will become damaged (unless it's NiCad). That is a reason why mine is built in with the motor controller.

A thought on my controller. It takes the voltage from my 12v LiFePO4 and boosts it up to several times. That is how it is progressing so far. An advantage of using a boost instead of a standard buck controller on a brushed motor (the type i prefer) is that if the power MOSFETs fail, the motor will stop turning (not go fully on). Because it is a brushed motor, the boost can be varied to drive the motor directly. With a brushless motor, a seperate stage would be needed for the controller, or three boost converters would need to be operated 120 degrees apart. Another advantage of using a boosting controller is that the battery can be less complicated because it needs fewer cells.

Had a setback though. The SG3525A SMPS controller IC inexplicably failed during testing of the circuit. I was building the voltage booster circuitry and testing as I was going along when the SG3525 quit outputting MOSFET gate drives. The oscillator on the IC was still working though. Later I recalled how the chip that I was using earlier as the controller, a CD40106, failed inexplicably too. That motivated me to switch to the SG3525. I might have unknown current spikes getting to the IC chips. Or, in the case of the SG3525, driving 4 BUZ110s MOSFETs per gate drive was exceeding the peak current rating of the IC. I can try 100 ohm resistors between the output of the SG3525 and the emitter-follower buffers because the gates are enhanced through diodes, but they are pulled low through 2SA1020 transistors, which are rated for 2 ampere collector current.

Re: Friction Drive Project

Posted: Apr 20 2016 10:09am
by Solcar
The new season of working on the project is under way. So far this spring, all work has been on the charging portion of the controller /charger unit. I have been wanting to use TLP352 opto coupled MOSFET drivers for simplicity in the transformer drive. Another feature of the charge circuit is how the cell balancing works by implementing separate rectified windings from the switchmode transformer for supplying charge to the individual cells of my 12v (nominal) LifePO4 battery. That way each cell automatically gets an equal no load (final settling in) voltage in a somewhat passive and relatively simple way. During charge, any cells that are lower in voltage will hog current and load down general transformer voltage output until in the final state all the cells settle in at full charge voltage. The voltage regulation only has to be done for the full series voltage of the cell's in the battery. Although each cell sees 3 1/2 volts, the regulation is done at the sum level of 14v.

Re: Friction Drive Project

Posted: Jun 06 2017 10:42pm
by Solcar
A lot of problems in life have been keeping me from working on the charger/controller unit. However, since I had the car battery jumper devices, I decided to place their outputs in parallel and try out the friction drive using the old 12v lead acid battery-powered controller powered by them instead.

The friction drive unit didn't hold up to the stresses. I didn't have time to examine it to see what happened, except I saw one of the pawls connecting the freewheel to the final drive shaft fall on the ground. I was surprised since friction drive has to endure much lower mechanical stress than a motor assembly and chain driving a rear sprocket.

So now two days later, after having been able to examine the friction drive failure mode, I was thinking about giving up on it. However, changing the drive shaft from 1/4" back to 3/8" diameter would likely fix the problem. The 1/4" one was stiff enough, but the nuts can't be made tight enough. Hopefully the extra mass of a 3/8" rod wouldn't adversely affect the maintenance of pressure of the drive wheel against a bicycle wheel that isn't completely round.

Those blue 2 1/2" diameter polyurethane wheels looked great for doing this project. However, they can't be drilled out for a 3/8" diameter rod. I've been looking for a comparable wheel with the bigger 3/8" bore without success so far.

Re: Friction Drive Project

Posted: Feb 12 2018 9:15am
by Solcar
Well, thought it would be good to update on the outcome of the project. It didn't work because the losses were too great. The greatest loss seems to have been from pressing the roller into the tire, even though I used a
2½" diameter one. I wouldn't want to use different tires besides the regular budget streets ones.