My custom sprag bearing freewheel sprocket

[Barncat]

Been pondering this problem sporadically for a couple years and finally arrived at a feasible solution. Spent the last couple weeks building this prototype. I usually don't post an invention prior to some testing, but this will work, it's just a matter of durability of the two keyway interfaces...

Advantages: INSTANT silent smooth engagement, and silent low drag coasting.

Even good quality freewheels have 3 degrees of backlash- which usually creates an annoying thwack to take up slack when twisting the throttle, plus the pawls buzz loudly when coasting which I don't like.

I began with several pieces of steel stock lying around, and a gutted cheap threaded freewheel. This design will screw on to any conventional 1.375 x 24 wheel hub. The star of the show is a CSK25PP sprag (one way) bearing capable of transmitting 63 lbft of torque that has two keyways.

The mechanism has an inner and outer hub. The inner hub was turned and bored 1" stock TIG welded to a machined heavy 3/4" washer welded to the threaded core of an old freewheel. I held good concentricity with various tricks... it's either that or spend hours making chips from very heavy bar stock. Flanged parts are problematic.

The outer hub was bored to fit the bearing with an outer lip left for location and an inner lip to locate the sprocket for welding. I used to own two milling machines but sold them due to various moves, so keyway and key work was PITA manual... the only reason this will transmit power is the two keys. Welded on a set screw assembly as pictured. Welding the sprocket to hub was a mistake as the hub ID shrank and distorted slightly but I managed to save it. As with any prototype a couple changes would be made next time.

It'll be a while before I can road test as this goes on my Mongoose Blackcomb build which is getting a new motor and new 20s Barncat battery.

 

[amberwolf]

I'll be interested to see how this goes.

I'm not sure I can actually manage to do the work well enough, but my ancient SB lathe could handle the size of the parts at least.

 

[Barncat]

Yes amberwolf, rest assured I'm looking forward to putting a few miles on this. If you were to try this on your cargo bike you'd need to go up a bearing size or two, but let me report back first... as mentioned there are a couple minor mods I'd do in a version 2.0 but I'm fairly confident this will fly.

As a bit of a checklist and further detail: I offhand ground some HSS lathe bits to make the 2 keys, so they'll maintain good engagement with the hardened bearing keyways. The welds will be good. The bearing torque rating is probably conservative. No published torque specs on the EC4P motor I'm using- I'm guessing 10-12 ftlbs, with a 5:1 sprocket reduction in this case, so probably in the 60 ftlb range. (Edit- that's a wild guess on motor torque and likely too low, but this bearing is actually capable of handling 125 foot pounds of torque or more for short periods...).

Also, I'm a trim 175 lbs, and ebike throttles need to be engaged and operated more smoothly than motorcycle throttles, especially off the line. It's advisable to limit major torque demands...

Assembly ready to roll on the Blackcomb pending controller and battery installation

 

[SlowCo]

Great idea. Silent coasting 'sounds' awesome

 

[Chalo]

I've considered using a CSK clutch for a front freewheel setup, but it looked to me like CSK35 would be the minimum that offered any safety margin. Even taking into account the fact that front sprockets are usually bigger than rear sprockets, I expect CSK25 is under spec'ed for freewheel use with above-human power. So I'm interested to see how yours holds up in the long term.

 

[Chalo]

I'm a trim 175 lbs, and ebike throttles need to be engaged and operated more smoothly than motorcycle throttles, especially off the line.

Mine don't. Even when I'm using >2500 battery watts in the rear wheel or in both wheels, I whack the throttle open as hard as I like, anytime I like. I only get front wheelspin, and only on wet, loose, or steeply uphill surfaces.

 

[Barncat]

Chalo- "considering" doing something and actually doing it are two different matters. Please show us pictures of your functioning freewheel prototype(s).

I've outlined my design rationale above. There is some possibility that the #25 bearing could slip. If so I'm not going to be happy, and will have to build a larger unit... I think keyway failure a more likely problem. It's going to be a few weeks til road testing.

I never whack a throttle off the line. I'm not about to dump 83.5V x 150 battery amps, or about 17hp, into an 11T sprocket and T8F chain and bicycle spokes. That being said my bikes build speed very quickly.

You're free to ride in whatever manner pleases you.

 

[lnanek]

Maybe Chalo just uses hubs and you use mid drives. I full throttle my hub drive based ebike every time I start from a traffic light. No gearing involved to damage like a mid drive. I don't have enough watts to lift the front either.

I would never full throttle my motorcycle even though the 250cc engine is kind of tiny compared to most motorcycles. I have a huge trunk on the tail that makes the back heavy, but lets me lock my expensive AR motorcycle helmet up securely. About only time I would W.O.T. is if an accident is imminent and I can't brake or turn out of it, but can accelerate away.

Mentors have taught me a lot of tricks that let me keep things gentle like approaching a red light slowly so I don't have to fully stop too.

 

[Chalo]

Chalo- "considering" doing something and actually doing it are two different matters. Please show us pictures of your functioning freewheel prototype(s).

No fotos, because it's buried deep in my container at the moment, but I decided to use a LH threaded ACS Southpaw freewheel with 16 teeth to engage a 144t #25 sprocket I milled on a CNC vertical mill. The humongous 3/32" thick aluminum plate sprocket makes it noisy, and my relative lack of chain management and long distance between centers made the chain somewhat likely to jump off. But I never had any problem with the freewheel itself.

I'll note that the freewheel is much lighter and less bulky than a CSK35, but almost certainly tolerates more torque.

If you're using 17hp and only getting 60Nm of wheel torque, I have to wonder what you did wrong.

 

[Chalo]

Maybe Chalo just uses hubs and you use mid drives.

True, at this time. I used commercial mid drives before, but eventually rejected them because they tear up bike parts. One of my buddies just taught himself that lesson (again) within the last couple days.



I used homemade chain drive bikes before, but let those go because they're a hassle.

I would never full throttle my motorcycle even though the 250cc engine is kind of tiny compared to most motorcycles.

I rode 1100cc motorcycles for a long time, and those rarely saw full throttle at any speed, let alone from a stop. I had an air cooled 850cc full bagger that did get a heavy throttle hand in traffic pretty often.

 

[Barncat]

Valid point Inanek re hub motors and throttle modulation.

I build only mid/chain drives which require a second or two of restraint off the line- the same can be said for ICE motorcycles. I've rebuilt two dozen of those up to 1000cc and ride them aggressively in the mountains.

Bolting a sprocket to a conventional 16T freewheel is a simple matter Chalo. I independently arrived at that solution to use a high-engage unit as documented on this site a couple years ago.

My recent invention, while simple in concept, is difficult to DIY execute- but solves the remaining freewheel shortcomings.

Furthermore, the 68Nm (or 85Nm depending on which chart) rating for this bearing is nominal continuous, a deeper dive indicates maximum 2.5 times nominal for short periods. That's about 126 foot pounds, or more. Should be plenty to get my battery/motor setup power to the ground with smooth throttle use.

 

[PaPaSteve]

Looking good.
My two cents ?
The only thing I might have done different is :
Not use the bearing clamping bolt as that's a likely point of slippage and definitely will distort the roundness of the outer bearing race.
Alternative method ?
Machine the ID of the housing a little small.
Welding housing to sprocket. (Welding distorts the part)
Final machine to .0005 / .001 inference fit.
Freeze the bearing.
Warm the sprocket assembly to 500F.
Drop bearing in place.
The warm / freeze trick is how bearings in a motorcycle engine are replaced.
Shrink Fit tooling is common for CNC milling as it's more precision and strongest way to hold the cutting bit.

But : what's done is done, give it a try.

 

[thepronghorn]

Here's another thread discussing using a sprag clutch for a left hand drive.

NuVinci Left Side Freewheel

I'm considering designing/producing a left-side freewheel adapter for the NuVinci N360. The intent would be to reduce noise and drag for designs like my folding Dahon Jetstream and the soon-to-be powered 2009 Norco A-Line The idea is to replace the disc brake mount with a machined 6061 adapter...

endless-sphere.com

 

[lnanek]

There seem to be off the shelf options too:

Tech

Video: How an Onyx instant engagement hub works Sprag Clutch – Onyx Racing Products uses sprag clutches in its rear bike hubs. Sprag clutches are unique because they engage instantly but operate silently. In addition, all sprag elements maintain gentle contact with the driver and hub body...

onyxrp.com

Silent coasting seems like a safety issue to me, though. Electric cars even have to add speakers and playing sounds while going slow to solve that.

 

[Chalo]

Silent coasting seems like a safety issue to me, though. Electric cars even have to add speakers and playing sounds while going slow to solve that.

I like my coaster brake single speed bike because it's silent. But it's also slow (mostly because of the motor's limitations), which makes it pretty safe.

 

[PaPaSteve]

There seem to be off the shelf options too:

Tech

Video: How an Onyx instant engagement hub works Sprag Clutch – Onyx Racing Products uses sprag clutches in its rear bike hubs. Sprag clutches are unique because they engage instantly but operate silently. In addition, all sprag elements maintain gentle contact with the driver and hub body...

onyxrp.com

Silent coasting seems like a safety issue to me, though. Electric cars even have to add speakers and playing sounds while going slow to solve that.

Beautiful craftsmanship on those hubs.
Adding a bell or radio would be easy enough if sound is desired.

 

[Barncat]

PaPaSteve- not only am I familiar with hot/freeze interference fits, I've used the technique to replace broken exhaust valve guides in cylinder heads. That approach would render the mechanism very difficult or impossible to remove from the wheel as wrench access to the inner thread-on hub is blocked until the outer hub/sprocket is removed. Plus you still need a key. Set screws are very common for key retention. Any bearing out of roundness is demonstrably negligible- as the wheel sits installed in frame it spins with less drag than a loud clicker freewheel, and the sprags engage properly. As mentioned, next time I'd bolt the sprocket to outer hub, but that complicates the machining...

Inanek- yes I'm aware of Onyx hubs. Nearly 500 bucks US, plus a custom wheel build, plus I don't see one compatible with a simple single sprocket. So no... And silence is golden IMO. "On your left" works for my rare bike path use.

thepronghorn- thanks for pointing out that decade+ old thread. News to me, but not surprised. The CSK style bearing is the only feasible solution to a DIY overrunning sprag clutch. One poster there claimed to have built a motor mounted CSK25 part, with no photo documentation, and the OP there had a special purpose prototype that ended so far as we know iterated in useless printed plastic.

My bike remains to be powered up. The only failure modes are keyway metallurgy, and possible under torque rating. The upsides are obvious as stated in my first post, plus universal fitment, plus a non moving chain while coasting.

 

[PaPaSteve]

PaPaSteve- not only am I familiar with hot/freeze interference fits, I've used the technique to replace broken exhaust valve guides in cylinder heads. That approach would render the mechanism very difficult or impossible to remove from the wheel as wrench access to the inner thread-on hub is blocked until the outer hub/sprocket is removed. Plus you still need a key. Set screws are very common for key retention. Any bearing out of roundness is demonstrably negligible- as the wheel sits installed in frame it spins with less drag than a loud clicker freewheel, and the sprags engage properly. As mentioned, next time I'd bolt the sprocket to outer hub, but that complicates the machining...

And silence is golden IMO. "On your left" works for my rare bike path use.

Seems well thought out
How do you remove valve guides ?
Hammer and punch or heat until they drop out ?

I've found "On your left" sometimes confuses the pedestrian and they step left directly in to my path.
"Behind you" always results in the person stepping off the path resulting in less conflict.

Power ON !

 

[lnanek]

Yeah, pedestrians misunderstanding "on your x" is annoying. I used to get that a lot too. All my bikes have bells, but when I really must call out, like when riding my wife's bike or a random 2x derailleur one with no room for the bell, I say "passing left" or "passing right", which I find works better.

If cost is the primary concern, I feel I could fake instant engagement with a pawl freehub in software. Sort of like how Grin's virtual electronic freewheeling feature always supplies some power to direct drive hubs so they coast more freely.

We could similarly just have the mid drive controller always keep some power on. Enough to keep the pawls engaged when stopped, but not enough to pull the bike when brakes are on or feet are down. Start performance would be higher too since sprag hubs are notoriously spongy. Takes a while for friction to rotate the sprags fully into position to achieve maximum lock, I guess. Whereas a pawl freehub with software to keep it always rotated forward to where it would catch would give you full throttle immediately. Most of my ebike motors are sized such that starting from a stop light at full throttle is the one spot where they lack performance. They can easily reach max legal speed and still be slow at crossing an intersection from a stop.

I've had clutchless CVT IC powered scooters that were like that and it wasn't a problem that the motorcycle wants to walk forward slightly without any throttle. Handlebar did have an engine cut out switch, though, and kick stand also cut the engine. So far more ways to kill the engine than my ebike where I have to feather the brake lever to tell the motor to cut. My usual controls position for my ebike at a red light is wide open throttle and feathered brake to prevent it from doing anything. Releasing the brake then gives you instant full throttle without the delay of having to twist the throttle. So instant engagement without the sponginess of sprags would be useful.

 

[Barncat]

PaPaSteve- re valve guides, the new ones go in the freezer overnight, the whole cylinder head goes in the oven for an hour at around 300*F as I recall. Either buy or make a custom fitted heavy punch on the lathe from a grade 8 bolt that inserts about 1.5" into the guide up to a shoulder. Put on gloves, drive hot broken guides out of head and immediately drive chilled new ones in, whilst praying your aim is good. It's very nerve wracking. Carefully chase the bores with a lubed hand reamer.

Inanek- there may be a slight perception of sponginess as the bearing engages, remains to be seen. If so would be analogous to motorcycle hub rubber cush drives, and of course superior to the "thwack" of conventional freewheel lash. Trying to manipulate lash electronically is complicated, not in my skill set, and would cost energy.

One thing I've not noted yet is that this EC4P motor has very high cogging torque, as in you need to wear a glove to turn the 11T sprocket by hand. Hence regen braking is basically out and freewheeling required for optimal energy budget...

 

[PaPaSteve]

A trick for removing / replacing bearings in motorcycle wheels or engine cases :
Remove anything that could be damaged by the heat, gaskets, seals, tire etc.
Place the part on a hot plate sourced from a second hand store ($20)
Leave a gap for the bearing to drop.
Run heat at highest setting - 500F
Cover with welding blanket.
Wait until the sound of the bearing dropping on plate, usually about 20 minutes.
Have the ice cold bearing(s) ready for the reinstall.

 

[Barncat]

Back to the matter at hand. I put my 15s battery pack back on the bike today after finishing up wiring the controller etc... no reason to delay testing til I get the new 20s pack built.

I'm pleased to report that my sprag bearing setup works exactly as designed during several quick miles around town. Instant smooth engagement, no slippage under full throttle up moderate grades, no sponginess, silent coasting. Obviously need to put more miles on it, and with higher power battery before any final pronouncement.

 

[Shockwave]

I love this style of engineering, great job