Centrifugal clutch for intermittent assist, *regen* capable motor?

[BalorNG]

I know people have built centrifugal clutch equipped bikes in the the past, to overcome 'sensorless start' issues with some fine electric motors like astro, and they are otherwise useless on electric motors.

However, I want a system that can isolate a motor from transmission, be reasonably light, and most importantly allow regen braking as well (my end goal is 500+ km rides in hilly terrain on very heavy, but very aero recumbent, so YES, this is absolutely viable, I've run the numbers - hopefully I'll get the shell finished this winter).

Minimoto clutches/gearboxes like this:

https://aliexpress.ru/item/32810843398.html
Seem a very good 'two in one' fit - 4.5 first stage reduction, relatively light and compact, doubles as shaft support! Designing and ordering laser cutting of an adapter plate is really not hard.

Problem is, however, that typical minimoto clutches are press-fit over a tapered 15-12mm axle and require a special instrument (clutch puller) to uninstall.

THIS, however, seems very interesting:
https://aliexpress.ru/item/32949414652.html

For 10mm shaft with a keyway!
You'll still have to get creative a bit with installation, but 8080 motor with 12mm shaft stepped to 10mm and keyway machined seems an excellent fit. I presume massive springs means it engages at very high RPMs, but than replacing the springs with lighter ones should be easy.

A question though - did anyone here experimented with such an arrangement? I know this is suboptimal when it comes to starting (lots of waste heat from pad friction engaging halfway), but I'll have a typical freewheeling motor for that, this will be relegated mostly for assist uphill AND regen downhill. Once back on the flat, slowing down a bit and commanding high regen current should instantly disengage the clutch from the drum... to prevent the same during braking, I intend to use 'duty cycle mode' of the VESC that keeps constant RPM, varying regen/assist watts to keep it constant.

This seems better than a friction drive - while heavier and more complex, this should be much more efficient in actual operation: (provided that I'll do most of the braking/acceleration on RPM range where clutch is solidly engaged), no tire wear, no rampant slipage when wet.

Did I miss something? After all, this system was never intended to be 'regen braked', IC motors are incapable of that, but than the pads should be indifferent to force vector, and I doubt that there are helical gears anywhere on that reducer...

 

[sleepy_tired]

I don't have any centrifugal clutch experience with electric motors, but I have some limited experience with gas driven bicycles from years ago. So take what I am saying with a grain of salt.

It is my experience that unless you are driving the clutch with the motor the times they choose to disengage is rather random, or at least uncontrollable. Like when you are coasting down a hill taking advantage of engine braking. Sometimes it works, sometimes it doesn't.

A option for a simple manual clutch is to use v-belts with a manually controlled tensioner. This is what is used on riding lawn motors and smaller tractors and things of that nature. I've seen this approach used on home-made mini-bikes and go-karts. So it seems suitable for fairly high-torque applications up to 15-25 HP or so.

The basic idea is simple. Pull the tensioner away and with no pressure there is no friction and the motor can spin freely. Apply pressure from the tensioner and it locks up and can easily provide reliable torque transfer either direction. It doesn't allow for the smooth take-up that you see with a proper motorcycle or automobile clutch, but it works.

---------------------

However a superior approach is probably to ignore the issue of the clutch completely and just tackle the problem electronically.

If you can setup a "Electronic Virtual freewheeling" by feeding the motor 1 amp or so when you are not using it then that will allow you to coast and pedal like there is no motor there. The losses are more then made up for with the regenerative braking. Ends up being more efficient in the long run.

The only controller I am aware of that supports this natively is phaserunning/baserunner from Grin. But I am sure that there are different ways to tackle the issue. Like a simple addition to a hand throttle to give the controller a small amount of throttle signal or something like that.

 

[BalorNG]

I don't have any centrifugal clutch experience with electric motors, but I have some limited experience with gas driven bicycles from years ago. So take what I am saying with a grain of salt.

It is my experience that unless you are driving the clutch with the motor the times they choose to disengage is rather random, or at least uncontrollable. Like when you are coasting down a hill taking advantage of engine braking. Sometimes it works, sometimes it doesn't.

A option for a simple manual clutch is to use v-belts with a manually controlled tensioner. This is what is used on riding lawn motors and smaller tractors and things of that nature. I've seen this approach used on home-made mini-bikes and go-karts. So it seems suitable for fairly high-torque applications up to 15-25 HP or so.

The basic idea is simple. Pull the tensioner away and with no pressure there is no friction and the motor can spin freely. Apply pressure from the tensioner and it locks up and can easily provide reliable torque transfer either direction. It doesn't allow for the smooth take-up that you see with a proper motorcycle or automobile clutch, but it works.

---------------------

However a superior approach is probably to ignore the issue of the clutch completely and just tackle the problem electronically.

If you can setup a "Electronic Virtual freewheeling" by feeding the motor 1 amp or so when you are not using it then that will allow you to coast and pedal like there is no motor there. The losses are more then made up for with the regenerative braking. Ends up being more efficient in the long run.

The only controller I am aware of that supports this natively is phaserunning/baserunner from Grin. But I am sure that there are different ways to tackle the issue. Like a simple addition to a hand throttle to give the controller a small amount of throttle signal or something like that.

Yea, I've been considering that, but than V-belts are kinda lossy in operation, and I doubt it is a good idea to use a toothed belt in this fashion... though when used intermmitenly *some* losses can be swallowed, provided that benefit from disengaging the motor is higher than losses incurred during functioning of the motor.

Regarding the clutch disengating at random when braking - yea, I've thought of that, that's why it is important to use not 'raw regen', but 'duty cycle mode' on the VESC - so if clutch slips, the motor will pick up instead instead of carrying speed down to zero and completely disengaging it (unless this is what I want myself). Again, combined with vesc and MCU this is really easy to program. bit how it will work in practice... well, will need to be ascertained in practice.

Regarding 'electronic freewheeling' - I've already implemented that on a VESC, that's very easy... however, losses are losses, even if you do not feel them when pedalling.

My goal a recumbent that can tackle hundreds of miles in one go, pedalling all the while, and my ultimate goal is 'offgrid touring', powered by solar panels - but than since I'm from Russia, not Australia/California sun cannot be always relied upon, distances are huge and terrain is varied - there are stretches of flatland followed by 'sine waves' of hills where efficient and powerful assist/regen is an absolute must on something heavy and aerodynamic.

My preliminatry experience that even with aero better than a typical e-bike and huge (170+ kg overall, most that being me unfortunately), on flat ground regen figures are pretty negligible and do NOT compensate even for motor drag, unless this is 'stop and go' city traffic - but than my goal is not city riding, I can use an unpowered bike for that...