Mechanism that engages when turning both ways, but releases when not driven

[Stormer]

TL;DR Summary

I want a mechanism with the "output shaft" being able to freewheel as long as you are not turning a hand crank, but when turning hand crank in either direction the input and output should be locked together.

I have an application in mind with a hand crank mechanism that will usually not be engaged to a trugh shaft so it can freewheel, but as soon as you start turning the hand crank in either direction it should engage with the shaft. And i want it to be silent (no mechanical ratchets) and with minimal slop between when you start to turn the crank and when it engages the shaft.

Is this possible?

The first thing i was thinking was to use two sprag clutches around the driven axle mounted in opposite directions with the outside race connected to the hand crank. And then some kind of friction mechanism that engages and disengages the springs in the sprag clutch when you start to turn the hand crank by a few degrees (friction pushes the sprags out, and when they contact the axle they bind and that takes care of all the forces from there and when you let go on the hand crank wheel some spring mechanism gets it back to neutral position so the sprags no longer contacts the axle so it is freewheeling again). The key thing here is that it should work in both directions so just a normal sprag clutch mecanism will not work. And it should be able to turn the hand crank all the way around as many turns as you want.

 

[Baluncore]

The first obvious solution is to require the crank handle be pushed axially against a light spring to engage a clutch with the motor shaft. The turning of the crank would then rotate the motor. If the engagement was achieved with a pair of crown gears with a 20° contact angle, mounted co-axially face-to-face, then the crank would disengage automatically when the motor was powered.

A more complex way is used on the pull start of two stroke motors. That expanding clutch inside a cylinder can be altered to operate in both directions. The system relies on friction against the housing, and rotation of the "crank", to drive a pair of engagement dogs out against what is usually a plastic liner.

 

[Stormer]

I don't want there to be any other movement to activate the clutch than the rotation of the crank. So the first suggestion does not work.

And i want it to work no matter how slow you turn the crank so a centrifugal clutch does not work either.

 

[anorlunda]

I don't want there to be any other movement to activate the clutch than the rotation of the crank. So the first suggestion does not work.

And i want it to work no matter how slow you turn the crank so a centrifugal clutch does not work either.

Please explain better. Suppose the device is initially freewheeling at 100 RPM. You then turn the crank at 10 RPM. Is it supposed to come to a crashing slowdown from 100 to 10? If no, what should it do when crank speed and freewheel speeds do not match?

 

[Stormer]

Please explain better. Suppose the device is initially freewheeling at 100 RPM. You then turn the crank at 10 RPM. Is it supposed to come to a crashing slowdown from 100 to 10? If no, what should it do when crank speed and freewheel speeds do not match?

Yes if both turn in the opposite direction it will be a instant crashing reversal. But in my application that will not happen because the motor will not turn when you are going to start rotating the hand crank (in reality there is not even a motor, that is just for illustrative purposes, in reality it is kind of another crank fixed to the shaft).
And if both turn the same direction the fastest one will set the speed. Just like the freewheeling hub on a bike.

 

[Dr.D]

Yes if both turn in the opposite direction it will be a instant crashing reversal.

Good luck with that! Are you acquainted with the concept of angular momentum?

 

[Stormer]

Good luck with that! Are you acquainted with the concept of angular momentum?

Did you read any further than that part?

 

[Filip Larsen]

If a non-zero rotation of the crank axis is the only input you want the mechanism to engage the drive shaft on, then the rotation alone must be converted into the a clutch engagement. I have no idea how this can be done as a passive mechanism (perhaps something converting a small percentage of the crank rotation into hydraulic pressure applied to the clutch), but an active (electronic) solution seems doable with a controller for an (electric/magnetic) clutch using a crank axis rotation detector to engage/disengage the clutch. If its a "normal" friction-based clutch (rather that direct gear engagement) it should also be able to handle some range of RPM differences at time of engagement.

Have you considered when the clutch should disengage? Engaging the clutch on non-zero rotation kind of implies disengagement when rotation is back to zero (for a while?) but is that what you want? An electronic solution has the benefit of offering more flexibility for the controller, e.g. its easy to imagine adding a disengage button that can be used to disengage the clutch when engage at speed.

 

[Rive]

If you want a hard lock with being sensitive for low rotation then maybe you can try something on the line of seatbelt locking mechanisms?

 

[Averagesupernova]

The requirements are vague and in my opinion not consistent. At first, any slight movement of the crank is supposed lock them together. Next it's compared to a bicycle overrunning clutch. Make up your mind.

 

[Stormer]

The requirements are vague and in my opinion not consistent. At first, any slight movement of the crank is supposed lock them together. Next it's compared to a bicycle overrunning clutch. Make up your mind.

How is that not consistent? Does not a bicycle freehub lock up with a slight movement? A normal ratcheting freehub has a maximum of 6-20 degree pickup angle depending on the number of teeth, and a sprag clutch freehub has almost zero degrees of pickup angle.

 

[anorlunda]

How is that not consistent? Does not a bicycle freehub lock up with a slight movement? A normal ratcheting freehub has a maximum of 6-20 degree pickup angle depending on the number of teeth, and a sprag clutch freehub has almost zero degrees of pickup angle.

No it does not lock up. If the bike has coaster brakes, reverse pedaling engages the brakes. Brakes take time to stop the bike, but a sudden lockup would destroy everything.

 

[Stormer]

No it does not lock up. If the bike has coaster brakes, reverse pedaling engages the brakes. Brakes take time to stop the bike, but a sudden lockup would destroy everything.

Yes it locks the cassette to the wheel when pedaling forward (faster than the wheel speed). If it did not you would not be able to pedal the bike... Anyway this is just pedantic arguing and is totally off topic and does not help answer my original question at all.

 

[Averagesupernova]

Anyway this is just pedantic arguing and is totally off topic and does not help answer my original question at all.

With the original problem stated as it is you cannot expect much.

 

[Baluncore]

@Stormer
If the motor is operating and someone touches the crank, what do you want to happen ?
Why does it require a crank ?
Why not just take better control of the motor ?

 

[votingmachine]

It seems that if you add a second belt to your crank and make the engagement depend on that, then the engagement is strictly crank dependent and the shaft is free-wheeling otherwise.

Another option is that the pedal is also the same as the hand-brake on a bike, which friction-ally clamps on the wheel. It doesn't matter what direction the wheel is spinning, the brake "engages" either way.

Continuing a bicycle analogy, the crank-handle is essentially a "Brake" that is unclamped from the wheel (which is fixed to the motor axle). When you push on the crank, you squeeze the "brake" which then allows torque transfer. The direction doesn't matter.

EDIT: I think it is easiest to separate the clutch functionality from the torque. You expressed it as a wish for a torque engagement. Another possibility, in your drawing the Hand crank is on an internal rotor, and the belt is on an external rotor. Squeezing the handle engages the clutch between those.

Another possibility is that the hand crank torque is on a cam shape, that engages the "pin". It free spins a small amount but then locks the two together. That probably requires a spring to reverse the cam. As long as you apply torque greater than the spring, the cam engages the pins. when you stop applying torque the spring pulls the cam backwards (a spring for each direction). There seems a flaw in this one ... I'm not sure of the disengagement. It seems like it could fail with friction overcoming the spring ... although with a good expanding gear-tooth design the friction might work for you.

I would have to sit and draw that design out. I keep thinking it just jams and stays engaged.

EDIT, EDIT: I just realized I am re-imagining your sprag clutch thinking. And my intuition tells me that has a jamming problem that is hard to overcome.

 

[cmb]

Summary:: I want a mechanism with the "output shaft" being able to freewheel as long as you are not turning a hand crank, but when turning hand crank in either direction the input and output should be locked together.

Is this possible?

Yes, but

I don't want there to be any other movement to activate the clutch than the rotation of the crank. So the first suggestion does not work.

And i want it to work no matter how slow you turn the crank so a centrifugal clutch does not work either.

... no, it is impossible for it to be operational at zero speed.

The explanation is quite simple if you consider the transfer of positive and negative torques. What you want is to transmit a positive torque from input to output but not negative torques from output to input. That is OK, it is a one way clucth.

But then you ALSO want to transmit a negative torque from input to output but not positive torques from output to input. You should see that is not compatible with wanting to transmit a positive torque from input to output but not negative torques from output to input.

The only way is to make it speed dependent and that it is possible, within a range of hysteresis, so long as the mechanism can detect and act on a differential rotation speed, but for that to be the case there can be no 'zero' differential speed as it would be unmeasurable and could not be acted on.

... that's if I understand your question correctly.

There are means to perform this function electronically. It's done in electric power steering systems, but requires some 'non-zero' (albeit very small) differential between input and output, usually detected with a torque shaft of some means, and measuring the differential angle between in and out. This would be quite complicated in a system that can rotate indefinitely on one direction but I don't foresee it is impossible. (Electric power steering only needs to rotate so-many-turns in each direction and then return to centre, so is easier as an engineering proposition).

 

[Filip Larsen]

There are means to perform this function electronically.

This is also what I hinted at in post #8, but for some reason the OP has only run with the negative feedback in this thread (which perhaps has been a bit too plentyful) and not shown interest in discussing any actual ideas.

 

[some bloke]

In order to make a centrifugal clutch more responsive to slow movements, you could gear up to the centrifugal clutch and then gear back down to the output drive. That way the clutch can be engaged with a slow rotation of the handle, as the clutch will spin much more quickly.

From there I would look at finding some way to apply a braking force if either the output is rotating faster than the input (to slow it down) or if the input is rotated in the opposite direction. The brakes would need to release as soon as the two speeds matched. It might not be such a crashing change, but it would be fairly quick. A free-spinning bicycle wheel can be stopped pretty sharpish by the brakes if it's not being pushed along by momentum.

The clutch would need to be designed so that it does not remain engaged when the input is removed - when you stop turning the handle, you don't want it to keep spinning. As such, your clutch will need to allow the output half to continue spinning whilst the input half slows down.

How you can get the braking mechanism to know the difference between you starting to use the crank and you stopping using the crank is beyond me, I'm afraid!

Further thoughts, prepare for terrible doodles!

This mechanism will require significant torque on the crank handle. How much will depend on the gear ratios and speeds, so it will need tuning to make it work properly, but I think it might be a system which will do what you want.

The way this would work, in my mind: (in all cases, input is on the left and output is on the right)

Crank - this is what you turn to engage the crank to the output, by spinning the centrifugal clutch up (which turns far faster than the crank so engages quickly.

Rotary brake caliper - this is designed so that if the torque between the input and output exceeds what you would expect for turning the centrifugal clutch up, the brakes are applied to a proportional amount. This links to the brakes by the output. This means that if the output is spinning faster or in the opposite direction to the centrifugal clutch, the brakes will be applied to slow it down or stop it. If it spinning slower, it should be calibrated such that this does not cause the brakes to apply. This will take some effort & tweaking! It may need a mechanism to mean that it only brakes if the output shaft is turning in the opposite direction or faster than the input shaft, and not engage if the output is turning slower than the input but in the same direction. It will need to base this off of a stationary outer casing.

Gear up gearbox - makes the input speed from the crank much higher, so the centrifugal clutch will engage promptly from even the slightest input.

Centrifugal clutch - this will allow the two halves to engage and release from one another as the crank is turned.

Gead down gearbox - brings the speed back to the amount you wanted.

Brakes - these slow down the output if it is not the same or slower than the crank, proportionally to the difference. If the crank is turned in the opposite direction, it gets maximum brakes.

Output - whatever it is you want turning & being turned!I hope that this offers some inspiration! It won't respond utterly instantly, but brakes are your solution to slow down and then the crank will speed up!

 

[Tom.G]

Take a look at a bicycle coaster brake. Two of these, facing opposite directions, on the crank shaft may be sufficient for your needs. Or you may be able to modify the concept to build a bi-directional configuration.

The most common documented style is the Bendix Coaster brake that uses brake shoes expanding on the inside of the hub. A variation, that may need less machining, operates similar to a multi-disc clutch, that is a stack of metal discs with alternate ones splined to the hub and axle and free to slide axially a small amount.

Here is a video that gives a quick explanation of the Bendix brake:


video, and others, found with:
https://www.google.com/search?&q=how+bicycle+coaster+brakes+work

Cheers,
Tom