Can you create a limited slip differential using only gearing?

[wimms]

I'm not educated in mechanics, and there is one nagging idea I can't get answered that's related to limited slip differentials.

I understand all types of diffs, and while learning them and requirements for best traction came to one idea about possibly good LSD, but, I can't imagine how it would be built. Actually, I'm not even sure if its not against some mechanical laws, although I don't see any violations.

So, basically, idea is this. Take open diff, and connect output shafts to each other with bidirectional reducer gearing. This needs elaboration, as I'm well aware straightforward approach would result windup and lockup.

Basic idea behind it is that when one of shafts speeds up due to loss of traction, the speedup is not stopped like with locking LSDs, but the speed difference is transferred to other shaft via reducer, like say 1.5:1. Necessary twist is that same behaviour must be present symmetrically for other shaft, ie. reducer effect should be present only when speed difference exists. Opposite behaviour, increaser 1:1.5 effect should be somehow avoided. Nonlocking transfer is needed to avoid frictional heat, unnecessary lockups in normal turns, ability to constantly differentiate and yet proportion torque.

Obsession is to achieve this behaviour without using any sliding frictional surfaces, ie without clutch, but use only gearing. Closest to frictional surfaces could be worm gearing, like in Torsen diffs.

Torsen is amazing device seemingly doing impossible - redirecting torque away from shaft that offers less resistence without any active clutching. Its nice device, but it has drawbacks that should be cured by idea above.

Questions I'm having for this symmetric two-way reducer are:
- are there any mechanical laws that forbid such device in principle?
- if yes, then I'd like to understand fully why.
- if there are no mech laws that forbit it, then does such thing already exist?
- Do you have any ideas of what's a closest gearing arrangement to look at?

 

[GENIERE]

I can think of a possible means. It would require 2 torque sensors, 2 conical gear sets, and 2 hydrolic or electrical actuators. If you want me to expand on this, let me know.

 

[russ_watters]

If I understand what you are asking, it sounds like what is already done with differential gears. I don't remember all of the terminology, but the net effect is if you spin one wheel one way, the other spins the other way without the drive shaft spinning. Or if you have the drive shaft turning you can hold one wheel and the other will still spin. Ever own an RC car? They work like that (good ones anyway).

 

[wimms]

Russ, you're underestimating me. Of course I know how normal open diff works. I want something different. I want that when you spin one wheel one way, then other wheel spins in same direction, but transferring torque in fixed ratio.

I can imagine one-directional solution easily with planetary gearing, but it would work in opposite direction as increaser. And crossconnecting two such reducers would wind up and lock axles.
Obviously, the nonavoidable component is input shaft of the diff, that offers the torque in the first place, and should be part of proportioning and kind of reference point. Torque transfer between axles would be 0 when they rotate with exactly same speed.

Problem with open diff is that when one wheel slips, all torque flows out it, nonslipping wheel receives only half of what spinning wheel can support. LSDs stop that by plain locking output axles, but here I try to imagine device that doesn't lock, but simply transfers rotation to other axle with ratio. It has several benefits compared to locking, and I try to find out if this could be achieved with purely mechanical means.

Of course one could do that within active diff, with electronically controlled clutches, but that makes diff quite complex and costly.

Open diff is necessary component in the device I try to figure out, but in addition, some mechanical gadget that doesn't allow opposite rotations you mentioned, only same directional rotation together with driving gear and other axle. Open diff gearing as normal proportioning of 50/50, and the gadget as limiter to axle speed difference. In total, such beast would allow normal differentiation to take turns, and would immediately transfer torque to nonslipping axle when one of wheels spins.

GENIERE, please, all your input is very welcome. I'd like to be able to visualise what you mean. Though, actuators seems like requiring active diff.

 

[GENIERE]

What I envision would have to be a smart system. I can see no other way to achieve what you’re considering.

Just to state the obvious, a conical gear meshed with a standard gear would have to be on a shaft that is angled to the shaft of the mating gear to achieve full mesh. Sliding either gear on its shaft would provide an infinite range of ratio between the extremes. Each wheel would have its own set. A linear motor or a hydraulic actuator would vary the ratio of each to achieve the desired effect; speed up one while slowing the other. This would also have to be tied in with the cars steering.

In essence one monitors the RPM of each drive wheel as well as the steering angle. A speed difference between the wheels would be converted to a voltage (error voltage), amplified and sent to the actuator that would vary the gear ratio until the error voltage goes to zero. This voltage would also be changed by the steering angle so as to allow one wheel to speed up while cornering.

Regards

 

[wimms]

Well, I don't think conical gear is very obvious. Gear ratio is determined by number of teeth, not diameter of gear. You can't smoothly vary number of teeth, they wouldn't mesh reliably. Conical could be only frictional smooth contact surface, as in variators. But such setup would have limited torque strength..

Very clever device is Torsen. If you have any interest, I suggest to take a look at it. Its purely mechanical device, and offers variable torque split on axles depending on traction available.

Anyway, I guess I wanted to hear some sort of mechanical law or principle that would state such device possible or impossible in principle. If anyone has some good pointers to mechanical gearing theories or somth to that kind, I'd be thankful.

 

[Bobby_302]

Universal gears

I have a design for a set of gears with variable gear ratios. This idea is strictly experimental. However, I do believe that they will work. If you would like to find out more e-mail me at bolindsey@earthlink.net