cam profile precision/wear

[wangp1283]

There is a cam profile I don't know if it's possible to do in real life. For one thing, it needs to be highly accurately machined.

If we take the circular cam profile and lay it "flat", it would be like a upward parabola, so that if a point goes from the left to the right at ,say 10 m/s. The cam profile will force the point to accelerate upward with an acceleration of 1m/s^2. This mean the parabola is very "flat". In fact, the highest point on the parabola is on the order of maybe 0.5mm.

Therefore, this take a lot of precision machining. I wonder if that's feasible, and how exact can it be.

Also, we also have to take into account of wear. Assuming the "point" that's moving across have a spring attached to it from above that "forces it" down on the cam, and the spring exerts 50lb. It doesnt' slide on the cam, but rather rolls like a bearing, well lubricated.

How serious will the wear affect the camprofile over the long run?

 

[GregLocock]

I think you need to revisit your basic dynamics equations.

A cam that could reach 10 m/s with an acceleration of 1 m/s/s would need to be 7m tall.


Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[wangp1283]

no, what I mean is the tangential velocity is a constant 10m/s. and the cam need to produce an RADIAL acceleration of 1m/s^2 (hence a parabolic shape) from a certain angle range.

 

[magnograil]

A roller cam follower is going to produce a lead/lag contact angle on the cam which, if you are trying to get a particular response, you have to account for. A parabolic cam profile will not produce a parabolic response on a roller follower unless the roller diameter is zero.
Grinding it is easy if you have a cam grinder. A master is made, usually at a larger scale, and the grinding wheel is moved into the blank cam as it turns. A CNC mill can produce a master with tolerances down to 0.0005" or better.
As far as wear, if there is no slipping, the only wear of the cam surface would be caused by fretting. That should be almost non-existant if both parts are hardened and polished.

 

[DaveKillens]

Can you incorporate a lever, so that you can get some kind of ratio between the cam and device it is supposed to actuate? I am assuming you have some vision of a cam that has lobes practically non-existant? With a lever, you could have those lobes four times larger, yet you could still get the desired movement of the valve(s).

 

[wangp1283]

The reason I'm asking this question is because I'm trying to build a CVT that uses a teeth mesh rather than friction.

In a regular CVT, for a given rpm, if you increase the radius, the acceleration is (rpm * the rate of change in radius of the output element). Let's call this the NORMAL acceleration.

If you have a regular frictional cone and a regular belt, then their point of contact will remain fixed and as you shift, you only have the NORMAL acceleration.

However, in my design, the point of contact between the conical element and the chain is not fixed, rather, it rotates along with the cone, just like a teeth on a gear will rotate with the gear, unlike the point of contact of a frictional drive (which stays fixed).

This means as I try to shift my CVT, in another word, as the radius of the output element gets changed, the acceleration produced will be the NORMAL acceleration PLUS an extra acceleration which I call SHIFTING acceleration. This SHIFTING acceleration is a function of the rate of the shift and the angular speed of the engine at the time.

This "shifting" acceleration is undesirable. For one thing, it makes the CVT harder to shift (extra energy used to produce the SHIFT acceleration).

I've already designed a cam profile that can "cancel" this SHIFT acceleration so the net acceleration will be only the NORMAL acceleration. Basically, the cam will produce a "counter" acceleration by varying a the portion of the chain between the engine and the load. This means the tension of the chain will never exceed a certain level, due to the superposition of the acceleration.

So the equation is : Acceleration on the conical teeth (NORMAL ACCELERATION + SHIFT ACCELERATION) - Cam compensated acceleration (SHIFT ACCELERATION) = NORMAL acceleration. (which is the acceleration under a normal theoretical, perfect frictional CVT with no slipping)

However, due to the imperfect nature of manufacturing, but most importantly, due to the segmented nature of the chain, it's basically impossible to cancel it out perfectly.

This means while the system is trying to shift, the net acceleration will always be slightly more or less than what it should've been if using a frictional drive.

This has a few potential problems, which I'm not sure if it's something to be worried about.

1. If the camshaft overcompensate the SHIFT acceleration, the net acceleration of the load will be less than what it should've been. This might mean inefficiency as engine rpm will tend to increase and this energy will be lost to the pumping loss of a gas engine. Then the load will have a smaller final velocity than the engine and the next time around, it will drag the engine back again.

2. If the camshaft undercompensate the SHIFT acceleration, the net acceleration of the load will be more than what it should've been. This will tend to slow the engine down. The tension in the chain will increase and the load will have a higher speed than the engine. So the next time around the load will "force" the engine speed up and some energy will be lost to the pumping loss. maybe...

It'll be nice to have a spring or other flexible device that can match the tension of the chain to the required theoretical tension of the chain under any operation condidtion in real time.

But this is hard as the engine's output varys so much.

Any solutions?

Since I want the system to be as efficient as smooth as possible. I wonder what my option are.

 

[wangp1283]

If a linear movement of a rack is 3mm. And this rack is connected to a
rack and pinion system, which is connected to a gear train with a gear
ratio of 1:60.

At the other end of this gear train is also a rack.

So basically, a linear movement (input) will theoretically cause a
linear movement (output) that's 1/60 of the input.

So in this case, the output rack will move 3mm/60 = 0.05mm. In theory.

But can this be expected in real life? If the system is under load and
it's unidirectional (with no backlash), how accurate can it be?

Does the result depend on any variables such as teeth size, teeth
machining accuracy, number of meshes...

Thanks.

 

[magnograil]

The whole point of an involute gear is to provide a continuous rolling load path between input and output. Gears cut on a hob are generally accurate tooth to tooth but have pitch diameter variation that is dependant upon the operator. Steering rack and pinion gears are preloaded to reduce backlash. For such a small movement, a screw thread drive might be more accurate.