how to produce such a response

[wangp1283]

I need to produce such a response to a load:

an constant acceleration of 0.5m/s^2 for 1/100 of a second.
The initial velocity is not important. What matters is that during that 1/100 second time interval, the acceleration is 0.5 m/s^2

I thought about using a cam to produce this response. Basically, a parabolic camshaft.

The problem is, such a cam would be very hard to machine, because the lobe would be practically non existant.

the distance traveled would be (1/2)(acceleration)(time)^2, which would be (1/2)(0.5m/s^2)(1/100 sec)^2 = 1/40 mm (0.025mm). This is very small distance so a cam lobe is hard to manufacture.

My idea is to use a gear train via rack, pinions, and gears. I will "amplify" the acceleration by a factor or 25 at the cam lobe (so the parobolic lobe is steeper), and then "bring it" down via a gear reduction.

I wonder if this is a way to do it? Please help. Thanks. I'd like to hear your feedback and opinions. This is an interesting and very important (to me) application.

Thanks

 

[sreid]

This sounds to me like a perfect application for a servo. How much mass do you need to accelerate? If this sounds interesting I can supply more details.

 

[wangp1283]

Hi, thanks. I think I'd like to do it mechanically with the way I said above, preferably.

A servo would be too complicated, expensive, and not as reliable, I would think. Also I'm not familiar with servos at all.

What do you think of the way I've mentioned above? It doesn't have to be dead accurate. Around 95% accurate would be good enough.

 

[unclesyd]

Instead of a cam you might try simple gears.

http://www.cunningham-ind.com/index.htm

 

[sreid]

If the movement can be repetative you could put multiple "Bumps" on a cylindrical cam and run the cam at constant speed. A spring loaded roller follower could turn the bumps into linear motion. Grinding a cam to have
0.025 mm height bumps is not hard to do (CNC Grinders) and you could probably hold +/- 2 um tolerance.

Mechanical start and stops in 10 ms are tough to do. I think you idea for a gear up gear down arrangement is probably full of problems (inertia, back lash, compliance?).

Think about servos, they are taught in the Mechanical Engineering departments of most Universites. They are "Servomechanisms" to use the original name.

To give an example of what can be done we maufacturer a machine with a fast tool servo. It is a 2"x 2" airbearing slide weighing 6 lb. including the voice coil actuator. The position feedback is from a 0.1 um glass scale. We run this slide with the following envelope.

+/-0.5 in stroke
+/- 28 in/sec
+/-5500 in/sec^2

24/7 if need be.

 

[wangp1283]

sreid,

the "bumps" on the cylindrical cam idea was what I was thinking. In fact, it doesn't have to be 0.025. I can give it a initial velocity to start with, so the lobe is a lot bigger. Also, I can do the "gear up and gear down" thing I mentioned.

You said mechanical start and stops in 10 ms are hard to do. But it doesn't have to start and stop in 10ms, in fact, I could have it "start" before the required time interval, and have it "end" after the 10ms interval. For example, think of the cylindrical cam as having 0-360 degrees. I need to have that acceleration response when the cam lobe is between 90 degrees to 180 degrees. The rest of the rotation it can act anyway it want, but just as long it offers that acceleration response between 90 degrees to 180 degrees of each turn.

You mentioned backlash. But I'm only worried with the unidirectional acceleration. When it's "coming back down" (reversing directions), it will be out of the 90-180 degrees range so I wouldn't care about that. The load is rather large too, so I wouldn't need a spring of any kind.

I don't know what you mean by "compliance".

I will surely look into servos too.

Thanks.

 

[sreid]

By compliance I mean that fixtures bend and shafts twist. To move 0.025 and not error more than 10% things can't move more than 2.5 microns.

The one big advantage a servo has is that one can change the acceleration, velocity, and position to what ever you want should that become necessary simply by changing the move profile parameters. If this is a test rig of some kind it never seems to fail that the minute you have something working, someone will decide they need a bigger or smaller jolt.

 

[wangp1283]

I can move more than 0.025. In fact, I can move as much as I can, since I only care about the acceleration. Therefore, I can give it an initial velocity to begin with. It doesn't really have much to do with the distance.

Imagine a parabolic displacement curve due to a constant acceleration. If it begins from zero, then the distance traveled isn't a lot. If it starts from a non zero position (with a non zero initial velocity), the change in distance will be greater for the same change in time.

You mentioned the fixture bend and shafts twist. I think this is similar to the backlash. I'm not sure if this will affect the result, since the backlash can be "taken up" in a negligible amount of time.

How much do they bend/twist any way?

 

[sreid]

As an example of compliance, imagine a steel cantilever beam with a square cross section of 1 inch by 1 inch and 12 inches long. Looks and feels pretty substantial. Yet a 1 pound force applied at the cantilever tip will deflect the beam tip 6 micrometer.

 

[electricpete]

I never understood the original question in terms of what are you really trying to accomplish. Maybe you can explain the application some more.

With my feeble understanding I'll throw out some more ways that might give 0.5 m/sec^2:

1- Acceleration of a vehicle (engine in neutral) down a ramp which makes an angle to the horizontal of arcsin(0.5/9.8)

2 - pulleys or levers to accomplish similar conversion of natures 9.8 m/sec^2 to your desired 0.5 m/sec^2

3- Acceleration of a charged beebee between two charged plates.

It would be trivial to come up with more details on #2 and #3 but I'm not sure if that's anything resembling what you want. If it is, just ask.


=====================================
Eng-tips forums: The best place on the web for engineering discussions.