Linkage to take shaft over top dead centre

[jouleflow]

Hi,

I have Two eccentric cams consisting of 100mm dia disks with a shaft hole 15mm off centre.
They are 400mm apart & I wish to rotate them together using two pneumatic cylinders each placed on either side of one disk. They are anchored at one end & have a pivot eye at the other. Each cylinder is attached to one disk at 90 degrees apart via the pivot eye. The discs are connected via a timing belt so I only need to drive one disk. Is there a linkage that will overcome the TDC situation so I can achieve a complete revolution by reversing the air flow. I am considering of something like the linkage on steam train wheels.

 

[Danger]

Welcome to PF, Jouleflow.
I'm afraid that I just can't envision this apparatus from your description. Can you possibly post a sketch of some sort through ImageShack or similar service?

 

[jouleflow]

Hi Danger,

thanks for the reply. Perhaps I should describe the problem as one disk that can rotate on an axle, the hole in the disk is offset to make a cam. However this offset is not important it can be treated as a concentric wheel on a shaft. In the face of the wheel there are two tapped holes through, 90 degrees apart near the circumference. On each side of the disk a pneumatic cylinder is attached via a pivot eye joint to the tapped holes 90 degrees apart. The two cylinders are anchored at the other end but can swivel & are designed to rotate the disk via separate valves. During one revolution there will be two places where the cylinders will be in equilibrium & will not pass over these points. The cycle should be one complete revolution & then stop then repeated when a fresh signal to the valves starts the sequence over again. As the cycle continuously stops & re-starts, there is no flywheel I can use to take the system over tdc. The pneumatic circuit has one valve per cylinder actuated by pneumatic sensors at each end of their respective strokes mounted on the cylinder that is actuated by the valve. If I could overcome the "locking" problem then the circuit would keep running, converting linear motion to rotary motion. I can probably send an AutoCAD file if it helps.

Here is a link to a walking beam feeder I have just built that shows the eccentric motion I am trying to replicate in a simpler way.

http://uk.youtube.com/watch?v=NdmSdFi7CkI

 

[Danger]

I'm not too sure about this, but it seems to me that if your linkage points were not 90 ͦ apart, nor 180, there wouldn't be a stall position. Can you move them to another angle?

 

[BenFur]

JouleFlow,
I am going to assume that your valves have only 2 positions, it would be smoother if there is a neutral 'float' position, but should still work fine. Let's say cylinder A and B are stroked completely out so that pin A is 45 deg before TDC and the pin of cylinder B is 45 deg after TDC. Begin by retracting cylinder B, then when Pin A gets to TDC begin retracting A. When Pin B gets to BDC, shift B to extend. When Pin A gets to BDC, Shift cylinder A to extend, and the cam will end at its starting position, unless you shift B to retract when pin B is at TDC, in which case you would continue rotation. If you have a float center valve, you would start with A extended to TDC and B would be at 3:00 and the cycle is started by retracting B, then A, immediately shifting B to neutral or float position.
I am looking for a way to do the same motion with one cylinder, like a steam engine, but I cannot use momentum to ensure that I continue the same direction. ?

 

[jouleflow]

Hi Benfur

you know exactly what the problem is & your suggestion will work. I have managed to get the system working with two valves & four sensors ( two per cylinder ). The sensors reverse the motion of their respective cylinders by reversing the valve controlling that cylinder. The action is a bit "lumpy" & it is difficult to determine the direction the unit will move & it can also lock at start up. It all depends where the unit was when the cycle was stopped. Regarding the 3 position valve arrangement, this will give a smoother motion but I will need 4 sensors per cylinder
sensor 1 - valve to mid point - both ports to exhaust - signal just before tdc
sensor 2 - send valve to reverse - signal just past tdc
sensor 3 - valve to mid point - both ports to exhaust - signal just before bdc
sensor 4 - send to forward - signal just past bdc
For this arrangement I may need two valves per cylinder as well.

Ideally I would have three or more cylinders - 3 valves & 6 sensors.

For your situation I have seen an answer in the books "ingenious mechanisms for designers & inventors" there are 4 books in the series. It was a long time ago when I first saw the solution & I could not figure out the mechanics then, perhaps you may have more luck.



Greg

 

[BenFur]

Thanks, Jouleflow, I will look for that book.
I guess for your application, I was not tieing the sensors to cylinder positions but to rotation positions. There is a custom set of switches for this, I believe called an Electro-cam box which my company has used before that has adjustable cams, as many as you need, which make switches at any angle of rotation. Each cam can be independently adjusted, then locked into position.
We may use 2 cylinders in our application, similar to what you are doing if a single cylinder application is not feasible.

Regards,
Benny

 

[wolram]

I think the idea must be to have a cylinder with a stroke longer than the center of wheel to cam point dimention, the last part of the cylinder stroke is used to trip it past 180 or 0 degrees.

 

[jouleflow]

The unit uses pneumatic sensors on each cylinder to operate the valves so the unit is entirely pneumatic. Unfortunately I am unable to use any electrical parts.

Regarding using one cylinder:
This can be done if you use spring loaded cams. Assuming the cylinder is horizontal with the left hand end secured but allowed to pivot & the cylinder rod is connected to the "flywheel" with a rod eye & bolt near the perimeter. This will give the basic connecting rod type of motion. To overcome the tdc situation as the rod eye advances it can hit a sprung wedge above & it will gradually compress the spring exerting downward pressure on the rod eye to force over tdc. The opposite can be applied at the bdc whereby the cam is sprung upwards to force the rod eye past bdc. This is assuming the wheel on the right hand side travels clockwise to reverse the rotation the cams will be reversed.
I would have liked to use this system but there is no room.

Regards,

Greg.

 

[Michael_E]

The cycle should be one complete revolution & then stop then repeated when a fresh signal to the valves starts the sequence over again. As the cycle continuously stops & re-starts, there is no flywheel I can use to take the system over tdc.

If you want this really small, I would skip the entire idea of a linkage and go to a rack on the cylinder rod and a pinion on the shaft with a one way clutch on the shaft to allow the rack to move back into the starting position. You'd need the pinion gear to be matched to the exact rotary movement desired on your cam, but this can be scaled to very small dimensions. I've never used these, but these guy's seem to use a similar mechanism in their product:
http://www.rotomation.com/ The nice deal here is that you'd need only one valve to reverse the cylinder and timing would be built into the action of the walking beam/cam movement.

 

[jouleflow]

Thanks for the reply Michael E,

The original unit did use a rack & pinion with a sprag clutch & worked well. It can be seen from the youtube link in my post #3. This unit also required a re set to enable the unit to start from the same position each time as the pickup point on the sprag clutch may drift.
I am presently looking for the most cost effective solution & I may now use a latch system in place of eccentric cams. The motion of the original cams is ideal as I get an almost vertical pick up & place motion but I will probably get somewhere near with a latch mechanism.