A couple of years ago I came up with a self releasing ramp clutch to disconnect an engine from the dynometer. I was contracting at a UK university and the idea was turned into a Msc student project and worked out very well.

I actually made the clutch but the student who did the modeling came to me about how to model the ramp. I tried to explain the shape to him but time was short so we went with a simplified model.

The actual manufacture was no problem as it is just a 10mm slot drill with a 1mm rad plunged to depth and then spiraled out on a cnc mill.

I am redesigning the clutch myself so have come up against the same problem the student did.

I have done a rough ramp which can be seen in the attached jpg, Is it impractical to expect to be able to model a perfect ramp, as a different part of the cutter is in contact with the surface at all times?

Any suggestions would be much appreciated.

Very good solution, thanks, I have tried a number of different ways and also thought about using the tool path but what I am trying to do is find the modelling way to do it.

I have been very close but have problems with the take off area as the rad of the tool makes it very complicated. It's kind of a paradox, if the tool started at a position through the part in fresh air it would be an easy surface.

I couldn't resist an attempt at modeling this device. Of course, I don't know any of your dimensions or specs, so I had to rely on eyesight calculations.

My approach was to model the overall piece and then model cutouts for the ramps. They look something like this:



Here, positioned on the model:



Finally, after final positioning, I used the Solid Subtract tool:



Here is a close up of the finished ramp:



Is this sort of what you are looking to do?

The ramp is curved if the tool's lead axis is parallel to the flywheel's. The ramp can be machined with a planar floor if the tool's lead is perpendicular to the flywheel's lead helix. If the clutch element's are balls, the contact area is instantaneous, if they're rollers, there's a contact line that will need a flat floor. This type of clutch gives a theoretically-analogue takeup, but it's actually dependent on contact pressure. Sprag clutches are more reliable, but have a degree-increment step between takeup points. A US company called Epilogics patented a notionally-sprag type of clutch that works in the same plane as yours, that they call "mechanical diode". It's simpler to machine, takes immense torque throughput with less endthrust, and can be designed with sprag-and-receiving-slot arrangements that use Vernier-style relationships to make the takeup increment angle minute, almost-analogue.

http://www.epilogics.com/md/index.htm

Almost. The question here is how to model the ramps so they'll have the same geometry as a simple manual or CNC mill setup will produce.

Because it's cutting a ramp, the last part of the cut is done by the back edge of the tool and that needs to be reflected in the shape.
Here's my (hurried) go at getting it.

http://www.screencast.com/users...6-49a6-8687-7b10b54bdf03


Martin.

Not true, as the z axis is moving in the positive direction the back edge does not come into play, if it was moving in a negative direction then both sides would cut.

Yes, but the line of contact between the pin and the groove is on a plane that is normal* to the slope of the groove. Modelling the grove by taking the profile of the tool will give the wrong groove profile because the leading edge of the tool is ahead of the widest part as it cuts.

*I'm not sure that this is exactly true, it would be if the tool end was spherical, but for a rounded cylinder the profile you want may be defined by a plane that passes through the tangent point where the leading edge of the tool meets the slope of the groove and the transverse diameter at the bottom of the cylindrical section of the tool.

Martin's solution is accurate.

I haven't tried this, but can't you make a copy of the tool and turn it into surfaces and position it at the bottom end of the ramp - then remove and/or trim any unneeded parts?

**Edit** Demo file added.

Martin.