How Would I Machine this CAD designed Ratchet

[peter1972Participant @peter1972]

Consider cutting the teeth with a 45° single-angle milling cutter, initially without the cylindrical inner: that can be added later if needed!  For the drawing below showing just the first cut, I have assumed 50mm diameter and 26 teeth.

The aim is to angle the workpiece on the milling table so the depth of cut reduces to zero at the centre of the workpiece but gives the desired angle for each tooth.

The angle at the centre is 360°/26, which gives the dimensions of a right-angled triangle as shown.

At point ‘A’ the depth of cut needs to be 5.98mm if using a 45° cutter. By my calculations, that means the workpiece needs to be angled on the milling table by about 14.3°.

The above should produce sharp teeth with no flat tops: not quite as shown in the opening post. Adjustments could be made to the geometry to give flat tops.

 

[Nigel Graham 2Participant @nigelgraham2]

The above should produce sharp teeth with no flat tops: not quite as shown in the opening post. Adjustments could be made to the geometry to give flat tops.

Or more simply, cut the teeth slightly deep then gently skim the sharp crests off?

[Nigel Graham 2]

On Nigel Graham 2 Said:

Or more simply, cut the teeth slightly deep then gently skim the sharp crests off?

Keep the workpiece angled the same but just cut the teeth slightly less deep?

 

[peter1972Participant @peter1972]

I’ve realised that my 3D geometry is not quite correct, so the 14.3° is not quite correct.

[Charles LamontParticipant @charleslamont71117]

So my initial interpretation was right all along, the pawl’s motion is ideed linear. In that case the geometry of the ratchet is suboptimal. As the ratchet is stationary when the pawl retracts, the sloping face A of the ratchet is not a curved surface but plane. The end, pushing face of the pawl wants to be radiused in plan so that it is always tangential to the face C as the ratchet rotates.

[peter1972]

On peter1972 Said:

I’ve realised that my 3D geometry is not quite correct, so the 14.3° is not quite correct.

Thanks Peter.

I expected this morning to post a summary of what I’d learned from the topic before bedtime yesterday.  Now I’m going to 3D model your suggestion to see how it pans out.   Think it’s easy to model in Solid Edge, but maybe not because projecting cuts into curved surfaces isn’t always done the obvious way.

I also want to explore John Haines idea, and see if I can understand the double curve geometry better.  (It results from projecting a straight line onto a circle, and then cutting at a fixed angle towards the axis,  requiring the tool and/or job to tilt across a range of angles in two directions simultaneously, and to do so without the tool colliding with the job as it does so.)

The suggestion also tempts me into exploring an SE tool I haven’t tried yet.  It directs a form tool shape along an arbitrary path.  I expect to find an enhanced sweep, but learning it might take time.

Found some real examples on the web by searching for flattened ratchet, mostly found on bicycles and IC starters I think.  I can see how they were made by following advice given in this thread,

Closest match to what I’ve modelled is this revolver cylinder, which I think Peter’s method can do:

Watch this space, I’ll be back.   Fingers crossed I’m not interrupted every 10 minutes as yesterday.  No peace for the wicked…

Dave

[GordonHParticipant @gordonh]

I might be wrong,  but to me, the revolver example looks as if the barrel was mounted on the Z Axis and a dovetail cutter mounted with its shaft parallel to the X Axis and the flat face  of the cutter in the YZ Plane.   As the cutter crosses  the flat boss on the barrel, it rises in theYZ Plane, making a narrowing, shallowing tooth form.

Gordon

[Charles LamontParticipant @charleslamont71117]

Where is the position of max rightward travel of the pawl? Does it go to the point where face C is square to the pawl’s axis of travel, or some other angle, such as 360/26 degrees each side of square?

Either way, it is unnecessary and undesirable for the notch to be sharp bottomed. Instead there should be a flat area at the bottom, in a plane perpendicular to the ratchet axis. This should be sufficiently broad to accomodate the tip of the pawl throughout its rightward driving stroke, so that the pawl does not ride up and down the sloping surface during the driving stroke, but remains engaged with face C to the full depth of the face.

[Charles Lamont]

On Charles Lamont Said:

Where is the position of max rightward travel of the pawl? Does it go to the point where face C is square to the pawl’s of travel, or some other angle, such as 360/26 degrees each side of square?

If I’d modelled it properly, square.   But it doesn’t matter much: provided the wheel is pushed 360/26 one click at a time.   (In the Enigma machine the rotors are locked in the correct position by another mechanism.)

Either way, it is unnecessary and undesirable for the notch to be sharp bottomed. Instead there should be a flat area at the bottom, in a plane perpendicular to the ratchet axis. This should be sufficiently broad to accomodate the tip of the pawl throughout its rightward driving stroke, so that the pawl does not ride up and down the sloping surface during the driving stroke, but remains engaged with face C to the full depth of the face.

Interesting point, and added to my TO DO list. But the examples I found don’t look as if that’s how they do it.  In my design, the pawl doesn’t necessarily ride down the slope until it engages with the push-face, rather it pulls back up the slope on the return, rides over the lip and drops immediately behind the next push face.  The distance travelled by L-R stroke is adjusted to achieve that by design.

Though I asked specifically about how to cut my ratchet, many other questions have to be answered when the rest of the mechanism is considered.   So all comments are interesting and potentially valuable.

Thanks,

Dave

 

[Michael GilliganParticipant @michaelgilligan61133]

This is very-much “out of the box” Dave … so just put it on that ‘all comments’ list

MichaelG.

https://www.tec-science.com/mechanical-power-transmission/cycloidal-gear/lantern-pinion/

Lantern pinion as a special case of cycloidal gear

[SillyOldDufferModerator @sillyoldduffer]

Only interrupted 4 times, though the work is taking longer than expected.

First off, made a blank with a big hole in the middle, as suggested, and added a tangent plane.  (In addition to the usual XY, XZ, and YZ planes, CAD allows tangent and angled planes as well.)

The 2D sketch representing the cutter is drawn on the tangent plane and can be projected into the blank in various ways.  I experimented with straight cutting at 90° and then with projecting the sketch onto the curved blank and letting the software cut at a diverging angle to an edge on the Z-axis.

Straight cut with a 45° form tool (as typical in a milling machine) produces:

And repeating it with a rotary table would give:

Though easy to make, the result is no good.  Tilting the job (or mill head) to 26°and going deeper, looks better:

but it’s another fail.  On the outer edge the pawl has to climb over 12mm, whilst only 5 on the inside.  Not ideal – the pawl should climb roughly the same height wherever happens to be.

Next I tried altering the cutter form, in this example to a round rather that straight 45° V.   This could be done with a shaped grinding wheel.   There’s no advantage in the yellow curves and the flat triangular tops spoil the pawl’s action.

 

Next, the V is projected on to the cylinder, changing shape, and doesn’t cut from the tangent at 90°.

The top is flat, hurrah, and cutting deeper narrows the top to the desired nearly parallel edged form, shown in red below.  The pawl rises and drops consistently:

But the yellow face is hard to cut because it’s profile curves right-left and in-out simultaneously.  That seems to require a cutter that changes shape and angle depending where is it.    Spark erosion or a laser.

Need to explore further, cos I don’t understand this properly, but it feels like the best answer is one of: 3D-printing a mould and casting it; 3D-printing it in plastic and hoping it’s strong enough; sending the file off to a 3D metal printer and having them make it; or sending it to a CNC shop who does sparks, lasers, and tiny cutters.  So far no sign of a way of doing this in metal with the equipment I have.

Oh, and as a few correspondents have pointed out, the pawl’s profile should match the ratchet.

Back to the drawing board is still an option, and I’m not convinced I’ve investigated Peter’s suggestion sufficiently.

Dave

 

[Charles LamontParticipant @charleslamont71117]

Dave, I think you have the cart before the horse. I think you have yet to fully understand the geometry. There is no point in working out how to make it before you know what you should be trying to make.

Your photographic examples on grey and red grounds above (in #785626) appear to be designed for a pawl that moves with rotary motion, not linear. Your ratchet design in the OP with the complex curved shape of the slope, is suitable for rotary motion of the pawl but not linear.

 

[JasonBModerator @jasonb]

Dave, if you used the 26 ratchet teeth that you say you want rather than about half that,then you won’t get the big flat pie shaped top pieces.

Projecting to the side of the cylinder means the path of the cut faces are radial so constantly changing so unless you have some magic cutter that can also change then you will have to think again. This is why most hobby engineers make bevel gears with the PARALLEL depth method as they can’t replicate the action of a proper bevel gear cutter which alters the cut becoming smaller towards the middle

[JasonB]

On JasonB Said:

[…] This is why most hobby engineers make bevel gears with the PARALLEL depth method as they can’t replicate the action of a proper bevel gear cutter which alters the cut becoming smaller towards the middle

Excellent point, Jason

… having never actually seen a proper bevel cutter in action <blush> … do you happen to have a video link that I could enjoy with Coffee ?

MichaelG.

[JasonBModerator @jasonb]

Here you go. The important point is that the two cutters are moving on guides that are narrower at the end they are being viewed from so the cut gets wider towards the outer edge of the gear.

Plenty more if you search for Gleason bevel gear cutter or machine. Hope that is not your coffee getting poured over the gears!

[JasonBModerator @jasonb]

The new Gleasons are a bit different

[Charles LamontParticipant @charleslamont71117]

I reckon this might be more like the animal you are looking for. This one would, I think, work with the starting position of the push faces square, and moving right from there.

[JasonB]

On JasonB Said:
Here you go. […]

Thanks, Jason … Coffee is brewed !

MichaelG.

[Charles Lamont]

On Charles Lamont Said:

Dave, I think you have the cart before the horse.

Not quite – I know what the requirement is!

 

I think you have yet to fully understand the geometry.

True, it’s why I’m asking.  BUT!  the geometry in my Opening Post works, the problem is I don’t know how to machine the necessary curves.   How to make it is the exam question, also helpful, though is any other profile that does the same job (approximately), that can be machined.

 

There is no point in working out how to make it before you know what you should be trying to make.

Your photographic examples on grey and red grounds above (in #785626) appear to be designed for a pawl that moves with rotary motion, not linear.

They are examples, hopefully similar, providing clues as to how they were made.  They don’t quite meet my requirement either!  Don’t worry if they aren’t helping.

 

Your ratchet design in the OP with the complex curved shape of the slope, is suitable for rotary motion of the pawl but not linear.

 

Sorry, I don’t understand that, my fault I’m sure.  However, modelling shows my linear pawl rides the curves OK and I think the real world revolver example works the same way.  Presumably when the trigger is pulled, a pawl rises up inside the gun and pushes the cylinder round to position the bullet under the hammer and fire the next shot.  As long as the pawl pushes the driving face far enough, the return slope doesn’t have to be straight or curved.  But the ratchet has to be wide enough to accommodate the fully up pawl tending to the inner edge, and when it’s dropped fully down where it comes closer to the outer edge.   The pawl moves linearly, it’s the cylinder moving that changes the relative position of the ratchet.

My pawl and that in the revolver also have to move predictably, which I think is a problem with your example, if the pawl side-slipped due wear etc.  I like it, but it’s very similar to some of mine!   I accept it might be a good compromise though!

To repeat, the geometry of the wheel in the opening post would work if I knew how to make one.

Dave

[JasonB]

On JasonB Said:

Dave, if you used the 26 ratchet teeth that you say you want rather than about half that,then you won’t get the big flat pie shaped top pieces.

Ah but remember I like to solve the general problem, not just one-offs.   Although my example calls for 26 teeth, if there is an answer, it should cover any number of teeth.   There’s an Enigma machine that only does digits, which would need a 10-toothed ratchet.  And a revolver needs 6.  The Germans dodged the problem on Enigma with a mechanism that doesn’t need a ratchet of the form I designed.  Might be because they looked at the geometry I’m asking about and decided it was too difficult to make.

In which case, I might well be going back to the drawing board.

Projecting to the side of the cylinder means the path of the cut faces are radial so constantly changing so unless you have some magic cutter that can also change then you will have to think again.

Yes, very true, and why I started the topic.  Although my design works in CAD, the problem is how to cut one.   Certainly does appear to need a magic cutter!   But it’s straightforward enough to make additively.  Thinks, surely I can afford a 3D-printer that spits metal…

 

This is why most hobby engineers make bevel gears with the PARALLEL depth method as they can’t replicate the action of a proper bevel gear cutter which alters the cut becoming smaller towards the middle

Going to watch your Gleason video next.   Bevel gear cutting is similar, so maybe all will become clear.

Ta,

Dave

[Michael Gilligan]

On Michael Gilligan Said:

On JasonB Said:
Here you go. […]

Thanks, Jason … Coffee is brewed !

MichaelG.

Those Gleasons are awesome, Jason … but of course the ‘total cost of ownership’ justification only works if you have a good business and a good Accountant ,

Fortunately, my second brew of Coffee was spent watching this:

https://youtu.be/RBco7PVuPDI?feature=shared

which is significantly more relevant to the rest of us.

MichaelG.

.

[Charles LamontParticipant @charleslamont71117]

 

 

I think you have yet to fully understand the geometry.

True, it’s why I’m asking.  BUT!  the geometry in my Opening Post works

 

Your ratchet design in the OP with the complex curved shape of the slope, is suitable for rotary motion of the pawl but not linear.

 

Sorry, I don’t understand that, my fault I’m sure.

 

OK. In spite of your protestations, I maintain that the geometry in the OP does not work correctly. If in that figure we take the starting position as shown and imagine rotating the ratchet one notch anticlockwise then the pawl rides up the slope and drops nicely into the next notch – all fine and dandy. But that is not what is happening in your arrangement. Now, if instead of rotating the ratchet, we drag it in straight line to the right until the pawl drops(ish) into the next notch then I think you can see that the result is less than ideal.

[Charles Lamont]

On Charles Lamont Said:

 

 

I think you have yet to fully understand the geometry.

True, it’s why I’m asking.  BUT!  the geometry in my Opening Post works

 

Your ratchet design in the OP with the complex curved shape of the slope, is suitable for rotary motion of the pawl but not linear.

 

Sorry, I don’t understand that, my fault I’m sure.

 

OK. In spite of your protestations, I maintain that the geometry in the OP does not work correctly. If in that figure we take the starting position as shown and imagine rotating the ratchet one notch anticlockwise then the pawl rides up the slope and drops nicely into the next notch – all fine and dandy. But that is not what is happening in your arrangement. Now, if instead of rotating the ratchet, we drag it in straight line to the right until the pawl drops(ish) into the next notch then I think you can see that the result is less than ideal.

Ah, wrong way round Charles.   The pawl is moved by an actuator, actually a paddle moved by pressing typewriter -like keys.  The pawl pushes the wheel round anticlockwise one notch, and is then pulled back to drop into the next position.   The wheel doesn’t move the pawl. It’s a linear to rotary converter, not a ratchet.   Ratchet-like as found in an odometer.

The action is quite simple.   The problem is that the CAD method I used to get the vertical and top geometry right does so by generating a hard to cut double curved slope.  The double curve is a side-effect, not an essential part of the design!  So far, methods that avoid the double-curve don’t do the vertical and top geometry well.

Does your CAD package support animation?   I knocked up this rough assembly earlier, it shows the pawl driving  the wheel and could be copied.   ‘A’ slides in a slot, ‘B’ follows on a hinge so it can rise and fall, and the hook end presses against the vertical face of each notch.  On the return stroke, the hook point rides up the curve, over the top, and then drops into next notch.

Might pay to develop my animation fully – get all the mates fully constrained, lengthen the pawl so it engages in the next notch, and have the pawl driven with the correct distance with a linear ‘motor’, so the action can be watched from different view points over many cycles.

Guess what, I’ve been interrupted twice, prevented from watching Jason’s bevel gear video, and haven’t eaten yet.

Dave

 

[JasonBModerator @jasonb]

Any reason why the pawl can’t be rotated 90deg and act on teeth on the side of the wheel?

[Charles LamontParticipant @charleslamont71117]

I understand how it works. I was describing the relative motion, but it does not matter because your latest image demonstrates my point rather better; that with your geometry there will virtually always be only one or two point contact between pawl and ratchet throughout the operating cycle. My ratchet will give you line contact virtually all the time.

I don’t understand your point about ‘pawl side-slip’. Mine has a vertical, radial face C, just as all yours do.

Animation is easy in Onshape, but I am doing it OK in my head.

[Author]

Posts