Generalized solution to machining 2D complex curves by hand?

Thanks! In the meantime I also found a very similar idea by member Marcel here some years ago that I’d pretty much forgotten about.

 
You can do an awful lot with a cutting torch and a 2x72" belt grinder.
Back in the day they used cutting torches on magnetic tracer machines. You just made a sheet metal template and a knurled magnetic stylus rolls around the outside (or inside) at the right speed for the cutting tip you are using on the thickness of steel you are cutting. Those parts don't even really need to be finish machined most of the time if you're making something like a hand crank lever. Nowadays laser and waterjet have replaced the mag tracer pattern cutters. My advice is to find a guy with a waterjet machine and make friends with him. A waterjet is basically a very high pressure wet sandblasting nozzle that moves in a 2D path CNC.
You can in a pinch get a bunch of identical parts lasered out, stack them up and weld them together to make a very thick part.
About how to imitate a CNC machine job by hand I have nothing useful to add.

metalmagpie
 
I would lay the part out then cut it with a band saw the last bit I then would mill it using a .010 step over. Mill to the line move away from part step over .01 then mill to the line like a tool path in a CNC machine.
 
I would lay the part out then cut it with a band saw the last bit I then would mill it using a .010 step over. Mill to the line move away from part step over .01 then mill to the line like a tool path in a CNC machine.
The "rough with bandsaw" has come up a couple of times. A previous had laying out the part (paper template I believe, but doesn't matter), bandsaw to the outline and then finish. That was on 1/2" thick aluminum which cuts fast enough to be tolerable.

Roughing out on a bandsaw is something I do, but this is REALLY only practical for thin parts and/or material that cuts quickly on a bandsaw. I did not mention in the original post - but did in a follow on - that one of the cases was 1" thick 1018. In that case I drilled 50 or 60 holes with a 1/8 bit about .03 to .05 apart and then was able to bandsaw my way through it in a reasonable amount of time.

So the roughing part is actually a whole other topic.

Doing the profile .01 steps at a time is going to mean 100 steps per inch of profile - also quite tedious. I assume you are talking of using a list or something for the target points?

The approach is not inherently unusable, but I would use the biggest mill that would work and then figure out how far the step can be before the cusp of overlap exceed some chosen value. It seems like it would still amount to hitting a very large number of points from a list by hand. Workable but tedious. Could also use CAD to approximate the curve as a series of straight lines, but then have to get the simultaneous feed in each axis right!

For theoretical concave curve, I recently tried to layout then number of circular cuts (to be done with a boring head) and it wasn't too bad. 5 or 6 cuts for the random convex curve I drew. Still would be a bit slow as the boring head has to be fed down in Z but not a lot of different cuts. This seems like the most promising approach for a concave / internal profile.
 
When I have to do a non linear shape on my mill I break out the rotary base for my mill vice. I really helps when milling to scribed lines. The rest of the time that thing is doing boat anchor duty. I have a rotary table but by the time I figure out how to get the radius and where to put the radius I could make the damn thing by hand with 600 paper.
 
If time is not a problem, you could mill the shape using short horizontal and vertical movements. The shorter the moves, the more accurate the profile will be. After milling you could use a grinder/file/stone/etc to make a smooth surface.
To get the coordinates, you could use a fine grid overlay, position the mouse pointer on the curve and read the coordinates from the screen.
Time consuming but doable.
 
My Grizzly DRO has the ability to cut arcs on a plane defined by any two axes\ by cutting steps. You enter in the the center of the arc, the start and end points, tool offset, and the number of steps. I used it a number of times to cut both convex and concave arcs. It is painfully slow and tedious but it does reduce the opportunity for operator error as when you load the next point, it tells you how far away from the point you are. The two remaining sources of error are moving the wrong axis first and overshooting the target point. When I used the function, I would generally set .020 to .050 for the maximum step and clean up the saw teeth with by filing and sanding.
 
a couple of parts with curves. that I had to make.
The first one is a block off plate for modifying the suspension on a motorcycle. It is all tangent radii and was done on the rotary table. It is out of .250 thick 304 stainless.

The second part is a valve adju8sting tool for a motorcycle. it was made from a 2 inch square x 12 inch long bar of 4140 that is not a radius on the outside, it is a cam profile, and the small step is also a cam profile. I had to make a bunch of them. It was done by milling the cam profile into a bar and then slicing off chunks to make the final parts. The cam profile was done by having the rotary table set up vertically and the bar between centers, I calculated the y dimension in one degree increments and then side milled the bar rotating 1 degree at a time, The setup is actually my profile pic. The finished result steps were actually a better finish that the tool marks that would be left by turning on the lathe. Even though this is a cam surface it required no cleanup or other second op.
 

Attachments

  • Anti Dive Blockoff MKII 002.jpg
    Anti Dive Blockoff MKII 002.jpg
    229.7 KB · Views: 7
  • Valve tool 002.jpg
    Valve tool 002.jpg
    36.1 KB · Views: 7
  • Valve Shim Tool 007.jpg
    Valve Shim Tool 007.jpg
    61.6 KB · Views: 7
You can do an awful lot with a cutting torch and a 2x72" belt grinder.
Back in the day they used cutting torches on magnetic tracer machines. You just made a sheet metal template and a knurled magnetic stylus rolls around the outside (or inside) at the right speed for the cutting tip you are using on the thickness of steel you are cutting. Those parts don't even really need to be finish machined most of the time if you're making something like a hand crank lever. Nowadays laser and waterjet have replaced the mag tracer pattern cutters. My advice is to find a guy with a waterjet machine and make friends with him. A waterjet is basically a very high pressure wet sandblasting nozzle that moves in a 2D path CNC.
You can in a pinch get a bunch of identical parts lasered out, stack them up and weld them together to make a very thick part.
About how to imitate a CNC machine job by hand I have nothing useful to add.

metalmagpie
I'm thinking it was 45 years ago when the fabrication shop my father managed purchased a pattern cutter that used a torch to cut heavy plate to a pattern. The patterns were drawn at 1:1 on drafting paper. They had to be sharp and black, so we had to break out the ruling pens and draw 1/8" lines in ink.

Of course, the part was only as accurate as 1.) the drafter and 2.) torch cutting, neither of which can work to typical machining precision. (I could draft to .010 but the machine followed 0.125 lines.)

That was early stuff, of course.

I have wanted in the past to make accurate cam plates for automatic focusing devices on cameras and enlargers, and these are often complicated mathematical shapes depending on the other geometry of the machine. The best I could do was using a band saw and a drum sander, both of which limited me to aluminum (which was fine in those applications). But it was tedious hand-fitting, never attaining the sort of accuracy really needed, and only providing a starting point for final manual focusing.

The timekeeping industry and the cam slots in things like zoom lenses used to be made to tenths or even millionths precision on production machines long before CNC. But the effort went into making the patterns the production machines would follow. Those jobs were all about metrology and skill. CNC has made that much easier. People think the wristwatch mechanical movement industry is mired in the past, but though the movements are an archaic technology, the machines that make them continually push the boundaries of the state of the art.

Rick "reading the responses with curiosity" Denney
 
Back
Top