Manually milling non-circular curves

[10] Like what you see?
Click here to donate to this forum and upgrade your account!
Hi All
I have a rotary table and I’ve used it for milling circular curves, when I have a center hole to pivot around. So I get the concept and use.
However, I have designs with curves that are not circular. Can I manually mill these curves?
I imagine the answer is yes but I would have to plot out every single thousand’s of an inch movement of the X and Y dials.
Is that true and is that how the old timers did it? Right now CNC is not an option. $$$


Yes, it just takes a little practice. Usually when I make a mistake with the movement it is when I have paused to move in another direction. You just have to make it a point to think of the effects of your hand movements before you move them.
I call it etch-a-sketching.
I just made this for my wife for Christmas. All etch-a-sketch.
Thanks for the reply. That bone looks pretty good and the ends are similar to what I would be making.
So for a little more detail, I imagine the procedure would be something like, down 1 on Y, left 1 on X, for example.
Wouldn’t that create a bunch of little right angle cuts, like teeth or serations?
And plotting that from my CAD program seems like a nightmare.
If you use a slow powerfeed on one axis you just need to constantly adjust the other axis by hand. Tom Lipton made a video as part of his wilton baby bullet build series where he explains how to use a template to create nice curves this way. Finding the link from my phone is too much hassle, sorry, but I think it was somewhere around part 15 of the series.

Sent from my SM-G930F using Tapatalk
Yeah, I imagined that if I could attempt both dials at the same time then the cut would be angled and eliminate the scalloped cusps. However, I’d have to operate with the accuracy of a machine to do that.
Doing it by X, Y, and cranking both at the same time or alternately can be a disaster unless you do a good amount of practice first, to get your hand and arm muscles and your brain on the same page without making any mistakes. I suppose that a good amount of practice on scrap would probably do that. I have had terrible luck trying to look at the line while my hands made the cut. It is just way too easy to make a mistake. It is not intuitive, it must be learned eye/muscle coordination. Doing it bit by bit carefully but off the cuff will probably lead to a glacial time frame for doing the cut, while also still being risky of making mistakes. Going slow is very hard on the cutter as well...

Consider other techniques to get you to the result you want as well, there are many ways to take a part to finished size, like roughing it with plunge cuts, followed by sanding and grinding it to final shape, just one option.
The ellipse has always fascinated me , in 85' we bought a CNC mill for the shop and in the manual was a page on programming a ellipse . I had no need to make one so I didn't pay it any attention , one day a tech was in the shop fixing the machine and I asked him about the ellipse program he said "don't waste your time on it , we can't get to work" . In 85' CADCAM was in it's infancy especially for a small time machine shop . I bought one anyway (total waste of money) . Anyhow Maplehead got me thinking about it again and now that Google is our friend I Googled it and found some interesting jigs https://www.google.com/search?q=how+to+cut+a+ellipse&rlz=1C2CAFA_enUS600US600&tbm=isch&tbo=u&source=univ&sa=X&ved=2ahUKEwj0n73Li77fAhWCHjQIHU9tAVUQsAR6BAgCEAE&biw=1366&bih=626
But it was all for wood , how did they do it in the pre CNC days in metal ? My guess is a pattern made by hand and then use a hydraulic tracer .
Or something along the lines of the Rose lathe .
Try drawing a angled line or a circle on a etch a sketch. It is exactly just as difficult. I have done some irregular shapes. The problem is when it’s a mirrored image or you need a duplicate. One way I found that helps is saw cut you part to your ruff shape. Leave a 1/4” around the part and use that to try out and learn the motions of what’s needed to get you around the part. Then hopefully you’ve gotten the hang enough that there’s not much to get to your scrub line. Just sand to line with what you have.
The basic fundamentals of machining require hand/eye co-ordination, that was the first thing we learned in shop class. Matter of fact, when we started working with mills, the first job we were given was to write our name with a Bridgeport. Clamped a block of wood in the vise and had at it with a 3/16" end mill about 1/8" deep. If you had a short name, or even got cocky, the instructor would make you do it in cursive. We did have one Polish student that never got his done........:grin:
I assume a metal bandsaw played a larger part in making complex shapes pre-CNC.
I recently bought an old Keller die filer at an auction with the idea of making square or hexagonal holes out of round ones (very slowly--it's not a fast tool). I assume when die filers were common they were used for finishing complex (especially interior) shapes. It would work well for smoothing the ends of the bone-shaped card holder after band sawing--faster than hand filing, anyway. Obviously, in industry they don't use any of these things for that kind of purpose any more and they are left for us to play with!
Last edited:
About 20 years ago, I had to machine an Archimedes spiral for a product I was designing. The first attempt was to use a straight line interpolation with the condition that the resultant curve couldn't deviate from the actual curve by more than .001". This required around 5,000 individual moves.

The second approach was to fit circular arc segments to the curve with the same condition of more greater deviation than .001" This was accomplished with 24 arcs. The construction was done using AutoCAD and a G code program was created. (Nowadays, I would just draw out the desired profile and let the CAM math engine calculate the path).

While this was intended to run on a CNC mill, it is also possible to create the tool path for manual milling. To do so, I would mount my RT on the mill table and mount a secondary x-y table on the RT. The purpose of the secondary x-y table is to be able to position the center of each arc segment at the center of the RT. Using this method, it is possible to precisely create very complex curves but it is tedious and requires a lot of calculating to determine the centers of the arcs, the start and endpoints of the arcs, and the translation to a polar coordinate system so the RT can be used.
Like how a universal dividing head couples the rotation of its axis to the travel of the table X, an ellipse can be drawn fairly simply by coupling the ratio of the displacements in both x and y to the rotation of the work. The old "machinist's do-nothing" motion is similar. There wouldn't be any interpolation or stair-stepping needed to draw, plot, or machine an ellipse. Only needs fixturing that follows the geometry for this one. Spirograph, an old toy that explores parametric relationships, has a real pre-digital application here.
My and many DROs have a function for milling arcs by waypoint. You program the start and end point, desired radius, number of waypoints tool diameter, etc. It then computes the path. You navigate to each waypoint till X=0 and Y=0 then hit a key to enter the next waypoint, repeat... Compound curves can be achieved by stringing arcs together. I used the feature only once, it worked surprisingly well though I had to do some hand finishing.
How about sticking a pencil in a collet, or drill chuck, and practice drawing your shapes on a clipboard clamped in the vise? You could do this all day, or twenty minutes a day for as long as it took to master the skill. Doing it this way, you wouldn’t even have to be running the machine. Just running the handwheels. I imagine a couple weeks of dedicated practice would do wonders. You’d also have immediate, non destructive feedback every step of the way.
There's probably a lot of easy ways to do it, only if we knew what you are working on....
I come across the need for curve cutting all the time. My latest project is pickup covers like this one.
If I had the means to make these on my own then I think I’d attempt cutting my own punch and die sets so that I could press these out of 24 ga nickel silver sheet.


All I see there are radii . Maybe I'm missing something because you are showing the part at 45* .
They are covers for a humbucking guitar pickup. Their shape would be properly described as "obround." Think of a rectangle with semi-circle ate each end.
Those would be hard to press form. Where the 2 radii meet would tear out.
Yup, that's why most of the humbucker pickups covers I saw when I was playing electric were square. I don't think they needed to form fit so tight. And because they damped the output a lotta folks tossed them.
"Obround" ? I have never heard that before . Teach a old dog a new trick , I don't understand .
You've got it right there from Wikipedia- a rectangle with a semicircle tangent at each end.
Or, the shape of a milled slot, if you prefer.

At one point, the term was pretty foreign to me. When I started with precision sheet metal fabrication back in 1980, it referred to punched slots. I was calling them ovals, but one of my colleagues corrected me. Obrounds have flat sides, ovals are more like ellipses....anyway, it still sounds strange.
Last edited:
The way I was taught to do this was...
Plunge milling, yes I said it! I know I'm beginning to sound like a fanatic, but hear me out:

If you have a DRO, you advance the cutter to each of the pre calculated points. when you are SURE (by checking your layout or cemented template, then you just plunge down. this is very accurate and doesn't involve any risky moves. Also, you get to lock the X and Y travel to make sure your cut is right.

just sayin'

P.S. I have done lots of non-eccentric cam plates like this from paper templates and a loupe. it works very well!
[5] [7]