The Redesigned CNC Scroll Saw Rides Again

When [Andrew Consroe] tried to build a CNC scroll saw, he quickly learned how tricky of a design problem it is. With a blade that only cuts in one direction, you can’t simply move the tool in the X and Y dimensions like you can with a laser or router; either the work piece or the blade itself needs to continuously rotate towards the direction of the cut.

He’s recently shown off the third version of the machine, and while it’s still not exactly a practical tool, there’s no question it’s a brilliantly designed one, or that it works, slowly. Earlier attempts used a rotating table to spin the work piece, but [Andrew] found this to be an imperfect solution. Building a mechanism heavy duty enough to spin the material being cut while remaining accurate enough not to break the blade was a tall order, though he did get pretty close.

The earlier version used a rotating table.

This time around he’s decided to simply rotate the blade itself. This can be accomplished with a single stepper motor and some suitably sized pulleys, while maintaining an exceptionally high degree of accuracy. The whole blade assembly moves up and down on an aluminum extrusion rail with a motor and crank arrangement. By synchronizing the rotation of the blade with the vertical movement of the saw, the software can be sure that everything is where it needs to be before the cutting stroke actually happens.

Judging by the video after the break, the system works quite well. The complex rounded shapes he cuts out of the piece of plywood look essentially perfect, and it sounds like this new version of the machine isn’t breaking blades due to positional errors like the previous one did. Unfortunately, it’s also very slow. There’s so many moving parts and careful positioning required that even when the video is sped up 10x, the saw still appears to only be creeping its way through the  material.

On the back half of the video, [Andrew] details another approach to rotating the blade that would reduce the amount of moving mass in the saw. This would give the machine a considerable speed boost, and we’d love to see him implement it. By the way, before anyone says it: using a spiral blade is cheating.

18 thoughts on “The Redesigned CNC Scroll Saw Rides Again

    1. Actually, there are ready-made square linear guide rails allowing both smooth linear movement and able to transmit torque:
      They aren’t very cheap, but will require less parts than the design you shown an will be much lighter. As for the up-down movement, i would try a pneumatic cylinder with springs mounted near the ends of the stroke to dampen the end stops, and sensors (eg. inductive) positioned to detect when the cylinder just hits the spring. Then it’s a matter of setting up an Arduino or whatever you’re using to monitor those sensors and drive the cylinder’s valve. When it detects the cylinder near the top on the up stroke, it switches the valve to down movement. When the cylinder enters bottom sensor detection range, it switches the valve again to top movement and so on. Those babies can be really fast and you can easily tune the frequency by changing pressure or adding a choking valve. Plus you’ll still know when it’s safe to rotate the saw since you control the up-down strokes.

  1. It’s cool to see the progression in this project.

    If a workable spiral blade exists with that same teensy amount of kerf, rather than calling it cheating, I’d call it smart.

    This update got me to musing about a really fine bandsaw blade. It seems like such a blade might be able to be twisted to cut from a different angle while running. Not sure that it could be done without flattening out the tooth set, but it was an interesting thought train for a few minutes.

      1. That is cool. I imagine that a shallower(?) blade might be able to twist further. On the other hand, you’d never be able to shove as much material through it as fast as that 1/2″ deep blade.

    1. It’s cheating when the whole goal is to try and make a CNC scroll saw.

      If it was a matter of just making a practical CNC cutter, he’d use a router/mill/laser/whatever and been done already. This project is specifically about solving the engineering challenges inherent with this type of blade.

      1. Hmm…would changing the blade geometry mean that it’s no longer technically a scroll saw?

        I figured the goal was to achieve a specific result in the cut. Tight turns and probably very low kerf loss. I imagined an automated process for creating custom jigsaw puzzles (scroll saw puzzles?).

        I got to wondering how thin the kerf actually could be with a scroll saw. According to the internet, .010″ thickness is commonly available in a thin kerf scroll saw blade. Other than a laser, I’d expect that would be tough to match with other types of cutting equipment. In wood, the laser is going to burn the edge that may be a plus or a minus.

        1. my take is that to qualify as a scroll saw for purposes of the design challenge, it has to be a linear reciprocating blade, so something like a Diamond Dust Wire Saw might qualify

          1. A minor quibble: linear reciprocating blade *under tension through being fixed at both ends*. Otherwise it’s more like a jig saw. Its being fixed at both ends is what allows the narrow kerf. And, yes, I agree that a diamond dust wire saw should qualify, though I don’t know how long it would cut wood before getting clogged.

            A downside of the scroll saw as opposed to the laser (or a router) is that the scroll saw can only make one continuous cut without detaching the blade, and to start a cut in the middle of the material requires a hole to be drilled in the material.

  2. This looks like a perfect application for a hollow shaft rotary ball spline. You’d get tensioning from the hollow shaft, rotation, and accurate up and down motion all in one package. Add a return spring to the bottom holder and you’d be able to have top and bottom not physically connected. Add a pair of timing belts to a shaft connected to top and bottom and you could go to a gantry style frame for added stiffness.

  3. Well shoot. Props to this guy and as someone said above it’s really more about the design challenge than the practical need for a cnc scroll saw. But the whole time I’m watching I’m thinking “this kid is overthinking this. Surely you can use the normal scrolling mechanism instead of all that precise timing and just…” Aaaaand I’ve got a half backed sketchup model. Guess I better save for some servo’s and put my money where my mouth is!

    Love this guy’s work though!

    1. It’s kind of painful, like watching someone design a motorbike from scratch, and pick a centrifugal jet turbine, powered by coal dust, and decide the back wheel can be a skid and make it front wheel drive, with an omni-wheel, made out of laminated bamboo, friction welded together, with a modified electric toothbrush, run off solar panels salvaged from 50 calculators…

  4. Great work. If you are looking for ball splines with pricing, you might want to consider Misumi ( I am using one of these for an oscillating tangential cutter that I am working on. The one that I ordered cost around USD $154 for a 6 mm shaft that is 200 mm long. I have run mine up to 2,000 strokes per minute, though I am still testing the configuration. If you want to see how I used the ball spline, I put a Fusion 360 project file here:

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.