Router Forums banner

Crazy Idea?

3429 Views 16 Replies 7 Participants Last post by  4DThinker
My compound miter saw will cast 2 red laser lines onto the board being cut to show me where on that board the cut will line up. My drill press projects a red X on the project board showing me where the drill bit will be centered when it enters the wood.

I woke up this morning with the wild idea of being able to accurately project the loaded g-code toolpaths onto the CNC bed where the board it will be cut into is clamped.

Both LinuxCNC and my CNC Shark Controller software have a g-code preview. Once X, Y, and Z origins are touched off on the CNC bed the tool position maps directly to the g-code image on the screen. This does let me indirectly verify the toolpath locations relative to the board and my clamps, etc., but isn't as directly obvious as having the toolpaths projected ONTO the actual board would be.

Twice in the last few weeks I've started up a CNC job only to discover the cut wasn't going exactly where it should. In both cases the parts had to be cut again on fresh boards. These were student-created toolpaths and in both cases the origin point wasn't set where we told linuxCNC it should be on the part. If the toolpaths had been projected onto the parts before we cut them we would have easily spotted this misalignment.

Yes, this may be an idea with no hope of actualization. A alternative might be to mount a camera over the CNC bed that sends it image to LinuxCNC where the toolpath/image superimposition can take place. I just know that if a projection of where/what will be cut is useful on a drill press or miter saw, then a toolpath projection would also be useful onto a CNC bed.

Laser image? 4K projection? Mounted on the CNC itself, or on the ceiling above perhaps? I don't (yet) know how to realize this idea, but I want one!

4D :)
  • Like
Reactions: 3
1 - 17 of 17 Posts
Interesting idea 4D... I like it. I thought I had seen something similar during one of my late night Youtube binge sessions but could not not find what I thought I saw. I did find this however...
While not directly answering the need I suspect what your asking for is quite doable. I hope other more seasoned folks will come up with a solution... Because "I want one!" too.
  • Like
Reactions: 2
I don't see why it wouldn't work as long as the lines are 90* to each other. It should be no different than the drill press. One line would define the x axis and the other the y axis.
For use on a CNC the idea I have is more than just for aligning to find X and Y origin location. In some (OK many) cases the board a student brings in to have cut ends up not the size they claimed it was when they sent me their drawing to have toolpaths created for. Having the toolpath projected onto the board will tell us instantly if they fit on the board, if they are aligned with how the board is clamped down, if they cross over any clamps, etc..

That sure would save having to run a job 'in the air' an inch above the surface to make sure it's going clear clamps and hold down screws. Sometimes the path I want most to check comes at the end of a 15 minute (or longer) job and I don't want to wait that long while cutting air just to make sure I don't hit something. I could just generate a toolpath for that single operation and check it out but seem to forget about that until I'm ready to run the job. At that point I don't want to go back to the computer and generate yet another toolpath. Of course, I could move on to repairing my vacuum pump to use for hold down and not have to be concerned about clamps and screws... but I have other work to do so that's probably not happening any time soon.

I want it. Great idea
Whenever I have a question about something lining up as expected, I'll run a test of cutting air to see where the bit is going to go. It doesn't help with depth, but then neither would a laser.
Depth of a cut is rarely what goes wrong, and when it is the problem it is usually a loose bit that no toolpath preview or air cut would have pointed out.

It did pop another idea for a CNC related product/solution.... a loose bit detector. Surely there must be a way to sense strain in a router collet that would vary with how tight an installed bit is in it. In our college shop I often am changing a bit on a CNC or more ordinary router table only to discover that whoever used that machine last didn't have the bit very tight in the collet.

Many months ago I pre-ordered a glowforge laser cutter. The singular unique feature it had that caught my attention was "Precision Preview", where a camera in the cutter lid shows the actual material on the bed. You drag your toolpaths (drawing) around with finger or mouse to align them perfectly on the material where you want the cut to be. Toolpath image overlays material.

I cancelled my glowforge order after they had delayed it 4 times. I'm not sure if they are shipping them yet, or ever will. That Precision Preview feature is exactly what I want for my CNCs though.

If the toolpaths had been projected onto the parts before we cut them we would have easily spotted this misalignment.
This does exist as a commercial product:

Pretty pricey though.

My solution for this is to create a simple “outline” toolpath that shows the extent of the cut area (i.e. offset the cutout profile to reflect the diameter of the tool I will use for cutting the profile). I create a simple tool path cut along this outline. I name this toolpath “clearance”.

On my cnc I place my stock. If starting with rectangular stock, I usually have it butted up against dogs in 2 directions (X=0, Y=0). I usually have my clamps loosely in place.

On my machine I have the ability to leave my VFD/Spindle powered off while the rest of the machine is powered on. Leaving the spindle OFF, I chuck a laser in the collet. Now there are $125 lasers designed for this (Laser Center/Edge Finder), but I always prefer my cheaper ways of doing things. Those of you who have rifles may have used a boresight laser to do an initial zeroing of a scope on your rifle. It so happens that the diameter of a .300 win mag round (.489) chucks perfectly into a .500 collet. You can find this type of laser boresight for about $20 on Amazon, I found mine for less than $10 on I chuck my “bullet” laser into my spindle and run my “clearance” toolpath, usually a few inches above my stock (the laser “bullet” is much shorter than the commercial laser center finder, so it works better if you have thick stock or limited Z height). As the laser travels around, I can make sure it doesn’t hit any clamps. I tighten my clamps and am ready to cut.

I have never (yet) cut into a clamp. And I sometimes tend to use nearly all my material.

The site for the commercial laser center finder has some additional uses that also work with my $10 version, namely verifying that my adjustable angle table, and adjustable fence on it are at a perfect 90 to the spindle axis. It is also useful to line up the center point if that is what you use for the reference on your carving.


See less See more
  • Like
Reactions: 1
It did pop another idea for a CNC related product/solution.... a loose bit detector. Surely there must be a way to sense strain in a router collet that would vary with how tight an installed bit is in it.
At the October meeting of my local CNC group (Colorado CNC User Group), a potential solution to the loose bit problem was demonstrated by Ron Reed of Tinker and Tinker (Precise Bits). He has been working on a prototype system that allows one tool tightening of the router collet to a precise, repeatable torque. Not quite ready for sale yet, still being tested, but it is close. The prototypes work with ER20 sized collets (requires a custom collet nut and a special tool), I don’t know if they will make other sizes available.

Nice that Tinker and Tinker is only 2 miles from the shop that is our normal meeting place. They have come to several meetings, and Ron Reed is very knowledgeable.
For the Multicam (with a tool changer) we have in our fabrication lab we've set up a bit changing station and use a torque wrench to tighten the chuck until it clicks. Set to manufacturer's specification. For the small CNCs which currently use dewalt routers the biggest challenge is routine and distractions. If I personally change bits they get tightened "enough". Enough, of course is simply from 4 decades of changing bits and muscle memory. When a student changes a bit it is a crap shoot. Most of the time they are lucky and the bit stays in place. My problem is the routine I go through each time a new project gets set up for CNC cutting. If I'm interrupted mid process (which happens often) I may end up completely skipping a step or two, and not be sure since I do have many memories of running through the same routine over the last few years. This never happens in my home shop because I'm never interrupted.

The only problem I have with laser projected dots or an X is precision. They are fine when a part is being cut out of a larger board and there is plenty of room around the part. Much of what I cut is for joinery where a few thousandths of misalignment can be seen and felt when a joint is assembled. I tried out a laser X that was intended for drill presses, except I mounted it on my CNC's router body. The center of where the red lines crossed was easily 1/32" wide though when best focused. I can get within .0025 just using thin paper between bit edge and wood.

Thinking about how Glowforge solves this problem I realize that they have a lid which always lines up in the same place over the project bed. A camera in that lid has known coordinates relative to the bed, so any pixel in the image it captures can be mapped to a known x/y location. No lid over the CNCs I use, and they are all on casters so we can move them easily to clean underneath. Mounting a camera on the ceiling above will not work for reliable/repeatable image to coordinate alignment. I've seen microscopes mounted next to the router that let you see when a mark or corner on your project board aligns with an X in the image. Offsets are then used to tell the control software where the bit is relative to what the microscope sees. I suspect pretty good precision can be found using this method. I'd invest in trying it out if I had any clue how to incorporate the microscope image into linuxCNC.

See less See more
You don’t need to integrate the image. You can run a separate application capable of displaying the image at the same time. Once you get the target point aligned, you just switch back to Linux cnc and apply the offset. You need to find an application that will display cross hairs.

I’ve played with this (using windows) for aligning a part back on the cnc, aligning to a hole. The accuracy with the usb camera is better than Laser dot. The laser is good for checking the clamps are out of the way.
A little google searching came up with links to where someone was able to add the camera image into a tab in LinuxCNC, and someone came up with a way to overlay a cross hair onto the image. There appears to be an app that can also overlay more involved images (not sure what format) that you can compare to what is being seen through the camera.


For small parts the limited focal range and the above strategies may make my idea feasible. For projects that cover a larger area of my CNC bed the low gantry height limits what any camera mounted on it could see at one glance. So I'm still looking. The router mount back plate on Probotix CNCs does have spare 1/4-20 threaded holes on it where a camera bracket could be mounted. Any fixed camera that could overlook the whole bed couldn't mount on the moving gantry. Such would be a very difficult mounting challenge.

Going back to the problem of using a projector mounted to ceiling to project toolpaths, and your issue of CNCs mounted on wheels that get moved, I found this YouTube video of someone who solved this issue. It would require 4 fiber optic light sensors mounted in the corners of the cnc, then a computer could align a video image to the bed within a few seconds. Maybe some of your students would like to take this guys thesis project and implement it.

Johnny Chung Lee - Projects - Thesis

If l were their instructor I’d give them extra credit!
  • Like
Reactions: 1
As part of our program (Interior Architecture and Product Design) the students have to do some sort of research project in their 5th year. It is rare that we get one who can think of a furniture related project, but I have several students express an interest in manufacturing technology and in creative uses of CNC tech. No need for "extra" credit when normal graded credit can be the reward for their work.

The difficult step here is finding a student who can come up with these crazy ideas/research topics on their own. Frowned on are project topics suggested by the faculty. This is one I may be able to get a youthful tenure track professor interested in pursuing though.

1 - 17 of 17 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.