Router/Laser Hybrid Project without unmounting

Accessories J Tech Laser
1

I’ve got a product stream using the 1F (Woodworker, original) that I think I could use the laser addon to significantly reduce my total processing time.

What I have so far:

  • Use Easel for the .nc files
  • I cut exclusively with a 1/8" bit
  • Stock is all hardwood, resawn and planed down to 0.30" thick
  • Every cut program uses the same X/Y/Z zero point. I don’t do any zero alterations unless my machine loses power. I use pegs to drop each piece of stock to the same spot on my spoilboard. At least, close enough, I factor in up to ~1/16" wastage around the edges in the piece placement, the pegs get clipped here and there. Exact placement of the stock is non-critical as long as the program area is covered
  • Due to the small size of the pieces I am cutting, no hold down method works for a single piece (0.75-1.5 square inch in area), and I cut 40-60 pieces per piece of stock, so I tape the whole board and only cut 0.29-0.295 into the stock, leaving 0.01-0.005 material remaining (leaves stock looking like this: https://photos.app.goo.gl/S7QQ1LybqpQ9PxhM6)
  • I then use a benchtop belt sander to sand off most of that remaining, breaking the pieces out of the stock.
  • I then use a disc sander to manually sand down the last of the remaining material, removing all of the extra stuff from the breakout.

These sanding steps are now up to about 80% of my time spent per saleable item.

What I think I can do with adding a laser:

  • After a router cutting run, before removing the stock from the spoilboard, run a laser (7w?) pass that trims the edge of every piece to cut that last 0.01-0.005" of wood cleanly. Removing the manual steps to break them out of the stock.

I expect:

  • The laser would need to be mounted so that when the router bit is at it’s fully 0.295" depth, the laser body is still above the stock to avoid crashing. Lets assume that is 1/8" (0.125)
  • When the laser program is running, the router bit will need to be at a safe height (0.15"?) above the stock, which by necessity of the above point, the laser body will be 0.30 (stock thickness) + 0.15 (bit safe height) = 0.45" above the stock.
  • Which then means that the laser would need to focus at 0.45 (above stock) + 0.30 (stock thickness) = 0.75" below the body
  • I will have to carefully measure the -Y offset from the router bit zero to the laser zero, and then offset the laser cut program +Y by that same amount.
  • I would end up with a faint burn line on the lower edge of every piece

What makes this a non-starter:

  1. Having to re-set ANY zero for either router or laser between jobs
  2. Having a mount that prevents router usage with the laser mounted, since remounting the bit, the router, or the laser will mean at least some minor variation in X/Y and/or Z.
  3. The laser unable to focus 0.75" away
  4. The laser can’t actually cut 0.01-0.005" of hardwood (everyone talks about maximum cut, and I think this is doable, but can’t find anything to confirm it)
  5. The laser can’t cut exactly down the side of the work pieces due to beam divergence or the upper edge of the piece clipped part of the laser. If I have to move the cut line out from the base of the piece, it doesn’t save me any time.

And because someone is inevitably going to ask about workholding options instead to allow full depth of cut:

  1. I use Taytools double sided tape as the only known solution for workholding for this, and I am only safe with around 40% coverage on the stock. Less and I have a high failure rate on the tape’s ability to hold.
  2. Vacuum is a nice thought, but given the final square inches, I’ve only got around 9-20 pounds of pressure holding each piece, and that just isn’t enough to prevent the pieces from getting ripped out and damaged at an unacceptable rate. I’ve tested, and I lose >50% pieces still. This is assuming I don’t have massive vacuum loss as I cut through the board all over the place.
    3B: It would technically be possible to have custom vacuum backing boards to ensure vacuum is only delivered to the pieces and no cut channels, although I would need 100+ of these boards, and the product updates and tweaks would mean remaking them regularly. Time and wasted stock material that I don’t see as functionally worth it given the difficulty of vacuum holding the pieces anyway.
  3. Clamps to the stock edges/corners would quickly no longer be holding anything.
  4. A high pressure clamp mounted to the body of the router that holds the middle of each piece down for the duration of the cut might work, swinging around as the router cuts around the edge, but it would have to be controllable to locate/assign the center of the coming piece prior to the start of cutting, and would have to be able to swing around faster than the X/Y movement. This is just a fantasy thought. I have never seen anything remotely close to this concept out there.

So that all being said … does anyone see any issues with the idea that makes it a non-starter before I buy a laser?

2

I have no experience with lasers, so can’t help with the laser-specific issues. But it did occur to me that the G52 temporary coordinate system offset command should be an accurate and relatively easy way to adjust for the physical offset between the spindle and the laser.

But back to your original challenge, have you considered using a drum sander to remove the onion skin? What I’m thinking is to leave the onion skin thick enough that the double-sided tape can be removed without breaking the parts free. Then run the full machined panel through a drum sander, with the bottom side up, to sand away the onion skin. The sanding drum will press the parts downward against the conveyor belt as they are being cut free, and the parts will also be captured laterally by the waste material.

I’ve never tried that, but it seems as if it could work. And if it does, you might be able to dispense with the double-sided tape entirely. I often cut parts with a downcut bit and leave an onion skin, in which case clamping only the perimeter of the stock has worked just fine for me.

I also like to use a spiral-ramp toolpath for the perimeter cut, instead of discrete step-downs. That results in a very gentle cut as the bit reaches onion-skin depth.

3

Thanks for the G52 idea, I could see that being very useful.

Yes, drum sander is in my list of potential ways to speed up the process. Although I could still see plenty of challenges. For example, would the drum sander’s belt be exerting more torque on the piece than it has left holding it, at any point? At the start of each piece, the answer would be no, at the end of the piece, the answer is definitely yes, because there is no remaining skin left. Where is the transition point? It’s the same problem with the router bit, or anything else. When the torque to the piece is greater than the strength of the remaining connection to the skin, it will get ripped off, rather than cut off, unless the piece remains absolutely stationary.

A planer is absolutely brutal, as a severe example. I tried seeing if I could just barely skim the back with the planer. A pass below 0.005" (I was literally moving the handle ~1/8th of an inch at a time to find the lightest possible contact point) resulted in every single piece being ripped off and chucked through the planer head and out the dust collector tube in fragments. If the pieces were big enough to be held down by at least one of the planer rollers, then I’d have more success, but at the moment of that transition between enough skin still holding it and not enough still holding it, there is nothing else holding the piece.

I’ve never looked into a drum sander to see if there is a similar gap between the sanding drum and the flanking rollers? I have to believe that at the moment of transition, the sanding drum is going to be the only thing in contact with the piece, and it’s going to go flying. Even if not ripped off, just the ejection could be destroying edges and lobes. These are puzzle pieces, plenty of them aren’t capable of withstanding such abuse. Look at the wing tips on this guy: https://photos.app.goo.gl/RiSc7S93yDHLPvXa9

A drum sander test is in my plans if I can’t find anything else, but I don’t have much hope that it will actually work.

Laser became appealing because the laser imparts no torque to the wood, removing the problem of hold down entirely.

Another possibility I would explore if laser isn’t going to work, is creating my own “drum sander” that has sub 1/8" clearance between the piece hold down and the sanding surface. Possibly even using the same concept of a flatting jig to have the stock stationary, and the sander the part that moves. I’m pretty sure that any ideal sanding solution would also be much lower speed than basically any belt sanding type tool currently operates at. Controllability is critical.

Barring everything, I just keep doing what I have been. But my throughput could be so much better if I could solve this. And I believe it IS solveable. Just a very unique problem.

4

That’s an friggin’ awesome puzzle!

I’d love to test the idea for you, but as I have a lot of my own irons in the fire at present, it would be quite a while before I could get to it.

So here’s a variant of the idea that has an extra step, but that I think would almost certainly work.

  1. Edge-clamp the workpiece on the CNC, and cut the parts with a downcut bit, leaving an onion skin to keep them securely connected to the waste material.
  2. Unclamp the panel, and flip it over. Now apply some sort of single-sided tape to the machined side. Maybe packing tape. Stick it down well with a brayer roller or the like.
  3. Now flip the panel over again, and run it through a drum sander to remove the onion skin. The sanding-drum force on the jigsaw pieces is almost purely lateral, so the tape should easily overcome any small lifting force.
  4. Harvest your finished jigsaw puzzle pieces by simply peeling them off of the tape.
2 Likes
5

I use the taytools tape to hold down right now. It’s sufficient when there is a skin left with about 40% coverage.

I have tried flipping a board over and taping the top of the pieces and using a surfacing bit to remove the skin. Aside from it being much slower than my belt sander approach, it also rips off pieces at that transition point, the amount of tape on each piece is just not enough hold.

I get that drum sander is lateral torque, and still think that it would result in the same outcome. Any torque at all where the piece isn’t held firmly in place by something other than the skin until after the skin is removed is always the problem. There will always be transition from held-by-the-skin to not-held-by-the-skin, and unless the piece is held some other way OR there is zero torque applied, this transition will continue to damage them.

I don’t mean to sound combative. I really appreciate your troubleshooting discussion with me.

Other digging is starting to suggest to me that I’m going to run into at least one of my non-starter points from my original post. Which means I’m thinking more about building a customized solution. Perhaps faint bevel plates with around 1/8th clearance to a belt sander (custom built one with adjustable speed) against rubber belt bed, also rotating. So the pressure applied by the decreasing clearance plates is magnified by the friction of the rubber belt, and pieces can be held until after the sanding belt is completely cleared.

This might be doable with a steel/iron bed and rubber rollers, but I suspect that due the 3 rotating bodies … roller, sander roller … I would have too big a gap between and a loss of hold down force. The rubber grip would need to move to the bottom.

Or, lightbulb moment. If I used a jointer with a rubber belt overhead to hold it down through the process. Each piece would be held by the downward pressure of the belt before the jointer cutting head reaches even halfway down each piece. Hmm… I think I’ll start here.

1 Like
6

just spit balling here but what if you flipped the puzzle over onto a rubber mat designed for holding small parts to sand them and used a random orbit sander? when pressing the sander down on the onion skin the force would hold the piece down.
Then when the onion skin is gone, lift up and the parts should stay there. Just a thought. I would try it myself, but my machine is just on its way.

1 Like
7

Not that much different from what I currently do, using a bench top belt sander and just passing the boards across that. Faster than a ROS, and since I am holding the board, I’ve found I can tell by the vibrations in the board/pieces when they are about to let go, and can remove by hand and return to sanding.

It works. Just again, this plus the clean up sanding after is 80% of my production time, and am looking to make progress cutting that time down.

2 Likes
8

what about tumbling?

9

I don’t know what you mean by tumbling?

10

the parts would go into a revolving drum with a specific kind of media appropriate for the material and tumbled till the onion skin is gone. you may have to research to find the right times and media but it could work. you might even have to carve them a bit differently to allow for some loss of material but It would work.

1 Like
11

Good job with describing your problem so well.

My experience with lasers is zero - but I have used an oscillating knife head on a machine which is a similar periphery device. To make them both work, you can make a quick-release device to attach the laser to, and then run the laser part program specific to that set of pieces. This touches on your non-starter, as you didn’t want to un-mount anything. I’m not completely convinced that a mount can’t be created that reliably puts the laser back in the correct location.
Machines I’ve used in the past have an air cylinder that brings an oscillating knife head down, reliably. A similar system for a laser would be a lot of work to get working, and might not be worth the effort. A quick-release system would be a natural middle step to verify functionality.

Sounds a bit like your problem would be solved if you found the right tape that sticks well enough to send the parts through a planer, or drum sander.
3M double sided carpet tape stands out as a good tape to try. You can get it at Home Depot. It’s insanely tacky, but after a few hours it starts to release on it’s own.

Interesting problem you have here. Good job methodically thinking through solutions.

1 Like
12

Hmmm. What happens if you forget the laser all together? You program the CNC to cut all your pieces to onion skin, run double sided tape over it all, flip the whole panel and stick it down. Then you mirror your program and run it again to free all the pieces, tada!!! Or no?

1 Like
13

@HardtailSr:
I can’t imagine a scenario where the skin would be subjected to enough removal that would not affect the pieces. It took a fair amount of trial and error to find that 0.006 artificial blade kerf on the pieces is ideal, feels good, and normal wood movement doesn’t cause problems. 0.007 makes things way too loose and they fall apart too easily. 0.005 feels good right after cutting, but normal wood movement results in puzzles that have inconsistent fitment from piece to piece. 0.004 makes a full puzzle lock up likely at some point in the future. These tolerances are less than the width of the skin.

@MikeXYZ:
Yeah, I see what you mean with your knife head, and agree that such a solution could work. But would also be quite involved to fabricate, and probably cost prohibitive.

Carpet tape does hold it, but the “few hours” part is the killer. It means each board is at least several hours to cut, rather than the 15-20min cut time plus ~30 min sanding time. I can’t see how that would be a time/efficiency gain. Plus the residue clean up needed anyway.

@Off-kilter:
Tried it. The idea is good, and it would work from a torque/destruction perspective. The barely-anything bit engagement with the wood is not enough to overcome the tape grip. However, it requires either

A) every cut program for every board has to be exactly centered on the X axis AND I have to cut each board to length exactly (at least to near-thousandth) of what the program is expecting, which is difficult at best since it is measuring by tape measure and eye.

or B) Every program and every board would have to add indexing peg solution that works with my spoilboard, and likely I would need the board lengths to become standardized. Right now they vary based on the tetris-ing of the pieces and how many I want to get per board/per wood from 9in lengths to 28in lengths.

The core problem is simply that when mirroring, however much the real board length deviates from the expected program length, the backside cut is too far left by double that deviation. I could of course still clean up each piece, but that then is back to where I am now with hand clean up.

Not impossible, but the risk of error is very high and when tripped puts me at a net loss in time.

The ideal solution would require no movement of the board at all to cut out the pieces. Every movement introduces unacceptable error rates.

So exploring a custom solution using a jointer, but still curious if anyone has any direct knowledge on if a laser could or could not handle each point.

2 Likes
14

Following up on Off-kilter’s idea, it seems to me that if you place the workpiece origin at a corner of the workpiece, and that workpiece corner is reasonably square, three simple fences could provide precise repeatability for two-sided machining, independent of workpiece length and width.

  1. One fence, presumably along the Y-axis, would establish the X-origin of the workpiece for both machining setups.
  2. For the primary machining setup, a near-side fence along the X axis would establish the Y-origin of the workpiece.
  3. For the cleanup-machining setup, a second (far-side) fence along the X axis would establish the Y-origin of the workpiece. It must be precisely parallel to the other fence, and you must measure the fence separation precisely when setting up your CAD model for the clean-up pass.

To be clear, the workpiece would be positioned between the two parallel fences for both setups, but with its longitudinal reference in contact with only one of them.

That fixturing design may get the workpiece positioning sufficiently accurate remove the onion skin with an end mill. But I’m thinking that instead, a very light chamfering pass along the puzzle pieces would reduce the positioning-accuracy requirements. Use a sharp V-bit, and shoot for maybe a 0.015" chamfer. If you’re off 0.005" one way or the other, no big deal.

The chamfers may even have a salutary effect on puzzle assembly. If so, you could even consider chamfering side one prior to making the perimeter cuts.

1 Like
15

That’s not a bad idea. My 30 degree pencil tip would work too for that. Although having to re-set the Z zero changing the bits would break my no-re-zeroing goal, so I’d stick with the 1/8"

My wife has been making an argument for me to order a 2nd 1F machine anyway. I’ve been arguing against it because it isn’t my bottleneck, but a 2nd machine could be set up just for this. I could even run vacuum for both, the first because I never cut through, and the 2nd because a v-bit’s torque would be really low. Although a pricey solution, workable.

Hmm. Worth considering.

1 Like
16

So everyone has beaten me to my rethink of this process. I went to bed and as usual instead of going to sleep right away, I pondered important thoughts like what if you used a V-bit set to shave just a hint of a chamfer. Wouldn’t that save a bit of cleanup? As for flipping the panel, I know you’re reluctant to do that but I’ve found if you include locating pins in your program you get excellent results.YMMV.

2 Likes
17

It’s unclear from your post whether you are using the CNC to detail carve the pieces or merely to cut them apart. If you’re simply cutting them, then given the ~8mm thickness of your stock, my advice would be to just get a 40W or higher laser module (or a dedicated CO2 or diode laser) and do the entire cut with the laser rather than trying to CNC them and then finish them off with the laser.

If this isn’t acceptable because of the charring on the edges, and you really want to stick with CNC cutting the majority of the way and then using the laser to cut the onion skin, this should be possible but will likely require some fiddling on your part. Lasers are extremely sensitive to focal length and the JTech lasers come with their focus set to 1/8" (~3mm) below the edge of the shroud. You will usually want to have the laser focused at the middle or bottom of the material you’re cutting, so for 8mm material, you want the focus between 4-8mm below the surface of the material. This will require manual refocusing of the laser on your part: a hassle, to be sure, but something you only need to do once unless you change the material. These lasers usually have an adjustable focal point between 1-3 inches (25-75mm) from the laser head, which means that when you are doing this refocusing, you can account for both the material thickness and the protrusion of the router bit at the same time. If the bit protrudes 10mm below the laser shroud, then you can calibrate the laser focal distance to 20mm from the shroud so that the bit can clear the material during the laser operation.

The larger problem you’re going to have to deal with if you want this to be fully automated is dust extraction/ventilation. Unless you plan to star in a Smokey the Bear commercial, you need to make absolutely certain that your dust collector never sucks up burning embers from the laser. This means that it absolutely cannot be running at the same time the laser is cutting, and the laser needs a completely independent ventilation system that is never running while the CNC is cutting, lest it suck up lots of highly combustible wood dust into its hose. Most of the time, people do this by swapping hoses between the CNC cut and the laser cut and manually switching the dust collector and laser ventilation system on and off. It is possible to hook up automation to make the Masso controller do this automatically (I have no idea about Buildbotics), but it will require a certain comfort level with wiring and G-code post-processing to do it as well as a dust boot that supports simultaneous connection of both the dust collector and the laser ventilation hoses. I’m sure someone has made one of these, but I’m not aware of one off the top of my head, so it might need to be custom designed/3D printed if you can’t find one. And again, this is a very serious safety concern. You need to be absolutely certain that those two dust extraction systems are never, ever confused and running at the wrong times or you stand a very high risk of burning your shop down as burning embers from laser exhaust ignite wood dust left behind by the CNC.

Personally, I wouldn’t take the chance on that and would instead run the operations manually, finishing the CNC pass, vacuuming off the parts manually to make sure stray dust is cleaned up, then swapping the dust boot for the laser exhaust and doing a manually-initiated second pass with the laser to clean it up. I suspect this will get you what you want without as many safety concerns or the hassle of trying to make the controller manage independent dust extraction systems.

One last thing: if you want a laser for cutting, do not get a sub-20w module. You want a minimum of 20W and preferably as high as you can afford. Cutting is possible with lower-power lasers, but it requires very, very slow speeds and tends to induce a lot more material charring because the laser has to stay in place and rely on evacuation of material in one place for so much longer. If you can afford the 64W laser for the Onefinity, I would strongly encourage going with that rather than trying to pinch pennies and get a cheaper unit. Without knowing the economics of your business, it’s impossible to say for sure, but my guess is that it will pay for itself many times over with speed of operation and the reduction in necessary manual cleanup.

3 Likes
18

@Cypren

I’ve been pretty busy, but want to thank you for your response. Ultimately it was the point about the risk of fire mixing the two that steered me away from pursuing this as a solution. Swapping the evac methods manually would never be a concern, but a year ago we had a fire (unrelated to CNC) that leaves it scars. The questionable success of the idea doesn’t come close to outweighing that kind of risk. It is a point well taken.

I picked up a smaller separate laser engraver, and while I found that at 10w, a laser can definitely cut the remaining skin necessary, the precision necessary probably isn’t transferable moving the stock.

I’m going to be pursuing the jointer with overhead slow belt option. I’ll have to design and build it, but once I have a 6" jointer, not particularly crazy to do within my skill set.

1 Like
19

@insedra

Quite welcome. Depending on which specific 10W engraver you picked up, you may or may not be able to get the necessary precision to transfer the stock and cut the material cleanly.On many lasers, you should be able to set up physical references and stops so that you can transfer the material from the CNC to the laser and align it to the same position and orientation every time.

Everything I’m about to say hinges on two critical assumptions:

  1. You’re using physical references to make CNC cuts in stock that’s square on at least two adjacent sides so that the outputs always have precise known offsets and orientations. If you’re just slapping down oversized stock on a CNC bed and aligning the cutter with the good old Mark I Eyeball, this isn’t going to work.
  2. Your laser has homing switches and the ability to operate from an absolute machine coordinate system rather than all jobs being based off the initial position of the laser head. If you have a cheap laser without homing switches, it usually isn’t that difficult to add the functionality so long as the controller board supports it. It just takes a small amount of comfort with electrical wiring and sometimes basic soldering.

Given that most enclosed systems are significantly more powerful than 10W, I’m guessing you went with a cheap open-gantry system like the Xtool D1 or Elegoo Phecda. You can use your CNC to make a base plate for the laser: machine a piece of MDF with precisely positioned holes that lock the legs of the laser into a consistent position, then use the laser itself to inscribe a grid from its absolute origin to the edge of its cutting area for reference.

Once that’s done, take the MDF back over to your CNC and ensure the laser grid is precisely aligned with the machine axes. Do NOT assume a cheap laser’s machine frame and legs are perfectly aligned with its axes! Use an engraving bit in your CNC, hover it just barely over the surface of the MDF, and jog the head to the outside corners of the grid to ensure that the pointed tip precisely traces the outside of the grid.

Once aligned, cut a set of holes along the outside of the grid for bench dogs or dowels to use as stops. With that done, you can lock the laser back into the MDF board, insert your stops along the edges that the CNC uses as references for the product stock, and now your laser has a repeatable and consistent way to operate from the same reference as the CNC. Once you align the cut pattern once, you’ll know the exact offsets to use for every subsequent cut.

Hope this helps.

1 Like