Drag knife not making sharp corners

Well, I already had the hicwic mount drawn up from some other projects so it was pretty simple to just add the bearing housing to it :wink: F360 is nice, but doesn’t run nearly as well on my computers as Onshape so I do most of my stuff in Onshape. But I want to try F360 for CAM since it will run on my macbook and Estlcam won’t…but I do love estlcam and my macbook is just an air so it doesn’t run F360 very well.

Look forward to seeing how it goes. Main reason I haven’t tried it is I don’t have excel…and don’t know how well it will work in google sheets. And it’s setup for imperial measurements and I do all my CAD/CAM in metric…so not huge on dealing with conversion issues. I keep hoping estlcam will get drag knife actions for non-estlcam firmware.

So looks like it will work:

[attachment file=46566]

I’m not sure about the spring I used…may be a little stiff. But I had one that was maybe a little too stiff and one that was definitely not strong enough to do anything. Figured I can always cut down the stiff one if I need it softer.

Now I just need to figure out the gcode. Since the donek tool is all in imperial I’m not sure how to go about using it if I even can as I do all my gcode in metric…Even if I worked in imperial in CAM my gcode is all output in metric since that’s what Marlin expects. So not sure what the best approach is here. I may just look at what their code does and see if I can whip up something that does in another language.

@James Donnelly - Any tips based on what you’ve done so far with your drag knife?

I’d like to try this on some poster board I need to cut…but I’m not sure that’s going to be thick enough to work well, may not have enough for the blade to bite into when rotating. I could try it on some dollar tree foam…but cutting the current foam with a blade is really sub-optimal as it tends to tear a lot…and I can seldom cut a full sheet of foam with one blade without wearing out the edge of the blade anymore.

Still, I do want to play with this some just need to figure out the best way to adjust the gcode.


Lookin’ good! My spring is quite soft, and I think it does help, but I’m not certain. I think the lower pressure on the piece leads to more stability in the blade. Lower pressure does mean more passes though, but I’ve found pushing the speed doesn’t hurt to a point.

Regarding the spreadsheet, I suspect it would be hard to translate to Google Sheets or Open Office as the macros do some of the work. I’m willing to try it out in imperial and use Excel. I think the best way forward for general use with that would be to implement it again with Python - not a small undertaking unless you’re jeffeb3.

My two hopes for cutting thick card (no so into foam board, no DollarTree here in UK, and it’s not so cheap) are:


Last night I managed to increase the offset a few mm by moving the blade in the current arrangement. It didn’t seem to help. I might end up trying to re-model the design to allow for adjustable offset. Would involve possibly splitting the bottom assembly, mounting the lower part in a groove and indexing with a screw. The problem here is that with increased offset, your minimum cut radius is going to increase.

Blade shape

The standard blade leaves a triangular protrusion. This means with deeper cuts there is quite a lot of blade length in the material asking to turn and snag. When I’m next in the shop, I plan to grind down the back edge of a standard blade to see if a much steeper blade angle (like an Exacto or steeper) will help. I’ll be wearing safety glasses when I try it out :open_mouth:

Well, I did have some softer springs…but they were MUCH softer. I’ll just have to experiment and see how it does I guess.

Well, I’m a web developer by trade and have done some similar projects…I’m just not very experienced with excel and am not sure if they’ve got the macros encoded or protected in some way that will prevent me from figuring out what they’re doing :wink: I’ll probably give DXF2GCODE another go. It can supposedly do drag knife corrections - but I’ve never had any luck getting it actually do…well…anything.

I noticed in the Donek video of cutting adams foam he mentioned that the top wasn’t super clean and that switching to an exacto blade instead would probably improve the cut quality. So you’re probably on the right track there.

BTW - I uploaded my HicWic mount version to Thingiverse: https://www.thingiverse.com/thing:2603958

Kind of surprised at how many people are adding it to collections and liking it already :smiley:

Good to see your creation on Thingiverse. Drag knives must be cool after all!

I’ll get around to measuring spring pressure with a scale once I start bringing things together.

I spent some time looking for evidence of what exactly the drag knife support amounted to in dxf2gcode, and couldn’t work anything out either. I didn’t look too hard because I’ve heard good reports from people using the Donek spreadsheet. I haven’t looked at that in depth either but at first glance, the macros are needed to support working inline on a file you supply. I don’t expect the calculations to be macros or protected.

Anyway, progress with the blade mods. I shaped the blade as discussed the grinder. First I mounted it oriented tall and steep like in this pic:

[attachment file=46652]

But I found the snagging was even worse than the standard. I was just going to give up on the blade mod approach, when it occurred to me that when the blade angle wrt the cutting material is too parallel, it’s going to be harder to cut through. So for an extreme test, I mounted the same blade (cut down a bit more) like this:

[attachment file=46653]

Well this time, it cut out the circle very nicely and reliably. Obviously there is too much of an offset - it’s fine for circles. But the key seems to be the shallow blade angle for thick material.

Before I did these tests, I’d started modelling the adjustable offset version of the knife:

[attachment file=46654]
[attachment file=46655]

But now I am thinking of working in a blade rotation angle mod, so you can independently control offset and entry angle.


I found that getting dxf2gcode going on macos is a bit tricky…there’s no binary distro and the source is tough to build since it has requirements that also don’t have easily installed versions available. So…

I waited until I got home and tried it on one of my windows machines. Finally got it going…and seemed to generate some drag knife friendly gcode…but when I cut with it the results were less than spectacular.

My phone battery was dead so I wasn’t able to get a photo of the results…but they were pretty ugly. Corners were torn and even some straight lines weren’t very straight. It seems the 8mm bolt can shift around in the bearings a little which causes some slop. But the bigger issue is the corner actions didn’t seem to clean up the corners very well.

I suspect part of the problem is that even my brand new fresh from the package blade wasn’t super sharp at the tip which cost accuracy. But the main problem seems to be it was keeping the knife down too low when doing the corner moves so it was ripping and tearing stuff as it rotated. Not sure if it’s just me not being familiar with the settings in dxf2gcode or if it’s just dxf2gcode doesn’t do great corner actions…or if it’s just my blades.

Will do some more experiments…but overall…not super impressed. My laser cuts poster board great other than leaving the edges darkened, and my needle does my foam (there’s no way the drag knife would do foam anywhere near as well, the blade just can’t do the tight curves I can do with the needle.)

It’s nice to have in my toolbox and I can see it being nice for things that don’t have sharp corners. But overall…I’m kind of meh on it.


Yes you said it right. I don’t think it should depend so much on an ultra sharp blade though.

But yes, this thread started about sharp corners, whereas you and me are also interested in thicker materials. But unlike you I don’t need sharp corners because I’m actually trying to make a scraplight (like Graypants).

I did say I’d look into sharp corners and I will, but I think I’m on to something about thicker material. I measured offset of the design we’re using and in a standard set up, it’s 4mm at the tip of the blade. This means that with 4 thick material, you’ve lost half your offset. I think cutting 7mm as I’ve been with multiple passes, when the blade is at the bottom, the top of the blade is touching the material with barely any offset. Having a more vertical blade reduces this effect, but makes it harder to cut through the material.

I think the answer is to calculate offset according to the thickness and blade angle, so at the top of the material, you have a good enough offset for a follow effect, but not too much to increase minimum radius. Then cutting in one pass I think. I’ll carry on with the variable offset version design.

Where I think you’re correct is that thick material + drag knife + tight radiuses / corners = fail. You can’t have a more upright blade, and that mean the cut line at the bottom of the material is always behind the one at the top. You can still to good stuff with the big bed cnc:

Might have to keep going until the giant laser gets built!

Well, I’m not even trying thick material yet. Figured I’d start with posterboard since it should be “easy”.

This was the first result:

[attachment file=46665]

Ick. Oh, and I forgot the worst part earlier…the gcode dxf2gcode created crashed Marlin before it finished. I ran the code with the knife lifted up first for a dry run and didn’t notice…I just thought it didn’t return to home after the cut finished - I did think it was odd that my LCD was showing all 0’s on the coordinates though.

Then I ran it on the actual posterboard…and when it went to do the last few inside cuts marlin rebooted. I haven’t looked at the code yet to see just what command it sent that caused that :frowning:

Because honestly the cutting results overall at that point were so unimpressive.

I probably will play with it more…but I’m really not in any rush since I don’t have a need for it.


@Jason, sorry to hear it didn’t work out great with the foam. I’ve found a sign maker near work, and they agreed to me grabbing any scrap I want, so I’ll be trying at some point with foam board. I don’t know what posterboard is called in the UK. Only UK links I’ve found reference plain old card, with no foam. Is that the stuff?

I don’t know why that would be a challenge to cut, but I will know if I try. I suspect something is wrong.

The 7mm cardboard is cutting very well on a prototype of my variable offset knife now. I didn’t bother with a fixing screw and just set the offset to +2mm of the initial design and glued things up. Pretty clean lines and no snagging now - might actually be practical for medium to large cardboard structures without tight corners

I plan to try running a router bit through the card as an alternative to the knife, but by all reports you get fuzzy / rough edges, and it’s very hard on bits. Not fun if you have 50 layers to cut.

I’ve also ordered the emax brushless motor + esc, so will be trying with a needle cutter too. Hopefully this can work as this would be the best. The nuclear option will be to make one of these:

PS: I agree about the play in the bolt seated in the bearings. I guess the answer is to get some 8mm rod and thread both ends to accept a nut, that way it’s a smooth surface travelling inside the bearing so the spring movement will be smooth.

That thingiverse link is nuts. If you do end up trying it please please share the results. Seems to be well thought out and complete.

Yeah, posterboard is just cardstock. It’s not hard to cut…just not happy with the performance of the drag knife overall. The slop between the 8mm screw and the bearings is the biggest issue…that and generating proper gcode.

That TOCK is pretty crazy. The extra axis to rotate the knife is pretty nuts…I see he included something for generating the required gcode (for mach) but wonder how flexible it is. And honestly overkill for anything I want to do :smiley:

There was talk about something similar in the flitetest thread about the needle cutter. There was also talk about using a similar mechanism for creating the needle motion…but in the end it just adds more complexity without any major benefits.

Honestly the TOCK just seems like more to go wrong than a needle. It’s basically the same thing just with a blade instead of a needle so it has to rotate. But with a sufficiently fast needle timed to your head movement there’s no need to deal with rotations.

My needle cutter does work on cardboard now that I added the glue to hold the needle on the bearing…it does do better a little slower than the speed I use for cutting DTFB though. Maybe I’ll try a bit more cardboard this weekend just for fun…



I’m just not interested in putting a lot of time into getting the drag knife working better since I already have tools I’m happy with that do everything I need. Still fun to have built it…and I do see how it can be useful for some things. Just nothing I do regularly.

The router tables from Esko have vibratory knives on them. You can’t really see the blade moving, it’s like a quarter millimeter or something like that. They’re also servo controlled to turn with the cut. They could zip through corrugated like a hot knife through styrofoam though. Wicked accurate too. Could cut vinyl sticker material without cutting the paper backing. Some of the blades they used were solid carbide.


Yes, chances of me making one are slim to none to be honest. I’d rather build a massive laser.

I mentioned the sign shop near my work for free foam board. They have a 10’x15’ bed machine and use tangential oscillating cutters. I didn’t see it in action but it looked alot like this: https://www.youtube.com/watch?v=q-zehi0UQUI

I asked to see the blade / cutting tool though, and was surprisingly small - looked a bit like the blades that go in the chinese small drag knives.

I guess they use these cutters industrially because they are very fast. I’m not sure the TOCK will ever be faster than a needle cutter or laser, and the blade looks too long to do fine work anyway.

I’d expect the TOCK to be faster. The needle isn’t super fast…but I’m not in a big hurry :wink:

A couple people have tried speeding up the needle to use it in production work but haven’t had much success. I currently run mine at about 9k RPM on the needle and 15mm/s feedrate. That lets me cut an average 20"x30" sheet of foam in about 15 minutes. Though some more complex designs take longer and some simpler ones take less…but on average it’s about 15 minutes a sheet.

The problem with speeding it up is tricky. First spinning the flywheel faster gets hairy quick, the forces ramp up and things start to fall apart faster. Second even with the needle moving quicker it doesn’t eliminate needle drag. In fact in sometimes makes it worse. Since the machine movements and the needle movements aren’t synchronized the machine is moving while the needle is at all points of it’s motion. The needle only cuts right as it hits…but the rest of it’s stroke while it’s fully in the foam it’s a potential drag.

So if you watch it closely you can see it start to deflect and drag on the downstroke, then re-align on the upstroke. It’s still fast enough that the individual punctures overlap and you get a clean cut - but as it’s moved faster laterally the drag on the needle goes up even if the speed of the needle increases.

That’s why I have to slow down for cardboard - it has more drag than the foam since it’s stiffer and the problem gets worse. A shorter needle would probably help (mine is currently 2-3mm longer than it needs to be - I left it long so I could sharpen it more times before replacing it.) but the best solution is just slowing down. I was running around 5-10mm/s to get reliable cuts in cardboard without stressing the needle. At 15 which works well for foam it was just knocking the needle off constantly.

Actually - now that I have the vac table and my cuts are looking even cleaner I should try speeding up the foam cuts a bit more…I’ve actually run it at 20mm/s but the cut quality diminished. So now that I’ve improved cut quality with better work holding I may be able to give it a bit more speed. I’m also curious to see how cardboard does on the vac table and if I can do any better with it now.

@Jason I’m with you on speed, not the top priority. Love the vac table by the way - I voted, but you didn’t need me, you’re killing it.

Keep us posted on the cardboard. If you get a chance to try the double wall stuff, 5-7mm, it would be great to hear your results. If it’s reliable I don’t care how slow.

I have a question now about whether oscillation is as important as tangential control. Interesting test here, which looks more accessible than the TOCK:

Industrial example with no oscillation (claims small radii are possible, I thought maybe oscillation helped with that):


The nicer vinyl cutters use this but they also have some crazy expensive software to go along with it. Do you have any leads on some software that can do the tangential gcode? I am sure it would be pretty easy to get a stepper on there, but the code is the hard part (says the mechanical guy).

As I watched your linked video the next one to play was tangential gcode from inkscape…dam…If I would have only waited a minute to post sorry guys.

I would guess oscillation mostly helps when the blade starts to get dull. With a fresh sharp knife just dragging it works very well - but as it gets dull sometimes a bit of sawing motion starts to help.

So I’d assume the oscillation is more a “backup” so you don’t have a big job go south if the blade gets dull partway through. That and helping production shops get the most from their tooling.

Interesting that inkscape gcodetools can do tangential…I have a spare stepper sitting here…and my wife and daughter are out of town this weekend so I can focus on my projects…hmm…

Ryan, yes I was imagining myself being willing to get coding to make a post processor translate tangential moves into the gcode - then I found that video/information that Inkscape can do it.

Jason, I also have a spare stepper. Interesting project, eh? What you say about oscillation makes sense. Are you thinking that the design we’ve used could be adapted or something else.

Seems to me to be a good starting point, although you need to be able to set a zero offset. I was planning to improve my initial adjustable offset mod by beefing up the material around the sliding dovetail and adding a screw. If that’s useful I can continue, but if you have a simpler approach in mind…

I honestly don’t have any particular approach in mind…not even sure if I’ll actually pursue it or not. I’ve kind of got a backlog of planes I’ve cut but haven’t actually built yet and some I’ve built but haven’t finished wiring…and now that the weather is getting nicer here I’m itching to get some of them in the air.

So…don’t wait on me to make any progress. But one evening I may get an itch and try something :wink: