X and Y axis don't want to produce smooth curves

So I just finished assembling my (very large, about 5 feet square) CNC. Everything functions very well (awesome design, I really love it), but I can get the CNC to produce smooth curves. I’ve been testing it out with a Sharpie instead of a router (since it’s just easier). I’ve been plotting a few simple SVG images and have noticed that when both the X and Y axes move at the same time, they don’t move smoothly. You can tell in the attached picture in the V that the lines filling it in aren’t exactly straight.

I’ve checked everything I can think of, but can’t figure out how to remedy the situation. I’ve tightened the timing belts and tightened the carriages. Both of these seemed to help, but I still can’t seem to eliminate the issue entirely. I’ve also watched the steppers on each side of each axis and they appear to be moving smoothly (there’s no visually obvious step that’s causing this). I’ll also note that the Sharpie is well secured and isn’t causing the ripples. The ripples seem to happen in the same place.

Does anyone have any ideas of what I can check or fix?

Did you buy it from me?
Check your pulleys, can you pull them off the end of the shaft?

I did not.

Thanks for the suggestion. I just double checked all of the pulleys and each is attached to its shaft quite well. I can’t move any of them independently of the shaft of the stepper motors in any direction (rotationally or along the shaft).

You might need to take the belts off and verify all the steppers are working. Seems like one might not be.
Are you using my firmware on a ramps board? what drivers? If you step rate is too Low it will produce jagged lines.

That actually just looks like there is something under the paper. Do you happen to have a small dry erase board and marker to try it on?

Do the axis move smoothly when the belts are off? It could be a rough spot on the conduit as well

Ok, I didn’t take the belts off, but I’m pretty sure all of the steppers are working. I moved each a very small amount and each stepper seems to be responding. Plus, they all resist me trying to move them manually after doing that, which tells me the motors are getting power and holding their positions.

I am using your firmware, the version called Marlin_16_32_090615 (I think that’s the recommended one?). I’m using the DRV8825s each with 3 jumpers each for 1/32 step mode.

Yeah, the bottom right looks a little funny because the paper wasn’t 100% flat (it was pretty close) and the sharpie was just barely skimming the surface. I have other examples which lead me to think it’s not the paper but the CNC that’s causing the ripples.

The conduit is smooth but could definitely be smoother. I had thought it might be contributing to the issue before, but wasn’t convinced. Would you recommend sanding it?

Thanks for your help!

Does XY move smoothly by hand in a diagonal? Any binding? If you draw a simple diagonal do those steps/zigzags have any pattern? Any differences in an XY move from lower left to upper right and back and lower right to upper left?

When I move the tool holder by hand I can move it relatively smoothly (no binding), but I do feel some vibration. I can’t tell if that’s caused by resistance in the steppers or bumps on the conduit. I’ve attached a few pictures of a single long diagonal line (about 3 feet). I don’t see any patterns, which I suppose might suggest that the steppers aren’t to blame. Let me know if you disagree with my assessment here.

I think this leads me to believe that the problem might be caused by less than perfectly smooth conduit. I picked up some emery cloth to try to sand out the minor imperfections, but I’m still not 100% convinced this is the issue for sure (I’m also not sure that this is the best way for me to smooth these out). Has anyone else had this kind of issue? Any other ideas?

Thanks for the help so far!

Are you sure your pen is not just loose in the holder? What surface is under the paper? A pen is a poor tool to try this test with unless it is spring loaded. The one in my video is a spring loaded liyu pen, made for this kind of thing.

I would say strap a cutter on it and make a test cut in some high density foam.

Don’t worry about the conduit, it is smooth enough, promise.
The belts and steppers should provide enough drag to where you can’t feel anything anyways, unless you take the belts off.

I just used standard Home Depot conduit that was definitely not smooth at the beginning and I never had any problems. Now the zinc plating is worn off on three sides of all my tubes by use. All my felt time pen plots have smooth lines, even when attached poorly with duct tape.

I’m also using DRV8825s with 1/32 microsteps. My steppers are 480m Nm.

I’m pretty certain that the Sharpie isn’t causing the weird ripple. It’s in an adapter, inside the normal tool holder, and tightened down pretty well. It’s also just barely touching the surface of the paper, so I don’t think resistance is causing it to move. I’ve been testing this with a sheet of MDF under a roll of thin paper on which the Sharpie is drawing. I’d say it’s very flat (not contributing to error). Yeah, there’s definitely no spring involved, but honestly, it really seems like this should work with no problems.

I’ve been testing this way for two reasons: 1. it’s easy with minimal waste, and 2. I’m still waiting for an adapter so the 1/8 in bits I have will work in my router (which normally accepts 1/4th inch bits). I’ll definitely try cutting some foam as soon as I get that adapter collet in the mail.

Yes, those ripples are strange. Did you use a regular fine point Sharpie or an ultra fine one? Just trying to get an idea of the scale of the ripples in the photo.

How are you generating the gcode from the SVG files? Can you preview/simulate the gcode? Can you use another method to generate a simple 45 degree box and circle outline in gcode just to cross check the software? I’m, kind of reaching for ideas right now.

It’s a regular Sharpie, so not a fine point. I should have thrown a ruler in there for scale. It’s making a line about 1mm wide, so the ripples are fairly noticeable. I took another picture with a ruler and uploaded it.

I’m on a Mac, so my (native) options are limited. I’m currently using jscut, which seems to be working nicely. I had the same thought that maybe it was generating gcode that had ripples in it, but jscut allows me to preview the output. I didn’t trust that either, so I downloaded another app called CAMotics (I think) that also previews milling gcode. I’m pretty familiar with this sort of thing from 3D printing, but the milling side of gcode is definitely new to me. I can virtualize Windows and run estlcam as recommended, but I’m pretty sure that jscut is doing an ok job for the time being and the virtualization route is just a few steps more than I’d like to do unless necessary. I’ve uploaded my sample diagonal test code (which I stopped after it drew the first line). As far as I can tell, this gcode contains straight lines, not ripples, so it has to be the machine not the code.

Oops, it won’t let me upload gcode, so instead, here’s a screenshot of camotics looking at the MARVEL logo I originally posted. You can see that the milling lines are straight (interestingly though, the edge of each character does have a stepped look to it, thought I assume that’s a side effect of the simulation).

You can .zip the gcode to up load it.
Try the test file at the end of this page. https://www.v1engineering.com/estlcam-basics

Good call. I zipped up my diagonal test code and uploaded it. I’ll try out the test code you suggested later. Thanks.

test_diagonal.gcode_.zip (3.33 KB)

I can’t test your file my machine is not that big, so I cut the first few lines down to 300mm instead of 600mm and it worked fine. So I don’t think it is your gcode.
You can preview your gcode in repetier make sure to check the show travel moves box.
Right now your clearance plane is .1mm you should raise that to something more like 2mm. Not that that would effect the diagonal just a tip.

What did you set your drv8825 drivers too? I had this issue with my extruder, I set the stepper too low and it actually stepped, I upped the voltage and it rolled through smoothly.
What size are your steppers? We use 76oz/in steppers and run .5v on the Z axis and .7-1.2v on the X and Y axis.

Note that Fusion 360 runs on a MAC and is free if you make less than $100K using it. There is a thread with post files on the forum.

Could the drivers somehow be doing whole steps? But then your scale would be 32x off so probably not.

Oh wow, thanks for pointing this out! I didn’t know about this one for some reason. Download now…

The big caveat with Fusion 360 is the steep learning curve. It’s really powerful and luckily there are a tons of video tutorials for it’s CAD and CAM. I think I posted a link to a good CAM tutorial on that thread. I can find it again if you want.

Great, I’m glad that looks ok. Yeah, I had it not going deep because I was just using a pen.

I hadn’t actually checked until you brought it up. I had initially turned them all the way down, then just increased them until the motors stepped without issue. Now that you mentioned it, I tweaked my settings to match what you recommended: 0.5V on Z and 1V on X and Y (I’m also using 76oz/in steppers), since this was higher than what I was using already. Unfortunately, it l didn’t seem to have much effect on the issue.

So I worked on the only thing I could think of that I hadn’t eliminated through all of these other suggestions, I started sanding my conduit with emery cloth. Now I’m about 90% sure that the bumpiness of my conduit is actually to blame for my non-straight diagonal lines. After about 30 minutes of sanding, my whole CNC moves much more smoothly when I move it manually and my test diagonal line looks much much better. I wouldn’t call it 100% fixed yet, but this seems promising, so I’ll do some more sanding soon.

Thanks again for all of the help!