Help Debugging Early Lowrider Cuts?

Hey All!

I finally did my first real carvings from the Lowrider, and it’s looking great, but there are some serious issues too, and I was hoping y’all would have the expertise to help me figure out what’s going on.

Here’s a photo album: https://goo.gl/photos/kRCSaKqtYQt4Fp5v8

I inspected the G-Code in Fusion360’s simulator, and everything looks fine, so I think something mechanical happened, but I didn’t notice what (I was milling about while this was running, it’s 7" wide and it took an hour).

The issues are:

  • Small scalloped shapes inside the circle. I think these might be due to the lack of a finishing pass?
  • The bottom left bone, and the bottom of the circle have serious geometry issues. This was done in two passes, and for the bottom left bone, the larger shape (the deeper shape) is correct. For the circle, it should just be a circle.
What confuses me is that the problem areas weren't the last places being worked on. If there had been belt slippage, it would have affected other areas too. For example, the eyes and nose are the very last thing, and those turned out fine.

Any ideas?

skull-coin.gcode (1.94 MB)

What is the first circle supposed to look like? Is that missing chunk supposed to be like that?

Which direction is the skull? It looks like the errors are all in one direction. Do you agree?

In addition to skipping steps (which is pretty noisy) you could have loose belts, or long zip ties or the wheels could be driving not perfectly straight.

For the last problem, I ended up installing some straight pieces of 3/4" plywood next to the wheels to force the wheels to track straight.

Do you have any stiff foam? That’s a useful test because it will remove any resistance from the equation. If you’re getting similar errors in foam, then it’s not skipped steps or belt tension (probably).

Your z rapids are too high by double (absolute must be under 8.4mm/s ), and try turning down you other rapids to 20 instead of 33 (just in case), see if that helps.

You problems are all on your table axis so make sure it isn’t dragging on the z rails too much and make sure your pinch rollers are not too tight, these don’t even need to be there for the most part so start by making them really loose.

I am constantly amazed by you guys (and Barry) in your reply speed! Y’all are some seriously valuable community members.

Point by point:

  • That first circle in the album probably shouldn't have been included. It was the same design, but much smaller. The GCode simulator in F360 showed that the 1/8" bit was only going to do the forehead, but I didn't mind, because I was mostly confirming the new collet and bits. (Technically, that original one was a different version of the same skull, where the skull was negative. But, the answer to your question is, yes, that blob shape was expected)
  • I do agree that the errors are all in the same direction. To move from the top of the head to the bottom, it would have to engage the rollers, *away* from the origin. That's actually got me wondering if it stepped on the 611's power cable, I have yet to clean up those cables.
  • The zip ties in some areas could be considered long, but they're pretty close to how long yours are, JeffB3. I'm not sure how to confirm belt tension, but they have a pretty good "boing" to them when I pluck them. Not quite acoustically tight, if that makes sense.
  • It's actually pretty likely that the bottom wheels are tight. I made sure to close them down on the rail. I'll try backing them off. They do seem to travel straight though. I had considered putting a dado stack in the table-saw to put a groove for them to roll, but it hasn't seemed necessary yet.
  • I still haven't bought foam, though I've meant to since the last time you mentioned it. We're headed in that direction today, so I'll swing by HD.
  • Message recieved WRT Z Rapid. I'm hunting around in Fusion for those settings, but I'm using the marlin post-processor from the software page.
All these things and more, I'll be trying today. Thanks to you both!

Things seem improved, but the issue certainly isn’t resolved.

  • Tightened belts as much as I could
  • Loosened the lower wheels so they aren't touching
  • Remade the GCode for the lower feed rates
  • Picked up some pink foam for testing
The new results are in the album, which is here again: https://goo.gl/photos/kRCSaKqtYQt4Fp5v8

The defect in the eye is because the cut finished and carriage just dropped in. Not sure what that little eyebrow chunk is. Once again I wasn’t able to see anything wrong.

Here’s what I’m starting to zero in on:

  • The defects are happening along the "roller" direction only
  • The defect happens at different Z depths, rather than affecting the lines drawn on a single depth (I think)
  • I suspect that at >8' long, the GT2 Belt is deforming a bit over the course of a job
  • I also suspect that the wheels aren't moving straight in some circumstances, though I can't see evidence of this
Just like before, the Z-drift happened towards the middle of the job, and seems to have corrected itself by the end.

Out of curiosity, what do you guys expect to happen when you push one roller assembly, but not the other side? For me it racks a bit, so I would need to push it from the center, or push both sides at once. The cut is happening in a corner of the bed, and the defects are happening on outermost areas. I wonder if the carriage being on the far side causes the roller motors to act asymmetrically.

Probably unrelated question: are you guys noticing any play on the carriage due to the asymmetrical layout of it? Because there is a (tiny) bit of gap between the tubing and the bearings, when the motor switches direction, it causes a slight pivot.

Also, do you guys happen to have a good calibration dxf or something? This skull is mostly for testing, I don’t have a lot of use for it otherwise, and wouldn’t mind using a file with more debuggable features.

Look at the gap between the side of the table and the z posts.

If the motors are on. I don’t push it. If they are off, you have to move them both. I always align it with the front of my table before powering on the motors and I move to the work area with software.

Not really. The bearings are touching, it moves freely, but doesn’t have room to rotate without flexing plastic.

I made a rectangle 10mmx100mm and that was really helpful for the time I found the wheels weren’t moving staight. The “log” of my problems are buried in my build log, but here is the photo album:

You’re saying it’s off by ~2mm in the direction of the long pipes, right? I am convinced you need some tracks. You can just clamp on a piece of scrap to confirm, but this looks like the same problem I had. I can’t quite explain all the problems, but for each z later, it was doing a few counterclockwise circles through the work. There was some hysteresis in the tracking that would cause it to travel to my left, which would cause a Monday cut on the left, and stairs on the right. That’s what happened to me anyway. I have had much better luck once I installed the straight edges.

If the motors are on. I don’t push it.
Agreed!
Not really. The bearings are touching, it moves freely, but doesn’t have room to rotate without flexing plastic.
This is not the case for me, I can give it a little shake. I've added a video of that to the album, it's 33 seconds long.
I did read your original log on the wheels, but I didn't really understand the issue until now. And I took some measurements (photos in the album), there does indeed seem to be a problem. The both rollers have a ~2mm disagreement from one wheel to the other. So, sounds like I need a track!
I wonder if this is a gyroscopic phenomenon or something. with both screws turning, the momentum
How has yours been treating you, anything you would have done differently? Thinking of either cutting a shallow dado, or using some angle iron from HD.
PS: How many Lowriders are out there, do we know?

And here’s a last minute theory before bed: on my machine, the wheels that are nearer to the table are also on the side where the carriage motor is, and therefore the belt. I wonder if the tension from the belt is slowly pulling that side together?

Do you remember which wheels were closer on your machine, Jeff?

If you tighten the bolt on the “backbone” of each of those rollers, it should deform a bit, and pull the side bearings into the pipe. Don’t overtighten (sorry, it has to be said). The other thing that might be happening is that the screws holding the rollers to the 611 plate might be holding the roller open. I’d make sure you loosen those screws a little before tightening the backbone. You might even have the holes a little off compared to your print, and you could widen them a little.

My low rider or my tracks? My low rider makes me very happy. I don’t use it every day, but I can get it up and running pretty well in a few minutes, which is really what I need. I’ve been trying to work on expanding my comfort zone with angled bits and working on setting up a pen and a laser. My tracks are fine. I don’t think about them anymore. They are 3/4" ply, which fits under the axle of the wheel and they are less than 1" wide, so they don’t interfere with the cart. I was careful when I installed them to make them perpendicular to the front of my table, and flush with the wheels (but not tight against them). I want all 4 wheels to be touching all the time, and I want it to be driving square to the front of the table, which is my reference face for starting the gantry square. If I were you, I would add a piece of wood before I cut a dado in the rail. I would want to make sure the problem was fixed before I made a permanent modification. You can just tack it in with a nailgun, or some screws, and take it off again if it isn’t straight or something. I would also prefer if it was made from some engineering wood, like plywood, but I am overly paranoid about wood movement.

If everyone that built one has posted here, there are less than a dozen. But I’m sure there are some that don’t post. Ryan would know how many kits he’s sold, but not how many people built it themselves or people that bought the kit, and haven’t finished it. I would guess there are probably 50x more MPCNCs than Low Riders.

For reference, my machine is 48" wide, but I only have 36" or so of the table. So my machine rolls front to back and rides the pipes left to right. My problem wasn’t so much that the gantry wasn’t centered left to right, but when it moved away from me, it would also move more to the right, until the Z pipes would hit on the left side of my machine. When it would come forward, it would move left, until the right side pipes were touching. But I think there was some hysteresis, because it would always end all the way left. I think the drag in one direction was better than the other. It was kind of like parallel parking, scooting over to one side by driving forward, then backward left, then forward, then backward left.

Sounds like Jeffeb3 is right.

What are your current rapids and feedrates set at, you should have Z rapid, and Z plunge, XY rapid and feerate. If you are still using fusion360 please use estlcam for testing, I can’t diagnose the million issues that can come from fusion.

Just want to be sure this is what you have going on.

1-To start a cut have the whole machine starting at the stop blocks at the end or otherwise squared.

2-Power all three axis by manually moving them with the LCD. I usually lift it to somewhere near the cutting surface, drive my yaxis to the corner (my table axis) then drive my X axis(my rail axis) to the starting point.

3-Then make sure to drive the Y axis (my table axis) further than your whole job will be. This is important, this ensures you will not rub or bind and if you did not start square it will show here.

4- Drive it back to the start, set your z height and start your job.

You should never move it by hand, unless it is powered down and you SLOWLY drag both sides back to the starting block/table end.

I use mine (large enough for a full 4x8 sheet plus some) all the time. I slap 1/4" MDF sheets on it randomly and cut lowrider parts at full depth at about 8mm/s. For now I would not exceed that speed, more on that later. This has a lot of moving mass you have to move it slow but you can compensate by cutting deeper.

If your side plates were not perpendicular to the bed surface you would find Jeffeb3’s problem where you cut was nice on one side and stair stepped on the other along the rail axis.

The only way to get star steps along the table axis is skipped steps. From your rails dragging too much or your tension wheels pinching too hard.

Belt issues show as ovals shorter along the table axis, or bit lift marks again only on the table axis faces.

 

1 Like

Try this gcode, it is for a 1/8" bit and a 10mm depth, the outside is 100mm the hole is 20mm diameter.

Run this on that foam.

test.gcode (56.4 KB)

If you tighten the bolt on the “backbone” of each of those rollers, it should deform a bit, and pull the side bearings into the pipe. Don’t overtighten (sorry, it has to be said). The other thing that might be happening is that the screws holding the rollers to the 611 plate might be holding the roller open. I’d make sure you loosen those screws a little before tightening the backbone. You might even have the holes a little off compared to your print, and you could widen them a little.
I'll try these things, thanks. I got impatient and bought all the parts (printed and flat) from Ryan's shop.
My low rider or my tracks?
I had meant the tracks, but both answers were helpful :)
What are your current rapids and feedrates set at, you should have Z rapid, and Z plunge, XY rapid and feerate. If you are still using fusion360 please use estlcam for testing, I can’t diagnose the million issues that can come from fusion.
I lowered the Z feedrates to 8, per your suggestion (50mm/min), but left the 333's. I will try Estlcam again, but I had some issues with the resulting GCode, and I didn't want to add that confusion to the mix. I followed your text guide + OldGuyCoding's video of the same, but the GCode just slowly raised Z (like <1mm/s) and it just kept going up, so I stopped it and switched back to F360. I'll work on Estlcam more soon.

My workflow is more abbreviated than yours: I turn the machine on, wherever the tool happens to be, then I use the steppers to position the bit on the work surface, and then start the program. I haven’t been taking the machine to the edges of the work-space. I’ll try that too.

For the tracks, I’m also thinking about buying some C-Channel aluminum, potentially with some printed parts to keep it straight and fastened. Like this, though that might not be the right size (just for illustration purposes).

I’ll load up this circle GCode and update y’all.

Okay, so I didn’t run the circle GCode yet, because I wanted to get a better sense of what was happening. Now I’m 99% Convinced that tracks would fix the issue. Here’s what I’m seeing.

Terminology, just to make it more clear:

  • Near side - The part of the assembly on which the RAMPS board and controller is on
  • Far side - The opposite of the Near side
  • Left side / Right side - Moving the gantry as far left/right as possible, when facing the Near side. The skateboard wheels are rolling when moving left/right
  • Rail - 1" OD Stainless tubing (as opposed to "track", the guide that Jeff has)
Experiment done:
  • Using the controller, move the assembly from the left side to the right side, across the whole table
    • I found that the far side assembly would get stuck at various points during movement. This would cause a counter-clockwise movement when moving left->right. This makes sense, as the steps in the cuts in the skull were most pronounced farther away from the stuck assembly.
    • Investigating the far side found the Z-Rails rubbing against the table
    • When moving left to right, the right-side wheels had a inclination to move inland, towards the table
    • Physically correcting the direction of the wheels on the far side, through a slight lift-and-twist, eliminated the problem until later in the left->right movement
    • Conclusion: The wheels on the far side are not straight, leading to regular jams against the table.
  • Confirming the Z-height is level on both sides, raise the carriage, then power down the steppers (with a roll of painters tape to prevent the assembly from slamming the table)
    • I found the far side assembly to be lagging behind the near-side when lowering, and once the carriage hit the roll of tape, it stopped moving entirely
    • Upon inspection, raising the far side assembly from the bottom caused the screw to continue turning
    • Conclusion: the far side assembly is not perpendicular to the table top, causing the screw to be angled, causing more friction, stopping the movement.
 

Doing a HD run and I’ll keep this log updated.

Ok, well the first thing you should do is make sure your gantry is square before powering up the motors. That’s why I go to the end of my table. You want to start the motors where the tubes are as perpendicular to the path of the wheels as possible. That will make things a lot better. I know my torsion box is square, and wheels ride on the torsion box, so I line the carriages up to the front edge of that box. You’ve got an 8’+ table, so you can use a tape measure. It doesn’t have to be perfect.

The Z might be tighter if the z pipes are rubbing. That might also push the z out of square. The down force is also proportional to the distance to the weight, so if the carriage was closer, the close side would have more down force on it.

I think I’m getting there:

I installed some track using some Aluminum C channel, like what I posted earlier. For now I’ve just installed 4’ of it, to see how well it would work. I was able to move the gantry from farthest-left to the end of the rail with basically no disagreement by the end of it.

I also tried loosening the cross-rail and moving the assemblies to be more favorable to going straight, then re-tightening.

I’m playing D&D until 6pm, after that I’ll run the circle GCode, and then maybe the skull again if it looks good.

50% Confused, 50% frustrated…

After installing the rails, the stepping and scalloping both seem to be gone. Awesome! Except, the test.gcode file didn’t go as expected:

The dimensions are all good, and the lines are clean. Except, WTF happened with that leg?

Other things I noticed:

  • With the rails installed, the far side now seems to move a little better than the near side, but it didn't seem to affect the cut. When the job was done, the gantry was not back at the edge like they started, but the far side was closer.
  • The Z height also is still disagreeing over time. The far side's lead screw higher than the near side at the end of the job.
More photos of the test gcode are in the album.

Start it at the end of the table so it is square.

 

Drive it longer that the cut so you know it won’t stick.

 

You can’t just power it up and expect it to be perfect.

 

Try it our way, before you start taking short cuts. Once you get a good cut do it your own way.

 

As for your z issues, did you buy it from me, what steppers, are they wired in series or parallel, lead screws or all thread? Or it just could be that you didn’t start square, you didn’t drive past your entire cut, so it is crooked and wedged the z axis against the table.

Start it at the end of the table so it is square.
I am
 

Drive it longer that the cut so you know it won;t stick.


I am not yet

You can just power it up and expect it to be perfect.
I don't, that's why I'm here
Try it our way, before you start taking short cuts. Once you get a good cut do it your own way.
I am trying as many of your suggestions as I can at once. I really hope I'm not trying your patience, because I really do appreciate the help you're giving so far, and I'm incorporating them as quickly as I can.
As for your z issues, did you buy it from me, what steppers, are they wired in series or parallel, lead screws or all thread? Or it just could be that you didn’t start square, you didn’t drive past your entire cut, so it is crooked and wedged the z axis against the table.
Every part that I could buy from you was purchased from you. I ensured that the Z height was square using my calipers, and that the gantry was square by pulling it to the far end. I did not, however, try running the machine past the job area to watch for snags. I will do that again and get back to you.

I’m not trying to be a punk, that is how I/we start and make every cut, for a reason.

The numbered steps I laid out above are how I use my machine. It is not meant to be a try this or that list, it is a do all of it list. At least until we get you sorted out. The test gcode choked at the beginning and finished fine, suggesting if you had driven the gantry the length of the cut before starting it you would have not had the pinch and the cut would have been fine. What I am getting at is I think your machine is fine, you might not have even need the rails added you just might have needed to drive the length of the cut first to make sure there were no snags.

 

I don’t care where you bought your stuff, I am trying to find out why you are having issues and can’t make assumptions as to your equipment. If you bought it from me I know what you have and don’t need to ask so many questions.

Lead screws or all thread?

What size steppers?

What drivers?

What voltage are your drivers set at?

Are your steppers wired in series or parralel?

How many teeth on your pulleys?