SHapeoko 3 XXL vs Shapeoko 5 Pro accuracy

CNC Machines
41

My apologies are due to you Jon, because I cannot start my analysis from an irrational position. If there was “NOTHING different” between all of the attempts that were both successful and unsuccessful, the results should not have varied by the slightest amount. Any significant variation from your desired datum demonstrates that CNC machining is a useless pursuit. Where you cannot rely on the final result, the implication is that something in the continuous chain of factors which MUST BE PRESENT for successful CNC machining… was not present.

Logically speaking, you will only discover that crucial factor by painstakingly removing one of the factors in your system and looking at the results obtained. Replace that factor and remove another one, then you can examine the results… ad nauseam. Something MUST be a causative factor and from this distance, I could not begin to guess at what that cause is. If you generated some log data, that would really help to pin down the cause.

Mechanical things do not usually give random results unless there is a fault in the driven elements, or one of the driven elements is poorly attached. Electrical things do not usually fail intermittently and cause interrupted results. They either fail and are non-functional or they work. Software can and does fail intermittently. It may have been tested insufficiently. A specific variable has not been considered and accounted for. This leads to unpredictable behaviour and unwanted results. It may even be very poorly written. Software can also be used in an unintended manner (by the user}. You may have to search the Carbide 3D forum archives to find a similar failure to the ones you have described and illustrated.

42

I agree that something is changing, but I am unable to discern what it is. I had hoped that I would get some insight. The comments about connectors seems promising.

43

Apologies if you stated this above, but, have you tried returning to X,Y zero after each toolpath to see if the machine has lost steps during the run? This would be before re-initialising the machine.

Also, with the machine powered off, slowly move both axes their full extents and watch carefully to see if the drag chains have moved to a position where they might obstruct movement, I’ve lost steps that way in the past.

As you suspect, you should be able to do much better on a 3XXL, I get better than 0.1mm on Aluminium parts.

Also, do you have the old mechanical home switches or the later inductive switches? On the old mechanical I’ve had chips, wires and other crud get in the way and obstruct the zero position.

1 Like
44

Troubleshooting is pretty straight-forward if one breaks things down sequentially and looks at it from beginning to end:

  • CAM software produces G-code which Carbide Motion sends to the machine — may be verified by using a G-code previewer such as the wireframe preview in Carbide Motion
  • Grbl determines the current which needs to be produced by the stepper driver chips and sent to the stepper motor(s)
  • the current is conducted to the stepper motors by the motor wiring (and wiring extensions on larger machines) — verify all connectors and conductors/wiring are in good condition and secure — note that the connectors should be placed so that the wiring leading into and away from them will not shift as the machine moves
  • the stepper motors then rotate based on the current sent to them, turning the motor shafts which in turn turn either a pulley or screw — ensure that the set screws in the pulley are securely in place and the pulley does not shift on the shaft — put a witness mark across the end of the motor shaft/pulley
  • the turning pulley/screw will then cause the machine to move by exerting force on the belts/nut — the belts will need to be tensioned appropriately per the instructions and the link above
  • the machine motion will be constrained by the Delrin V wheels on the aluminum rails for X and Y, and the linear rails and blocks for Z — ensure that they are well-adjusted/lubricated

The machine movement of course will require that feeds and speeds are appropriate to the materials being cut and the tooling being used and the toolpaths used for cutting.

1 Like
45

Liam: thanks for your thoughts.

If I can remove the stock, I can visually verify whether the end mill lines up with the starting position, but that is only the case after the finishing passes. And I doubt that I can see 0.1", so not all that helpful. I thought about drilling a hole at zero and then trying to drop a center finder in, but have not done that yet. Is there another technique that I should consider.

Suppose that I determine that the zero is off. Now what? I already suspect that. How is this information actionable?

I will move the axes by hand as you suggested.

I believe the switches are inductive, but I would need pictures to be sure.

46

Hey,

These are the inductive switches

They’re somewhat more repeatable than the mechanical ones, although the mechanical are Omron I still find that I can get 0.1-0.2mm error on initialisation sometimes.

I use an MDF fixture for complex parts on top of the spoilboard. Use dowel pins or similar to align it repeatably on the machine, then machine front and left edges to give a repeatable zero point, then zero from that and bore out some dowel holes for the workpiece. This way I’ve got an accessible zero at all times to re-zero or check.

To actually set zero, I have the bit-zero which is OK for most work, but for verification I find it’s useful to put a chamfer bit in and use my phone as a magnifier to really see where the tip is at

https://www.amazon.co.uk/Trend-Tungsten-Carbide-Chamfer-Groove/dp/B0001P11H0/

What you could also do is bore a 0.25" hole in the fixture plate at a known X, Y location, then manually jog to that location and jog down in Z to see if the cutter is lining up cleanly. I do this occasionally with a dowel pin to ensure I’ve got everything where it ought to be.

If you find the machine is losing zero during the cut then it’s time to walk through the ‘where and when am I losing steps?’ elimination process. Will has already described quite a few of the key steps but something like

  • Check the X and Y axes roll freely, the V wheels are clean and have proper preload
  • Check the belts and belt anchors are in good condition, ensure the V wheels don’t clip the belt anchors at the ends of travel
  • Check again to make sure you don’t have something hanging up on a drag chain or some other embarassing oversight, done that
  • Use a marker to mark across the end of the pulley and stepper shaft so that you can visually assert the pulley has not slipped and rotated
  • Use the rapid positions to move the machine around, listen carefully to the steppers, if the sound is not continuous then that’s a warning about wiring
  • I’d probably check the spring tension on the socket side of the motor connections as well as looking for any signs of darkening or contact wear on the pins

HTH

2 Likes
47

The short version: everything is working perfectly.

The long version:

First off, thanks to all of you who have tried to help me. This community is wonderful, and I am especially grateful to the Carbide3D employees who supported me while using an out-of-date machine. When I need to upgrade, I will turn to Carbide3D first. This kind of support is important to me, and deserves to be rewarded.

I am the conceptual designer for this project, but my friend, Brian, translated my concepts into CAD in OnShape, generated the STL files used for the 3D prints, and generated the STL files used as input to MeshCAM to create the NC files required by Carbide Motion. He created the various CNC jobs and we cooperated on the spreadsheet that I used to coordinate and run the jobs.

We discovered that I was using a Nomad (Carbide3D) 3 flute end mill, while he had specified a 2 flute end mill in MeshCAM. I then switched to a 2 flute end mill, so that the feeds and speeds matched better.

I reported that I had created one perfect part in spruce, but then when I tried to create another part in maple, it failed. It turned out that Brian had accidentally generated a MeshCAM job specifying a 1/8" end mill but had told me (via the spreadsheet) to use a 1/4" end mill for one of the 4 jobs. The result was a part with a visible lip, which I then interpreted as the machine losing steps.

Once we used the right end mill for the job, all three parts came out flawlessly. The machine did not need to be dis and re assembled, or even tweaked: it was already fine (after tightening the set screws). I did a number of the checks recommended above, none of which revealed problems.

It is frustrating that the NC files contain comments specifying the size of the end mill to use, but that Carbide Motion does not share that information with the user. Some of this could have been detected or prevented earlier had the details of the job been more visible when it was loaded.

We are now moving onto other 2 sided jobs, with increased confidence.

Thanks, again

Jon

3 Likes
48

The new beta of Carbide Motion:

https://carbide3d.com/carbidemotion/beta

adds a feature working toward that:

Added code to read a tool table from the G-code.

1 Like
49

As it usually is, the fix was apparently simple and we all over-thought it. That’s a classic move for me. :smiley:

Glad you got it sorted.

50

Thank you, Will: I will check out the new Carbide Motion. And thank you for your help.

Jon

1 Like
51

Glad you got it sorted.

If your debugging experience was anything like mine you’ve probably learned quite a bit about the machine and got more confidence in how to use it, service it and program it too.

1 Like
52

I did learn a lot! Whether it was worth the torn hair is another matter!

2 Likes
53

This topic was automatically closed after 30 days. New replies are no longer allowed.