Carriage Tolerances - Unusable Over Distance >75mm From Center

[brent113]

After endless fighting of bed levelling during 3D printing, and now endlessly unable to get the laser to focus, I took a straight edge to the carriage and estimated a 1.25mm drop. The ruler is in Freedom Units™, don’t mind. I used F360 to measure it:

The 4 inline machined flats that run along the Y axis makes the following shape (exaggerated here), and is easy to see and feel as you roll the straightedge along the pairs of points:

Additionally, some of the machined flats look like they were machined at an angle. When I leave these loose all of the laser trays are in line. But as soon as I tighten it the slightly, finger tight on just the round shaft of the tool, this happens:

For grins, I took a video of a dial indicator tracing the flat in a circle, and there’s 0.003" deviation (0.08mm) over the 8mm diameter. That explains why tightening down the screw cants it to the side.

New video by Brent Scrivner

Basically, the tolerances on this carriage render every tool nearly unusable. For instance, with a >1mm variation, there’s no chance the laser stays in focus as it moves over material.

I measured dozens of triplets of screw holes, and nearly all of them are out.

The measurements were pretty straightforward:

{EDIT: Removed dial indicator measurements as they are being skewed by a non-perpendicular X axis mount, detracting from the carriage deviation point I’m trying to make. A corrected measurement is in a comment below.}

There are additional measurements that could be made, but I believe this paints enough of a picture that further measurements are redundant.

I have reached out to support.

[Atom]

I love your thoroughness… I haven’t checked mine… I’m too afraid of what I will find.

[brent113]

I just updated with a video of the flatness of the individual mounting pad…dude: https://photos.app.goo.gl/RBdsL857erAAuckbA

Over the width of 1 of the laser plates that will result in +/-.5mm due to the angle of the mounting flat.

[Atom]

Oh God… how do they sell these for so much!

[Atom]

I must be one of the lucky ones… seems like that is one of 3 parts on my unit that isn’t defective lol

[brent113]

Let’s combine forces - everyone can start a pool of working parts and, like the Power Rangers assembling into Megazord, we can assemble into a single functioning machine!

[Atom]

We will have so many extra psu… its the one part that works on every machine because its the one part that SM bought from a good company

[brent113]

I realized a measurement error and updated the mounting flats above - they are much much worse than I originally measured, with a maximum deviation across the corner of 1.4mm:
{EDIT: Removed dial indicator measurements as they are being skewed by a non-perpendicular X axis mount, detracting from the carriage deviation point I’m trying to make. A corrected measurement is in a comment below.}

[Snapmik]

It happens that I recently dismounted and mounted the bed again for other tests and now I find it difficult to cut large pieces with the laser. There is always an area where it doesn’t cut and this could perfectly be the answer.

Before this last dismount and mount it was working reasonably well; cutting 3mm plywood in 4 passes, but now I have to make 6 or even 8 passes (with the feeling that the laser in not well focused: and each time I calibrate it, the SM returns a different value between 23 and 25mm).

[ITmaze]

Some questions:

• Is the surface on which the printer stands flat?

• Is the printer base flat?

• Are the tops of the Y-axis flat?

• Are the tops of the Y-axis mounts flat?
  When I say flat, I’m referring to the fact that your straight edge shows no gaps at each above check point.

When I take delivery of mine, I’m intending to start at the bottom, working my way up, one level at a time to ensure that the individual levels are flat and at the same height as the level below it. If you have an indicator this will be much easier, but for the error you’re showing, I suspect that a second ruler will be enough to get you much closer.

(Note that technically the same should be true for the Z-axis and the X-axis, which should be 90 degrees offset, but you’re showing a problem of the bed, so I’m suggesting to start there.

The aim is to determine if anything (unexpected) has warped or is not installed correctly.

If I’m telling you how to suck eggs, I apologise.

O

[ITmaze]

Hi Brent,

I see you have an indicator That will help with my previous response

What you’re showing on the video could be caused by an error in Z-heights, combined with an error in Y-heights, made worse by an error in X-alignment. My point being that the error you’re showing could come from anywhere, including a poorly machined hole, a warped frame and other things.

The only way to figure out what it is is to do each dimension individually, starting at the bottom.

Over the last few months I’ve been looking at various approaches at tackling this and the best examples I’ve found come from the CNC community who are checking all manner of dimensions and errors. The 3D printing community appears to (by enlarge) think of the problem as “bed levelling”, but that’s not what’s actually happening.

So, in my opinion, you should start with the base on which the printer is standing, the table/desk/floor/cupboard and make sure that it’s flat (or as flat as you can get it) - it doesn’t have to be level as such - as-in, the front and the back, the left and the right, don’t have to be at the same height off the ground, as long as the base doesn’t have any dents or bumps. A straight edge should not show any gaps in any orientation. This will be your “reference surface”.

Then you put the printer base on that and ensure that it’s at the same height all the way around off the reference surface. You then expand that to include the tops of the linear actuators for the Y-axis and so-on.

Eventually you’ll get to the point where you know that the bed that you’re indicating off in your video is actually flat and level in relation to the reference surface.

Then you can check that the Z-axis is 90 degrees off the reference surface (in left-right and front-back orientation), then once you’ve done that, you can do the same for the X-axis.

o

[brent113]

Thanks Onno. I omitted some measurements in my previous post, I’ll try and fill in the missing gaps here.

The machine is sitting on a float glass table that is extremely flat.

I repeated some of the measurements I previously took with a precision 24" straightedge. The table, base, and all linear modules are extremely flat. While holding the straightedge in position on the modules there is virtually no light visible. I’d estimate the flatness is better than 3 thou. On the base casting it is also very flat, for an unmachined casting. There are a few small dips light is visible under, but it looks better than 6 thou. I’m not interested in getting my feeler gauges out for more precise measurements, especially difficult with the rough surface of the casting. Suffice to say, they are quite flat

Y axis mounting plates, from far to near, in line with the center of the carriage:
Y axis Left, Left side: 0.000"
Y axis Left, Right side: 0.008"
Y axis Right, Left side: 0.023"
Y axis Right, Right side: 0.034"

I also repeated the measurements directly on the Y linear module case extrusion, with identical results.

That could be X axis mounting being misaligned on the Z axes, so I disabled the steppers, and using the micrometer aligned the two towers so that the X axis is true to the Y axes. After that I was within 0.003" across the previous 4 measurements:
Y axis Left, Left side: 0.000"
Y axis Left, Right side: 0.002"
Y axis Right, Left side: 0.003"
Y axis Right, Right side: 0.002"

So now with this alignment, I redid the previous measurements.

Remeasuring the trueness of an individual flat is the same - 0.003" variance around the circumference.

Remeasuring all of the pads results in this planar deviation:

0.4mm deviation across the corner. Side to side is better, front to back is obviously unchanged. That may be an acceptable level however, particularly if the laser movements use the same bed levelling mesh (which they do).

The biggest question I have now is how to reliably align the X axis on the Z towers. I’ll try with 1-2-3 Blocks sitting on the y linear modules, up to the X axis.

Regarding the tilt in the mounted laser trays, I may have to attach them together somehow to create a flat surface, unless the carriage flats can be flattened or shimmed. I do have shimstock, perhaps I need to build up a layer of shim on the low side of the flat to flatten it.

[ITmaze]

Hi Brent,

Now we’re talking!

In your answer I did not see any discussion around the alignment of the Z-axis.

I’m planning to use a machinist square, but also found some interesting methods using dial indicators, for example mounting one on the X-axis and running it up and down along the length of the machinist square by raising and lowering it.

And then there’s the alignment of the tool mount and the tool itself.

That said, you seem to have flat and level under control.

Based on what you’ve shown, it looks like the mounting plate for the bed is actually bent. The plate itself - not sure if it’s cast or machined - could be the problem, or it could be the linear rail mount plate that’s not flat, bending the plate when you tighten it down.

As you say, shim stock could be a way to get the laser plates level on top of the mounting plate. Another might be to actually machine the plate, but I wouldn’t attempt that until I was certain that everything was trammed in. An intermediate step might be to super glue a 1mm aluminium sheet to the mounting plate and machine that as an overlay, like a custom machined layer between the mounting plate and the three work surfaces.

I’m suggesting super glue because a little heat will detach it. The holes can be oversized through holes, so you don’t need to tap them.

Would love to hear you thoughts.

O

[brent113]

The angles were talking about here are so small, you can’t accurately measure the Z towers with a machinist’s square. They are square enough, and have no adjustment. I focused measurements on things that are adjustable. Additionally, the Z towers could be rotated in 3 orientations, each - and all permutations of those within 1 degree or so will not affect the Z height measurements of the carriage if the X axis is trammed to the carriage, which it now is.

I did say the base, and many other things, are flat. I don’t agree with your conclusion the base is bent.

Decking the carriage is an option if I get support to send me a replacement first.

[ITmaze]

Sorry, just realised that I didn’t answer your X-axis question.

The 1-2-3 blocks would work, but unless your bed mounting plate is actually flat and level at a particular Y location, you’ll get incorrect results.

You could align the X-axis to your reference surface, but until you know that the Z-axis is actually vertical, that also won’t give you a real answer.

Come to think of it, if you have a dial indicator mount, you might be able to find a spot on the bed where left and right are at exactly the same height off the reference surface, you could mark those and use them as a reference for the X-axis.

[brent113]

Again, I wouldn’t reference off of the bed, it would be referenced to the Y linear module housings.

[ITmaze]

I don’t think the base is bent, I think the mounting plate is bent.

[brent113]

Are we talking about the same thing? Mounting plate = webbed undercarriage (‘platform’ in the manual)? Or the plate on each y axis module that the carriage screws to? I see - you said mounting plate for the bed - must be the ‘carriage’ I’ve been referring to it as, or platform in the manual. Yes, it likely was bent in shipping, and being cast metal I’m not going to attempt bending it back.

If it was the plates on the axes that webbing connects to you were referring to, which is a thought I didn’t think of until now, I’ve checked those and they are quite flat, but they could be twisted slightly. That probably doesn’t explain the measurements I took fully, but that would be something to look at for shimming, possibly.

Also, I went to the shop and grabbed my digital angle meter. On the close Z tower the 3 rotations that should be 90.0 degrees are 89.75 (rotate around X), 89.85 (around Y), and 90.35 (around Z). That’s extremely close. I checked the far tower as well and it’s measurements were nearly identical, within .1 degree of all measurements.

[FlyByWire]

Sorry for getting off-topic. After all the issues I have with bed leveling on 3D printing I was thinking of giving me a break on that and switch to the laser engraving tomorrow. Fortunately I saw this thread before, so I’ll quit this adventure.
As Laser and CNC should just be a nice add on, I don’t think I’m willing to invest another couple of week with getting frustrated.
I think this experience is coming to an end for me. Good thing is I learned that I could really have a lot of fun with a good working 3D printer.
So I will probably focus on that and buy a CoreXY printer, most likely a Vivedino Troodon.
I really admire those guys around here that have the knowledge and ambition to get the Snapmaker 2.0 working. I like getting thing to work too, but in the case of the Snapmaker I’m lacking light years of electrical and mechanical experience to move further on.

[Jesi15]

Shimmed the bolt surfaces on the carriage to make it closer to flat for now. Still hoping for some information from support regarding tolerances, @brent113 has a ticket in.

Lucky for us blue painters tape is 0.003 thick, significantly easier than cutting shim stock to test.

Compared to the original post above, this is significantly better.

Here is the initial test with the laser following the tape installation, results look promising.

The lower left is a bit off due to the paper lifting off the bed surface.