On achieving a perfect level

It looks like it is truncating the lower values. You can’t even see 1/3 of the chart. It would be more useful to have a flatter chart which is what you would get with a different z scale.

Cant verify at the moment, will have a look at it, but my digitalprobe shows nearly the same values.

It’s not the values, it’s the scale of the z axis. Look at the two graphs just up from yours in this thread. They are both the same data, just different z scales. Look at the range of z values on those two graphs. Obviously if the z was at the same scale as your x & y you would’t even see the warpage.

2 Likes

@stefix based on your graph the maximum difference between the highest and lowest point in your bed would be 0.1mm (range -0.02 and +0.8). I think most of us would just call that flat and be done with it :wink:
I think that’s what @Tone is trying to say.


You change the axis values in the plugin settings with the Graph Z-limits value. “minvalue,maxvalue”
By default it’s using “-2,2”.

(Before I misunderstood that setting and had just one large value in there, I think you might have it configured that way too; without the relative Z-offset values you’ll need to set absolute values in there I think)

1 Like

AAAAAAHHHHHH! yet I understand what @Tone had tried to explain to me!
Had better took of my sunglasses :sunglasses:
When back to 3D I will have a look at it.
Tanks for pointing this out.

1 Like

It’s not great, you have about 0.75mm difference between the highest and lowest point. However, there seems to be a nice relatively straight slope from one corner to the other. This means the bed is probably relatively flat, but not parallel to the x-axis. The good news it’s not full of asymmetric bumps etc.

Try to tram the x-axis to the bed to get both parallel and I think your results might already improve quite a lot. (Carriage Tolerances - Unusable Over Distance >75mm From Center - #33 by pta00066)

2 Likes

@Tone @brvdboss
Hi I did what I have been told. Here are two graphs.
with Z Limits 1,-1 and 0,25 and -0,25.

Is it now useful? I have done my best :wink:

Forgot to say its at roomtemp

1 Like

Much better, don’t you think?
The biggest problem with your other graph is that it clipped some of the data. You couldn’t see just how far down the z went on the near Y side. The clipping was starting at about Y175. I agree with @brvdboss that I don’t think your too far off. The leveling software can probably deal with that surface without problem.

Let me mention a couple of points here, you can perfectly level your bed physically (by shimming) and then the leveling software doesn’t have to work too hard if at all. You could effectively turn off the leveling function. But, if your leveling matrix is accurate and your bed isn’t too lumpy, then you will get good results. This is the real question, how accurate is your leveling matrix? If you have just one single point off it disturbs the whole region and probably affects clear up to two grid points away in all directions. The leveling software is creating a smooth surface over all the grid points and it doesn’t take too much to affect a whole region. A couple or three points off and then it starts to get bad. Because play, backlash, and heat affect the accuracy of measuring each grid point, it’s hard to be accurate. That’s why I developed the “Check Level” gcode files.

1 Like

Super late to this party, But I had a random question regarding the redesign you talked about for your dial.

I can’t say for sure that ALL do, but don’t they typically function with the use of at least 1 spring? I would think that it’s not a great idea to compress the spring for hours while you’re printing, all the while, being subjected to high temperatures.

Just a thought. Thanks for all your efforts! You are a gem in this community!

Forget the Prusa I have one too and the procedure is totally different.
With the Prusa, after you set the approximate Pinda height, you run a test print and adjust the Z on the fly until the squishiness is just right.
With the Snapmaker, it does the auto levelling (not before each print like the Prusa does though) and the last step is to use the card provided to fine tune the reference point.
Hope that helps.

1 Like

For the manual 11 x 11, since the heating stops during the process, should I heat the bed up to say 80 celsius rather than the usual 60, so that as it cools it sort of averages out at 60 halfway through the process?

Does that make any sense?

1 Like

That’s what I do. Also keep it in an enclosure so it’s able to maintain temp a bit longer

2 Likes

So I’ve read the first few pages. And orders a dial.
Have not had issues yet. But I want to test it. And might use it for other QC. GDNT stuff.

So I had a random idea here. To plot it anyway.

I know the igaging brand of calipers and test equipment have a BT dongle. That inputs the selected measurements, in to a cell in excel.
Click next one click next one etc.

Thinking if I get a dial compatable.
And the bt dongle. I could run the bed leveling sequence. Mapp all the data to the Excel sheet.
And then use that data to generate a 3d graph. The graph should be level in theory. But it would be a good visual representation of what your talking about with the bed warp.

Also to overlays this with the thermal imaging camera image.

I wonder if there is any correlation to heat up areas and the bet warpage?

Thank you for the dual dealios.
I might try the map myself if I roll for initiative jk

Second thought to add.

And I wonder how many have done this without thinking

Every time I’ve assembled the heat bed to the platform.

I barely finger thread all the screws. And then I do like a bolt pattern. Start in the center and work out from there. I don’t ever actually torque any of them down snug tight. Just enough so the hub centric counter sunk holes and the counter sunk bolts meet up right.

It’s hard not to torque them down. But it’s self restraint.

It barley needs to hold. The # of them and the hub centric nature. The sheer strength is there. And it’s not shifting. Lol.

I wonder if most have messed up there bed in first assembly and caused some warpage.

So like. With aluminum. Steel. Etc sheeting. Welding whatever.
If you tack all your parts together too fast and heat it up. The metal expands and contract’s. If you tightened from left to right. And first half tighter than the other half. You would have to expect some material change. Tensions etc. That’s why there are so many bolts. So close.
But still.

So I tighten from center few bolts. Diagonal to diagonal. Across across. Etc. To tighten in thought of this. But at that point. The bed is snug. Centered. And all the bolts are finger tight almost. Like I could shift it if I tried. But not much in it’s own kinda tight.
And let the counter sink hub centric thing do the rest. .

I wonder how many beds are over torqued. And then heat cycled a few times in the unknown.
And the star to get warpage and defects.

And . Idk. If they incorporated a dial in to the mount plate for the tool head. Could be used for any tool. Depth gauge etc.

And it would be super simple for them to map the data to the display grid and show you where there is warpage etc. Or just show you and ask if a +/- % tolerance is ok. Or re level. If same results.
Then re try the fastening pattern.

Idk I have the Kickstarter a350. Original bed. Platform etc.
Throug I’ve been debating a mirror top.

1 Like

If you are using octoprint, that’s exactly what the bed visualizer plugin shows. Also, the gcode “M420 V” gives you the current bed leveling status as measured with the existing sensor. (so inaccuracies of that sensor may apply, just like with a dial indicator). You can plot that one in excel as well.

And in this topic there are already spreadsheets floating around to do that visualization as well.

I also created a piece of sample code to read out the digital dial indicator I have which is the typical cheap chinese brand. (Arduino code to read out digital dial indicator with data output port · GitHub). The serial cables that you can purchase as well cost a lot of money and offer very little functionality. My initial idea was to create some arduino or esp32 code that also connects to the SM2-usb port and sends gcode, does the probe on different points and automate what you want. But it’s one of those many things to do, that aren’t my priority.

me over here torquing everything as hard as my weak programmer fingers can, as nothing tells you not to, while seeming reasonable given the full metal build of the machine.

i been torquing them hard. and i do a radial pattern similar to lug nuts on a car wheel. results seem to be my corners/edges are my bad zones.

when i wasn’t driving them as hard as i can, after a few prints, at least one of the 9000 screws on the bed will come loose enough to rattle, causing the machine to make screaming noises.

i now though wonder if i loosen up the edges if they’ll chill out. keep my center screws tight but let the outers move a bit. already doing the thing where you let the bed warm up first, but, like, the moment it hits 60 i was screwing it back down, not waiting the 30 minutes. the primary reason i did this was it also made the machine more quiet.

following up because this seemed stupid successful for me.

  1. loosened all the heat element screws
  2. cooked my bed to 60C, and then forgot so by the time i remembered it was super done thermal expanding.
  3. lightly tightened the screws just like “ye ok that seems to have friction” and no more. kept the pattern of doing inner-cross-corner then outer cross corner just like it was a car wheel.
  4. recalibrated with the 5x5
  5. re-drew a giant square around the entire platter.

my back right corner is still the worst, higher than the rest, but its no longer so high that its choking the nozzle causing clumping as it leaves that corner again. and over all, all four corners and edges seem to be good enough that a 0.2 initial layer is allowing the leveling mesh to iron out the rest.

going to do my best to just make it a habit of letting the machine fully cook itself before actually committing a job to it.

1 Like

I will definitely need to try this as I can still only print in the centre of my bed. Thanks for sharing. One thing Snapmaker could do for us is to build in the 11 x 11 point calibration but with the bed being kept at a constant 60 degrees.

update your firmware, it has this now. when it asks choose 60, and then just forget about it for 30 minutes. then come back and hit calibrate.

I don’t think the built in machine calibration goes up to 11x11, I think you need to run gcode for that. The previous releases kept the heater on if using the touchscreen to calibrate but not if you send a gcode command. This has been fixed in GitHub and will most likely be in the next firmware update.