Yeah may have some of the same issues i am seeing. I honestly haven’t printed a thin wall version yet So ill put that on tonight’s list.
X-Y calibration, if you are honestly seeing only .3mm error over 140mm movement then that’s not your issue. Though if that .3mm error is all from a smaller part of the acme screw then it could be depending on where on the plate you are printing. But that .3mm over 140mm for a pre x-y step calibration is really good. So unless there is some error in how you are getting your data, that isn’t the issue.
I personally during all of this this have my z seam set to random. But it doesn’t seam to make a difference that i can remember. I do think the flat that i am measuring and seeing is the same as you said. A expansion in the offset of the wall from one wall to 2 walls.
One thing to try would be trying changing from outside wall first to inside wall first or opposite of what you have right now and see if that changes anything.
I printed a 2 wall thin cylinder. The wall thickness goes from .83mm +/-.01 all the way around the edge until it hits that common location where it bumps out to about 1.1mm
Is this a slicer issue after all?
Or Another theory im trying to think out is. Is it a coasting or k-factor issue at the end of each line?
So test plan.
Change to outer wall first.
Try spiralized.
K-factor / liner advance calibration. Is this enabled on snapmaker?
But if it was anything to do with coasting, or retraction, or anything to do with the z seam, then why is it creating the same flat/bulge opposite the seam as well?
I’m going to carry on with my tests, which have various settings including random seam and a vase mode, lined up to print. But I’m not going to be surprised if they all come out the same (except maybe the vase mode).
Are you using cura as well? Is this just an error in the way cura is creating the circles?
I’ve done 8 samples, all of which have come out exactly the same, with the flats on the sides (these flats run parallel with the y axis) and the gaps between the walls just above (on the left) and below (on the right) the flats. The only difference is on the vase mode, and with the z seam set to random, you don’t get the actual seam line, but they still have the same shape.
Yes, K-factor/linear advance can be adjusted on the Snapmaker…uses the “M900 k#.##” G-code command. I have found the default value to be VERY close to the calibrated one (for the SE). Default should be 0.04, and mine calibrated to 0.05 with no noticeable aesthetic change in print quality…much older versions of the firmware had the default at 0.12.
Has anyone here tried a test a vase mode? I did a search of this thread and the word vase is not found in the entire thread. Vase mode does not have a seem, thus it is the only valid test for roundness although backlash will still have an effect. I do find it interesting that people here are expecting a perfectly round circle from an FDM printer. The physics behind FDM make that impossible for anything but vase mode, so no one ever tries. If you need a perfectly circular hole, slightly undersize the hole and then drill it out like everyone else does. Is there something that I’m missing here?
@CNC-Maker Do you mean “FDM” printing? Never heard of FTM, though I am still a relative newcomer to this universe, so maybe it is just something I have missed.
Vase mode was mentioned by @davidE1A a few posts back; apparently has the same outcome.
Whether it should be expected or not, there is certainly nothing wrong with noticing a flaw and striving for solutions through experimenting with processes and settings to get the results it as close as possible to what is desired. It does seem weird that the flat section of the prints are only on a certain sides, and are parallel to a specific axis, confirmed as repeatable on 2 different machines.
@Spaced I haven’t noticed it yet, but I will test it today or tomorrow. Finally finished up my extensive bed level testing, so I can finally get back to printing.
vase mode really has no real world use besides vases or decretive items. much more to the world of FDM printing then just vase mode.
Why is vase mode different from the hardware point of view than regular FDM printing? It’s still a x-y coordinate system in the g-code. nothing that is Binary 0’s-1’s in computer code can ever be perfectly round, true. But the resolution of the model is so much higher then the print to make that a mute point.
My printer theoretically (haven’t tested it) would show this same issue with the laser and the cnc. I might have to go down the road of the laser to burn a circle to test, but ill have to think about that. since i am not currently environmentally setup of laser until summer.
Your also missing trying for a level of quality that the system should be able to provide mechanically. You are missing that basic tolerance tests are failing at very large gaps. Much more then are advertised.
Its not backlash. (I personally have backlash numbers that are < +/-.01mm more like +/-.005mm as stated earlier in the conversation.
Its not XYZ calibration. This has been covered I believe.
X-axis is tuned down to +/-.06mm locational tolerance. but repeatable. on my first attempt more this weekend.
Y-axis is tuned down to +/-.03mm locational tolerance and repeatable. on my first attempt more this weekend.
Z-axis is tuned down to +/-.015mm location tolerance and is repeatable. on my first attempt more this weekend.
Its leaning towards a slicer maybe. I need to set up prusa and tune it. but then do the same test. unless someone wants to send me g-code to test from there prusa slicer?
@Mxbrnr you are correct about the acronym, and I’m still trying to figure out why autocorrection keeps changing it.
@Spaced there is a lot of physics that goes into the mechanics behind FDM printing, and to put it simply, there are pressure changes in the molten material that occur with each revolution of a circle, primarily at the beginning and the end. Because of those changes alone, it is not possible to create a “perfect” circle. In vase mode, those changes do not occur at all, thus the reason that there are no “bulges”. However, backlash can still have a hand, and turn a circle into more of an ellipse or an oval. There are many other factors as well, but those are the easiest to convey.
Repeatability and accuracy of <= 0.15mm should be obtainable by any good printer. In fact, no tolerance test goes beyond 0.15mm, as there aren’t any known printers that can attain a better tolerance of like 0.10mm.
I respect your thoughts but have allot more experience in this area then i let on. I would need to see evidence that this is a pressure issue that only shows up in circle arc profiles. Maybe print 4 quadrants separately with defined start stop points?
the issue seams to show up in the upper right quadrant and lower left guardant from the central point of the sheet for me.
I agree good printers should be <.15mm tolerance test. mine isn’t in a circle but passes it with blazing colors in a cube.
Edit for clarification
Also for everyone’s clarification i have only been doing all my print tests in the middle 50mmx50mm square section on the A350 Mainly to minimize the other variables i haven’t approached or wanted to attack yet. Mainly the bed lift issue when the y axis changes direction.
The X axis and the Y axis are at or near the mins, correct? If so, what you are describing is backlash, which is the affect that I mentioned. It is very possible that your Linear Modules have either worn down a bit, or were not properly adjusted. There is an anti-backlash screw in each Linear Module that can be adjusted to reduce the backlash to almost 0 (zero).
However, you need to be very careful as many people have completely trashed their Linear Modules simply by opening them. Also, if you over adjust the screw (too tight) it will cause the nylon carriage to wear down even faster. The should be a document written by Snapmaker on how to do it. I think it was posted about 2+ years ago on their support pages.
There is also a backlash setting in Marlin that Snapmaker added a couple years back, which you can also use. Some people liked it, while others didn’t. It’s a non-invasive option that may, or may not, work for you.
Im open to the help but but insisting it is something that we have tested. Confirmed and retested as a group isn’t helping. You have brought no new info to the table.
You said to print in vase mode i did. Same issue.
Using a dial indicator with 0.01mm resolution measures back to 0.00mm 75% of the time and +/-0.01mm 25% of time no matter the amount of y movement (10mm, 20mm, 30mm, 40mm, etc), or in a separate test the same for the x movement. using a machinist magnetic base and a steel plate bolted to the snap maker base.
So cnc-maker how dose one find backlash with that data? what’s missing? I’m not being sarcastic I’m being open, what am i missing?
I agree my x and y calibrations are still not perfect. But they are not in the magnitude of .25mm deviation of the dia of a round object. Since both my x and y are sub 1.2% error over 20mm.
I honestly paused and had a think after that last sentence. Is 1.2% enough to cause this out of round. I would think technically yes. But that doesn’t account for davidEa1’s photos that show really well what I am seeing also. So im torn from a software firmware issue or hardware. I have to think with the dialing in and testing i and other have done that it is leaning towards the former. But at the same time other people seam to be able to print something like clearence/tolerance test by Jace_fireborn - Thingiverse
and i cant. I have no elephants foot to speak of. the pla i am using is dimensionally accurate in a 20mm calibration cube (+/-.05mm overall).
So im at a loss. Am i asking to much of a snapmaker? maybe. But i think it has the bones to be a decent printer. I have already done medical prototyping with it for my job and it works fine. I decided to push it and dive down the rabbit hole of precision and here i am.
So my single extruder prints a 20mm x 20mm cylinder showing a diameter variance of between 19.74mm to 19.92, widest at the z-seam; no calibrations of any sort done with the machine at this point (except bed level). So even uncalibrated I am getting the same 1.3% error you are. Slicer settings were 0.4mm nozzle, 0.2mm layer height, 2 walls at 0.4mm/wall, outer wall before inner wall, and walls before infill.
That linked post is definitely worth a shot. I’ll keep working to get my printer calibrations completed, then test again and see if anything changes.
Yeah in reality .18mm delta is not allot. Did you notice a flat spot like davidE1a and my self show?
I mean if it was round and oval but smooth then my mind goes to like cnc-maker said back lash. But since there is such and abrupt change (on mine the max and min are only like 2-4 degs rotational apart) i dont think its backlash. But im reading up and doing more tests to see if it is.
but then do the same test. unless someone wants to send me g-code to test from there prusa slicer?