Hi, I made lots of tests with big parts and auto calibration of the A350.
I’m using PETG and Simplify3D slicer - I try it with 3, 4 and 5 points of auto calibration.
I have the newest hardware of the extruder head with an “… enhanced proximity sensor …” (words of Snapmaker company).
It seems, that in the middle of the heated bed everything is ok and the calibration is fine. But then a little outside there is a problem with the bed level. It seems to be closer to the nozzle.
(see picture)
The “white areas” shows, that the nozzle is to close to the bed. And the rest of the part is ok.
There is the same effect if the part is vertical. Like a “ring area” around the middle, where the bed leveling is not ok.
For me it is a hint, that the auto calibration software does not work propperly. It seems, that the measured array of test point are not correct calculated to a “compensation matrix”.
I have an Ultimater 3 too and it has a auto leveling too - but there is no problem. All the parts of the bed are working ok - no extrusion problems.
K.D.
Auto-leveling in software cannot correct a hardware platform that’s not flat. What auto-leveling does is to sample a number of points on the bed, average them, and create parameters that allow the print as a whole to be corrected as much as it’s possible to correct. The result gives minimal variation in thickness in a first layer; the top is flat and the bottom following inherent bed variation. What auto-leveling does not do is to attempt to track the variation in bed flatness so that the first layer is of constant height-as-extruded.
Now I made some more tests.
I found, that I have to lift the extruder about 0,15mm - 0,2mm in the “white areas”. (z-axis offset +0,15mm - +0,2 mm)
Now for me it’s now proofed, that the software calibration of the bed does not work correct.
The value is nearly equal if I use the auto calibration with 3, 4 or 5 points.
If I do this I have a perfect surface at the printed parts.
@eh9:
But what’s the sense of that “auto calibration” you discripted ? If it works so, then I do not need to measure different points - one in the middle would be enough.
This point would be the endpoint of the auto calibration routine, where you have to set the absolut level with the paper card.
The rest points are measured via an inductive proximity sensor - looking for the metal part of the bed not the plastic survace…
It is measuring relative values - no absolute values - but it gives you the difference between the points and so the “orientation” of the board.
(That’s my understanding of the calibration routine - I’m not sure if I’m right…)
Maybe a SnapMaker guy can give more and exact information.
At my Dremel 3D45 and my Ultimaker 3 the “orientation” of the bed is measured absolutely and the first layer is all time correct. So I thought, that auto calibration makes the same “job”.
K.D.
You’re assuming that your bed is flat. “Flat” means that all points on the surface of the bed are contained within a common geometric plane. They aren’t; they never are because it’s a physical thing. The question is “how far out of flat is my bed?”. Put a straightedge across the bed where you printed your test piece and illuminate it from behind in a dim room. You’ll clearly see the high spots and low spots.
Double check your bed assembly.
Usually a bow in the middle like that indicates the frame is upside down.
It should be nuts down. Also make sure you’re using flat head screws so they’re flush with top of heated bed.
I checked everything - I disassembled the A350 complete - cleaned erverything and assembled it new.
I will try again.
But back to “auto calibration”.
What does it at A350 ?
I have/had 5 3D printers, XYZ, Dremel and Ultimaker.
They all have the “auto calibration” / “auto leveling” and there was / is no problem.
All prints are ok. So what the difference - I think these 3 companies are no wizzards. The mecanical design is comparable good like SnapMaker. I use the same slicer software and the same filament.
I my opinion the only difference ist the way they are doing that “auto calibration”.
But what can I da at my A350, if the software don’t helps me?
Small parts are ok - but big parts (that’s why I bought the A350) have a terrible surface - absolutely not comparable with Dremel or Ultimaker.
I think I found the problem …
I found, that about 10 screws are not really fixed - I had to fix.
(5 srews of the heated bed.)
I’m nearly sure, that they were fixed after assembly. (I cannot guarantee 100%)
Maybe they get loose because of vibration caused by the stepper motors.
I’ll check it now every day.
Now the print result is much more better - I’m satisfied.
K.D.
