also thanks for the info about the corners, i just tried my first print in the lower left corner of the plate after doing 7x7 auto calibration and in lower left corner of the object it wasn’t adhering. I have a PEI plate in a box - not sure if that will make a big difference?
For now I am printing nothing larger than the 4x4 center squares of the build plate, maybe i will avoid outer perimeter of squares until i have more experience…
Same here.
What I was experimenting with last night, before I tried the method @Mads0100 proposed, was disabling the auto-leveling via the touchscreen, and doing a manual calibration (21 points? 21x21 grid? I forget which, it was from one of @brent113 's posts).
The idea there is that you do the calibration by entering gcode on the Luban console (not as bad as it sounds, really), then print the grid of calibration points to the console, then use gcode commands to modify the ones that are off. So if the front left corner is the four mesh points (0,0), (0,1), (1, 0), (1,1), then you would use four gcode commands (one for each point) to raise or lower the value of each point.
This should work, it can be verified by some quick first-layer prints, and it can be saved as a text file for applying as-needed in the future (subject to the usual caveats: make sure it needs to be applied, double check the values, and perhaps just perform the steps manually using the saved gcode as a reference when re-caclulating the values).
Leveling might even work properly soon, since hopefully the next firmware release will support measuring the bed at the correct location instead of 20mm away 
yeah when i saw that bug (same day i was setting up printer) i was like - well that would explain a lot. Makes me wonder what other bugs they have in code based on systematic way they were using co-ords. I see from the PR they changed to RAW coordinates? Fix: fix bed level coordinate system by xhycode · Pull Request #121 · Snapmaker/Snapmaker2-Controller (github.com) i hope all their code in Luban, controller that uses coordinates has been examined for issues too?
i did my first larger 5 inch by 2 inch print today - perfect so long as kept it in the 4x4 area, doing a bunch of flow calibration cubes, same, no issues. I can wait for them to sort out the wider issues with leveling meshes etc.
I had the print head off last night to remove a clog, and while going through the gymnastics of putting it back on and adjusting the probe height (seriously, why is the set screw for that not on the side of the toolhead) I re-visited the bed issues (0.32 difference from left to right, with a big jump around the right).
Using 1-2-3 blocks set up on edge (3" tall), I lowered the print head to them and found that the right side Z-axis module was between 0.1mm and 0.2mm higher than the left side Z-axis module (method: lowering by 0.1mm until the lower side traps the calibration card). This means that my previous alignment had reverted, as another user reported (I think it was @NilartPax but can’t spend the time to look, got eight pounds of pork shoulder to smoke by sundown) - not gonna look into that for now.
I couldn’t adjust the X-axis alignment by hand: it just didn’t change. So I loosened the carriage screws on both Y-axis modules, lifted up 2mm, and gave the right-side of the X-axis module a whack from the dead-bow hammer. The 2mm lift was so the movement from the hammer hit would not cause the X-axis module to strike the hardened steel 1-2-3 blocks and damage the module.
Applied my usual backlash-resistant (like bullet-resistant, as in it may not actually work) method of up 10mm, up 1mm, down 10mm, down 1mm, down 0.1mm to taste. I had to repeat this process twice, using gradually heavier blows with the mallet, until I got a reliable result (card pinches on both sides). Home’d the print head, check again, all good.
I have not yet set up an indicator to see how much this helped. Not sure when I’ll have time. Gotta retry the print tht failed due to the clog and get that prototyped part out to the shop to a) see if the idea is sound, and b) implement it in 1144 if it is (nothing special about the 1144, I just have some the right size).
UPDATE: Huge waste of time. Print head still stops extruding on the first or second layer of a small print. Might just take the Snapmaker out to the range and shoot it. Or change out the nozzle and hot end AGAIN and do all the calibration AGAIN. And everything was working great just two days ago - 3D printing really is a pretty awful technology, at least at the consumer grade.
Did you turn the power of when you tried to push it down onto the blocks?
If the power is on, the stepper motors keep it fixed at their current position!
Yep! I didn’t for the dead-blow though, as that was more about seating it on the mounts. And, really, I got tired of waiting the damn thing to power up, home, and so forth every time I made a change.
Given how frequently these things need to be calibrated (damn near every time you use it, judging from my experience), there should be a mode to bypass all that garbage. Hmm, maybe there is, I should check the developer-mode options.
EDIT: I should amend my previous post, by “it just didn’t change” I mean that the right side wouldn’t go any lower. I could have moved the “left” side up, probably, but having done that before (followed by tedious back-and-forth as the two Z-axis modules take turns being the high one) I decided not to, as clearly it didn’t resolve the issue permanently. I’m going to give the print head one last go with a new nozzle, and if that doesn’t print right I’m packing the thing up and moving it to the attic. This POS has wasted enough of my time without introducing any time or cost savings.
@edf I continue to have a physical gap of over 1mm between the top of my 123 blocks on the left side and the x rail. Tramming does not fix. Calibration mesh mostly accommodates (and new firmware is definitely an improvement) but i have under extrusion (presumably from nozzle height) on the far-left side and toward the far back.
As part of an open ticket for support where they are proposing replacing both the platform and possibly the heat bed i found the following during more testing for them.
I measured the left side gap between top of blocks and X bar (using feeler gauges):
• 1.19 mm when measured from 123 blocks placed on heated bed
• 0.65 mm when measured from 123 blocks placed on center line of platform
• 0.48mm when measured from 123 blocks placed on the linear carriage mounts
(i only tested this once per test, so if this is slippage on the left Z module i won’t have detected that)
I continue to not understand why the bed levelling mesh doesn’t account for this - i can see the same sort of slope / difference in the bilinear mesh data…
–edit–
i am not sure that using 123 blocks under the x bar and measuring with feeler gauges is a good approach?
using a digital dial meter attached to the print head i am unable to replicate the 0.48mm result
–edit 2—
i finally figured out easy way to prove if deformation is coming from platform, heated bed or build pate.
rotate each item around the y axis 180 deg (means for heat bed you have to keep it unplugged) - if its is any of those items then the slope mesh should rotate 180 deg (it didn’t).
I’m going to decommission the Snapmaker.
I realized I’ve been approaching this problem entirely wrong: I expected an “it just works” device that I could use to print parts and prototypes in order to save time and material in the shop. Instead, it appears that the Snapmaker is a Project, and one which so far has required orders of magnitude more time than it has saved.
I am going to treat it like a used (or import, heh) piece of machinery: tear down, inspect, clean and rebuild. I might be able to make a reliable machine out of this, someday. As received, it’s just a toy for making more garbage (quite literally: every calibration print is destined for the landfill).
Some day I hope they release a full QC document so we can fully tear down ad rebuild properly.
I know things can be frustrating, especially when some machines seem to have lots of problems and other work out of the box just fine.
Have you check the platform yet? The platform will also cause the adhesion issue. You can contact the support by sending an email to support@snapmaker.zendesk.com.
Should have followed up here. After some prodding from people wanting to print things, I decided to fix the A350.
I took the Y-modules off the base and re-seated them, also took the end-caps off the Y modules to check for cable pinch and re-seated those. Did not remove the platform or even the heated bed for this. I was looking for problems and did not find any, but a lot of the erratic behavior I was seeing for a week or two before my previous post went away after the reassembly. I think a module got loose or a cable got pinched or something.
I used a dial indicator to measure Z-offset of the center of each grid on the print sheet, starting with one of the center squares as the zero-point (as those always seem to print the best). I wrote this grid down so I have a reference of low points, and place larger workpieces so that the edges are as close to the same Z-offset height as possible. I also determined that the difference between low and high points, excluding the edge squares, was something like 0.21mm. I started using a first-layer height of 0.24mm, and this has fixed most of my adhesion problems.
As for the platform, I did check it previously and found the machined bosses for the bed mounting screws were rough and not of equal height. This likely contributed to the height differences in the bed. I used a 3000-grit stone to smooth them, using a long enough stone to bridge two of the bosses so I had something resembling a reference for flatness. That seemed to help a bit: just smoothing the tookmarks from the machining process probably knocked 0.10 off the height differences (not all at once obviously - 0.002 here, 0.001 there). After doing this, I was in the 0.21 range referred to above.
One caveat when troubleshooting Z-offset: NEVER USE THE TOUCHSCREEN. The touchscreen is convenient and applies the change immediately (or close enough), but it saves the result across reboots, and there is no way to view this saved value, nor to clear it. If you are working off handwritten-measurements like I was, and calculating possible z-offsets to try, it is better to apply the Z-offset in Luban (USB mode works best), and only use the touchscreen to set and store the final offset that you have determined to be correct. I got bit enough by the saved Z offsets that now I change them in the slicer (Cura) when it’s job- or filament-specific.
I should report my experience. Since receiving the A350 i was never able to achieve a good leveling. It was good enough for small or medium parts, but never good for large parts or multiple parts covering (almost) the whole bed. No matter what I tried - 3x3 automatic, 5x5 automatic, manual with USB cable and Excel spreadsheet, etc. Every time I calibrated, started to print - one side of the bed had no adhesion, another side had so much adhesion that I destroyed 2 Snapmaker PEI sheets (the prints literally fused into them). Until I found this thread, read through it and someone mentioned loosening the bed and tightening again. I did that before a couple of times, but today I remembered that I’m also a “gun nut” and I have tools for tightening optic scopes, which include not only hex bits, but also torque limiters. So what I did, I heated the bed to 80 (I print PETG and this is what manufacturer recommends) removed the print sheet, loosened the screws (all while the bed is heated) and then tightened all of them in a criss-cross pattern starting from the center to precisely 15 in/lb torque. Put the sheet back on, calibrated with 5x5 automatic and bingo - perfect print from the first attempt. Had it a bit loose so dialed in -0.05 z-offset while printing the skirt. And it’s a full bed print, barely an inch spacing from the edges.