Since the built in bed leveling calibration is useless, I’m thinking about attempting to install a z probe of some kind. Has anyone done it yet? I realize that it won’t be effective for dupli/mirror modes, but it would be great for single extruder and multi-material in default mode.
The original bed leveling works perfectly fine for me (and largely also did while my printer was running on the original firmware), as long as I make sure the nozzle does not drip in the least - which I cover by unloading any filament, wiping the nozzles, waiting very long (about 20…30min) with heated nozzles before wiping them once more and only then continuing for every z calibration. Thankfully the system is very stable, I only need to recalibrate after a hotend switch.
Is your glass bed even and are the plastic holders of the heat bed okay?
I have just thoroughly checked everything I can think of. Everything is tight and in good shape from the clip brackets to the rails and frame. No slop in the bed guide rods either. Cleaned nozzles, recalibrated, and started another print. No change. Back right corner is still high.
I put a straight edge on the PCB after calibration and the touch points are all in alignment. The back corners are quite a bit lower but that shouldn’t make a difference because the PCB conforms to the glass when it is clipped in. Straight edge on the glass shows that it is surprisingly flat in all directions.
I’m really not sure what to try other than putting the machine on the flattest surface I can find and loosening all of the frame connectors and then retightening them in case the frame just isn’t square. That is probably the proper fix but it sounds miserable lol.
@Wombley
Leave that, I would be very surprised if that would be it. Unless Snapmaker has issues with their manufacturer all of a sudden, the J1 frame is really sturdy: one massive rectangle on top, one on bottom, and four columns connecting them. There is not much that can go wrong there if the cast parts are okay.
Since you say the back right corner is always too high (how much?), regardless what you try, I would expect the glass plate to be suspect no.1.
- What happens if you turn the glass plate by 180° (logo facing to the right back, black side still up)? What happens if you turn the plate 180° vertically (logo facing to the back left, black side down)?
- Do you have a larger flat surface at hand? If your home happens to have e.g. polished granite windowsills, those are perfect. Otherwise a quick stroll to the next cemetery might do the trick, at least in spring/autumn when outside temperatures are not too far off
. Find a place with the straight edge where the stone is really even, thoroughly remove any dust and then place the plate on top of it. I also had one print plate which looked close to perfect when using the straight edge in all six directions, but when I placed it onto the windowsill I found one edge was feelably bent - just about 0.2mm in total if I measured correctly, but that was by far enough to cause issues… (Interestingly, that problem vanished after about a year with the plate just lying on the shelf, it is even by now)
I’ll have to try flipping and rotating the glass later. It’s worth a shot I guess, but that glass was dead flat when I checked it.
I don’t think any/many manufacturers check for square during QC inspections. I think the good ones design their parts to make it hard for them to not be square, and I do appreciate snap makers design of this machine. I just know that it would require flat tooling plates and jigs at every assembly station to ensure that these are square when they are torqued up, and I cant imagine that happening with a hobby 3D printer manufacturer.
Hm, if you design something correctly, the CNC machining of the parts is enough to assemble them as they should. Jigs and such are only required if the design engineer has not done a good job or if the part size or function made it impossible for him to reach his goal - at least of you are not working in an absolute high precision range. And FDM printers are definitely not to be seen there since the precision the process of making a lot of plastic spaghetti stick onto each other is well away from laser optics and such 
But I still fail to see where a rectangularity error of any type will end up in a bed leveling error as you seem to suggest if I understand correctly. As long as the bed and the linear bearings are straight in themselves, the 3-point bed leveling which uses the hotend nozzle will always result in the bed being aligned parallel to the XY gantry, if it has been done correctly.
There’s no possibility for misalignment in the way the rails are mounted to the frame, assuming no error during the machining process, so I have to assume that there is misalignment in the frame itself. It is designed well for rigidity but it isn’t exactly designed for perfect fitment. That would require notches and tabs, or something similar, and tight tolerances. As you said, it isn’t something that should be expected in a 3d printer, so I’m not complaining about the design. However, there is the possibility of movement in the frame when the parts are being assembled and torqued. A slight twist or a single vertical extrusion that ended up slightly out of square could result in one corner of the frame being very slightly higher or lower than another, and that frame is what the x/y axis rails are mounted to, so there would be variance in that axis. It could also result in a slight skew in a z guide rod that could effect the bed alignment differently at different heights. It doesn’t take much to throw off a first layer. We’re talking about ~0.2mm difference between the left side of the bed to the back right corner and ~0.15mm from the left side to the front right corner.
Yes, but you ignore that such a displacement would result in the print bed being offset in the corresponding edge by that amount, since that error would be already there when calibrating and cause you to adjust the bed a little higher. This means you would get a slight angular error of the z direction of a printed part from this - but not a bed leveling/first layer thickness error.
Well then, I stand corrected and don’t have other suggestions. This is how it is on mine as well.
The three lines between the probe points on mine, checked with a straight edge, are all flush and define the highest plane. My back corners also bow downwards (about equally, and very obviously).
I do not believe the PCB or frame flexes appreciably. If the two back corners are lower, that does not to me explain why the back corners would raise up though either. The bed is clamped to the triangle defined by the probe points in my thinking.
Because I agree, the glass is dead flat (and rigid). And if it isn’t, that would be obvious.
I do usually need to adjust my bed up by almost 0.2mm. I think from the PEI coating roughness, and a laziness to just fix the glass thickness setting since I’ve only had to reset this a few times, then it’s usually stable for months.
(I say because of the coating, but I also recall the Snapmaker Cura profile had some absurd flow settings that seemed to be thinking to smash way too much extrusion into the first layer, and I reset this all to unit flow and thickness.)
You’re right, and this conversation has brought me all the way back around to the belief that the bed leveling calibration is useless lol.
If it cannot be explained by any aspect of the machine up to the point of calibration, then the error must be introduced after calibration. Clipping the glass onto the PCB must cause some movement in the PCB and underlying bed support. Bed leveling needs to be performed on the actual print surface. Unfortunately, I can’t even level the bed manually with the glass in place because there’s no way to adjust the back corners independently. I’ll probably either design some adjustable clamps for the back corners or I’ll use Marlins manual probing feature to create a mesh for z compensation.
The heat bed bending down was done intentionally, in order to press the bed centre against the glass plate.
What I learned and what I suspect to be the next thing to investigate if the bed itself has proven to be flat:
- after wiping, the printer probes first in the centre back, then front left, then front right.
- the longer the time after wiping, the higher the risk a tiny amount of filament oozes out of the nozzle (hence my remark above to wait, wait, and wait a bit longer to be sure)
- as soon as there is a filament drop hanging at the nozzle, the contact height will be later than it should, because the filament drop will momentarily harden when coming into contact with the bed and therefore prevent contact. The bed continues rising until the hot nozzle has melted the drop again and pushed it out of the way, and then the printer “sees” the contact and reports the height to you on the screen, telling you to rotate the screw to bring the bed higher.
- since a) even a tiny drop of oozed filament will cause this to happe and b) the larger the drop, the larger the effect and c) the largest drop will always be on the last measurement position, such an error will cause you to screw the front right screw higher than you ought to.
- finally, if you take a look at the three-point fixture and the kinematic it creates when adjusting the bed screws and keep in mind that the front of the bed has a little more vertical flex than the back due to the lever arm acting on the z axis, this means this may cause your issue.
For the Duet in my J1, I improved the calibration process by decreasing z dive speed (beneath other optimisations such as checking for probability of the single reference measurement results before calculating the average value), and I remember using plated nozzles instead of the unplated original ones also helped a lot. And I always wipe using a clean cloth, not just the funny wire brush Snapmaker wants to be used.
IMHO the most stupid idea Snapmaker had was to keep users from driving to z=0 via the touchscreen. If you could do this, you could easily use the old manual “paper between nozzle and bed” calibration process to check if the bed is a) really even and b) correct or at least even out any tiny leveling errors as far as possible in case you need it.
Thanks, I’ve got an unused set of hot ends that also happen to have plates nozzles. I’ll try the calibration with those soon, without any filament in them, and see if that makes a difference.
Great that you have this option! Just to name the obvious: Make sure that the three golden circles on your heat plate are clean as well, and then you will get the best calibration the J1 can do - should be well below 1/10 mm. And if that is not okay then, contact Snapmaker support - because then your printer does indeed have an issue.
Oh, one more thing: check the distance between your print bed and the support frame before you start, the springs must be preloaded sufficiently. I know of one case (yes, @Miq19, yours 
) where the distance was off that much that the springs were pretty much unloaded which resulted in calibration running off frequently. The J1 assumes that the centre back screw is set to a correct height by the user before starting calibration. Pictures are in this german thread (sorry, Snapmaker never cared to repair my account, I can only post pictures if I posted them before in an email answer - uploading does not work for me): J1 - Layer shifts nur mit PETG-CF - Page 2 - Snapmaker - 3D-Druck Forum
Watching this as it’s a common enough complaint on the Facebook group too. Leveling works very well for me and it’s difficult for me to see what’s wrong from my own printer. When it works it’s very satisfying, and I think the overall rigidity/stability of the system is a strange benefit of the glass bed design choice compared to most printers. Using the probe points is convenient, I think, since I usually also have the bed off and hot to do the other offset calibrations at the same time when needed.
Would like to understand what could be “out of whack” to help others. It seems it’s always “the back right corner is too high.”
Mine’s just flat though and the PCB frame seems very rigid. I can only imagine some bolt is not tightened properly. But I don’t want to screw up mine to find out. 
Update:
Rotating and flipping build plate gets the same result. Swapped hot ends to a new clean pair with no filament and did the bed level calibration but still got same result. I’m stumped. I’m going to reach out to support but I’m not optimistic.
i had a weirdly similar issue and the center of the bed was always where parts were at the right height and the edges of the plate seemed too low. flipped the plate and that had the same results, i have no idea what changed cause a few levelings later and leaving my printer off for a few days somehow did the trick and im not sure what the deal is