Hi there.
The print bed is more or less uneven. What will the shape of the printed material look like with automatic leveling?
Good question. I would definitely want option 2.
I would be surprised if it was anything other than #1, since the height of an object is determined by tool-paths which are created externally to the SM.
There are two possibilities for No. 2.
One is the case where the unevenness of the print bed is not taken into account at all.
When printing the first layer, there is no correction of the Z-axis.
If the gap between the bed and the print head is small, the filament is crushed; if the gap is large, the filament is thicker.
Thereby, the first layer is printed on a plane parallel to the moving horizontal plane of the print head.
The first layer may remain a little uneven, but the second and subsequent layers are similar, so a flat surface is eventually created.
In the second case, the position of the Z-axis is also corrected according to the unevenness of the bed, and the Z-axis is corrected so that it becomes progressively flatter toward the top surface during printing.
This is unlikely, however, because it requires advanced calculations.
Great discussion, fun read.
The answer is both 1) and 2) are possible, depending on how you have the firmware set.
In practice, without doing anything, the answer is 1).
If you want 2) you will need to adjust the mesh as the way Snapmaker stores the mesh is incompatible with Marlin’s M420 Z Fade Height.
Z fade height gives funny results when the mesh values are very far away from 0. Snapmaker uses the mesh values around 6-7mm because it combines both the bed unlevel characteristics and also the home offset into a single number. For Z Fade Height to work you will need to adjust the mesh so it is near 0, and the home offset moved into M206 Z, for instance, by subtracting 6 from each mesh point, which can be done with G1029 D-6.
The fade Z height is used in the firmware to compute a scaling factor which goes from 1 to 0 as Z goes from 0 up to Z_FADE_HEIGHT, which is applied as follows:
raw.z +=fade_scaling_factor ? fade_scaling_factor * bilinear_z_offset(raw) : 0.0
You can see the problem, the way Snapmaker applies a 6mm-ish offset into the mesh to compensate for the home offset, as you get higher you are moving the nozzle 6mm closer to the bed as you reach the Z fade-out height, which makes things much worse.
Word of warning - when the mesh is set up per the standard calibration there is a software endstop at Z=0 that prevents the firmware for crashing into the bed. If you move this into M206 and out of M420 I believe you can drive the nozzle to negative coordinates if you’re careless, crashing into the build plate.
Thank you for explaining this in detail.
If the answer is 1), then it seems to me that the SM2’s auto level adjustment is only intended to keep the distance between the print bed and the print head constant and out of contact.
I use ABS more often and use rafts.
Here’s a case where there is no Z compensation with automatic leveling.
If I create a thicker raft, the raft itself absorbs the unevenness of the print bed.
As a result, the top surface of the raft is perfectly parallel to the horizontal plane of the print head movement.
We thought that this would allow us to produce a distortion-free printout.
We thought it would be interesting to have a way to change the compensation method depending on the presence or absence of the raft.
In practice, I’m satisfied with the SM2’s current state, since I don’t print anything that requires that much accuracy.
Thinking about this a little deeper
I had just assumed from the moment I plugged the machine in that it would be 2, but the more I think about it the more i realize it won’t be - and it begs the question what’s the point at all if it doesn’t actually compensate for the sheet not being level
The only time I could see this being of any use at all is if it is so far out of whack that the nozzle would drag in one area of the sheet that it wouldn’t in another - which would probably mean its time to change sheets.
The point is for bed adhesion. A variation of 1/10 or 2/10 of a millimeter on the initial layer will wreck bed adhesion.
Ideally your bed would be flat enough that the top varying by 2/10 of a millimeter would not be critical and you could just ignore z fade out.
Many companies latched on to the increasing sophistication of bed leveling methods and used that to make cheaper printers with even less level beds leading to the addition of z fade out.
Somehow we’ve now gotten to the point where even that’s not enough.
I can only laugh at my bed varying by almost a full millimeter from corner to corner.
Cant just tweak the frame a bit or shim?
You’ve got a 3d printer to make all sorts of thicknesses hehe
How about I send it to you and you make this flat
Print bed
Mounting platform (webbed carriage):
This is insanity
Lol, just smoosh it between a couple sheets of steel!
thats funny stuff
The heated bed is made from what looks like FR4, an epoxy-glass composite for printed circuit boards. It’s notorious for being neither particularly flat nor of particularly constant thickness. There are folks who route very fine features on copper-clad FR4, but they need a surface map to get good consistency in trace and gap thickness. The bed that came on my machine is bowed in the center almost 2 mm along the y axis and about half that on the x-axis. I haven’t tried measuring it as mounted, but I anticipate only different deviations, not much smaller ones.
The reason glass is effective is that it’s pretty flat by the way it’s manufactured. Float glass naturally levels because of gravity, and molten glass’s surface tension tends to pull surface irregularities straighter. So when you slap a piece of glass on the heated bed, you’re averaging out the high spots. The glass will deform a bit, but not nearly as much as the deformations of the underlying heater.
One way of flattening the bed might be annealing. I don’t know if the thermoset epoxy can anneal, nor at what temperature. If the temperature’s too high, it will demagnetize the embedded magnets. Plus the equipment needed isn’t simple, as you need a thermostat-controlled platen press. So I’ve put that method aside.
The other way of flattening is to use external forces, that is, bolts to force it into position. At the very least you need something stiff enough to counteract the forces exerted by the heated bed as it’s moved into flatness. The heated bed isn’t hugely stiff; I can feel it bend in my hands. On the other hand, the subbase frame isn’t that much stiffer. Some other mechanical element would be necessary. Forget about using plastic; it’s not stiff enough.
And then, if you want to use shims, you need one under each adjustment point and you need a way of adjusting them. This certainly isn’t impossible, but it’s not a quick job, either. Adjusting the shim means many cycles of bolting and unbolting. The shims need to hold their adjustment across iterations. And don’t think that plastic work here, either. Plastic at elevated temperature and under strain tends to creep, throwing off the calibration.
Glass is definitely the most expedient way of improving this.
That got a laugh, lmao.
I’m trying to get a replacement undercarriage for 2 reasons: Hopefully the replacement will be flatter, but also because with a spare I can take this existing one and try and sand it flat, either fixtured in the machine or removed, on a glass plate with sandpaper.
Separately, glass bed with new heater and cork spacers off the existing FR4 board. Perimeter supported, for ease of shimming, with some support in the middle to resist sagging.
Separately, I need new laser work plates because the updated design has additional mounting screws that, when combined with the flattened carriage, will stop curling at the edges.
Has anyone already tried to use the wasteboard as a bottom for the print and laser plates?
I was thinking to try that.
Using then cnc to mill the wasteboard completely flat. (or at least in parallel with x-axis and then fix the heated bed on that one.
I would expect to also help with the heating of the bed as it will be better insulated.
Only in the center a shallow pocket would be required because the thermistor is just glued on it.
Most of the mounting points actually aling with the cnc holes, so it looks easy enough.
You could question if it’s a good idea to put a heating element on wood. However you do exactly the same when you get something out of the oven and put it on a wooden coaster. And something out of the oven can easily be twice as hot as the heated bed. (and wood combust at around 300°C)
I intended on trying that but haven’t come around to it yet.
1 Like
I thought about that and I’m concerned with long term exposure to 50C+ bed temperatures affecting the wood. For those printing ABS running at 100C+ I have stronger concerns as all unbound water will be driven from it.
At 180C wood begins the decomposition process. That doesn’t leave a lot of margin for an accidental thermal runaway event. A fire would be certain.
Additionally, MDF, as an example, expands and shrinks with humidity at a rate 0.2-0.5% for a 40-50% change in humidity. After starting a print job in the fall or spring, as the wood heats up and the bed dries it will shrink in thickness up to 0.05mm, which could cause issues.
I would recommend scraping it flat rather than using abrasion. The buy-in to scraping is higher, but cheap compared to how much time you’ll spend on it with whatever technique. A 12" x 18" surface plate can be had for around $100, and that’s the most expensive piece. Michael Ward had an excellent series in Home Shop Machinist several years back on the subject. I would have thought Village Press would have turned them into a book, but not yet. Searching to see if they had, I came across this page, which is chock full of references, including the very first, an 1840 paper.
The problem with abrasion is that it’s very difficult to use this to make something actually flat. Abrasion rate is quite sensitive to contact pressure. It’s why optical flats are ground in triples, because the flat on the bottom gets more pressure on the edges and tends to concavity, which grinds the top into convexity. A surface grinder has a similar problem, that you need the work surface to be evenly supported to the base. Probably the best machine would be a horizontal lapper, but it would need to be a big one. If you still want to try abrasion, you’ll want a way of evening out contact pressure. One way would be to cast the frame in a block of plaster of paris, and then only push the block around from the sides. The block will restrain flexure in the frame and the mass of the block provides even contact pressure.
Scraping has the advantage is that it uses a measurement (spotting) that happens at zero force, so you could actually scrape a reference surface onto a spindly frame. Even you manage to abrade one side flat, the next step is get the other side flat and parallel to the first face. Another flat is easy enough, making it parallel is much more difficult.
And to reiterate, even if you had a perfectly flat subbase frame, it’s not very stiff, so as soon as you bolt a wavy heated bed onto it, it’s going to move and not be flat any longer. That doesn’t mean it wouldn’t be useful, but it’s also not a panacea.
Thanks for that thoughtful reply.
I’m familiar with scraping, and the dykem transfer measurement technique is very applicable here.
I think even without scraping it can be used to ensure flatness, as smaller and smaller areas can be sanded (for example) as it approaches flatness.
Since the contact area is small, and relatively large height changes are required scraping would not be a good choice initially, usually that would come after some other machining process to flatten it.
I also agree that the deflection significantly hampers any machining or sanding effort.
I have a 38" drum sander. What if I took photogrammetry of the bottom surface to create a 3D model. Then I could make an MDF or HDPE carriage CNC carved for it to nest in. Then I could run the entire assembly through the drum sander to flatten the surface, with the base mold resisting the applied downward force.
Maybe if you could find a manufacturing business with a time saver you could just run it thru the conveyor (with the support you are discussing). We have one but its only 9 inches wide so it wouldn’t work very well
Couldnt you just as well mount it to a cnc and mill it down?
Maybe. Except I don’t have a CNC, and I do have a sander. Specifically this one:
Between the conveyor and tension rollers it’ll feed through just fine.
Hehe fair enough, it might be tough to run it evenly unless you make some kind of jig to roll it across it evenly but thats doable too 