Please give us the 11 x 11 bed calibration with heated bed!

They’ve already made the changes required in the GitHub to allow for the bed to stay heated when running a calibration gcode command. I would expect this to be in the next firmware release.


Hi John,
while I fully support your request: I tested recently and found that my bed, when calibrated cold (i.e. @~18°C) with 11 x 11, stays fairly consistent with that calibration when heated to 60°C. So while heated calibration would be the icing on the cake, I found it not as crucial as I expected. My assumption is that the numerous screws of the heated bed hold it down good enough to counter the thermal expansion.

1 Like

my firmware already has this option 5x5 is 20 points I think and 11X11 and heated or am I Missing something My version is V1.14.2_20220113.

Sorry I guess i meant close to full bed maybe you are asking for 11 points by 11?

You can do 11x11 via GCodes. Currently not available via Touchscreen. Would also be nice, but notnthat important :slight_smile:

I guess I need some bed leveling 101 . if more points is better taken to the Nth degee a point offset for every point within the movement step constraints of the printer would be perfect (like a laser measuring the distance on the fly of the point immediatly preceding the point of extrusion in the direction of movement.
would the printer adjust print speed and or extrusion rate in order to achieve a perfect top surface of the first layer ? isn’t there a limit within the capabilites of the printer that this would become irrelevant?

My understanding is that the print head moves up and down for the first few layers to compensate for the unevenness of the bed. The printer interpolates between measurement points to smooth these changes.

At some point, the smoothing is fine and you don’t need to add more points. In addition, there may be firmware limitations to the grid size, I recall seeing somewhere that 11x11 is the maximum, but I’m not sure if that’s accurate or why that may be. But 11x11 or less is typically fine for most users.

The printer moves the head up and down on every layer by the same amount. What you’re referring to is called “fade out” and because of some silly design decisions it cannot be used so its disabled.

That level for this machine would be determined by the smallest bumps, which have been measured to about 5mm. Nyquist then tells you you would need at least a 124x124 grid to perfectly compensate. At minimum a 60x60 grid to adequately compensate. Both of which are hilariously infeasible as there’s nowhere near enough memory on the machine to store that.

As such, the firmware has enough memory for an 11x11, so that’s as good as we get.

Many manufacturers opt for providing a flat surface, like glass. Then you don’t need more than 3 points to sample the plane and apply a correctional matrix compensation.

The history of bed levelling in software came about when manufacturers realized you can get “decent” enough results with a less-than-perfect bed and just apply a mesh compensation.

So basically an algorithm that averages , maybe inversely compensates with Z and assumes a level of flow during cooling to achieve a certain level of fade out then stored in memory in a 3D mesh matrix. we are tasked with getting the best data to put in to that algorithm by using whatever tools are available to us .
actually that is a pretty smart way to achieve and adequate result.

Thanks Guys very concise answers , it helps me to be able to visualize it in my head to help me understand what im doing and why im doing it. Thank You.

so I might be horrified if i use a precision straight edge on the bed itself ?

Prologue and important note:
I still 3D print just fine.

You can compare to my carriage in this thread: Carriage Tolerances - Unusable Over Distance >75mm From Center

and my bed here:

Bonus picture of the carriage on a known flat surface:


Oh Man I would think that that much deviation in the carriage when tightened to the bed would result in enough stress to make cool heat cycles completely unpredictable . wish they would make it so I can just leave the bed hot for my entire work (more play) cycle. seems like a no brainer but I’m new. I’m defiantly going to check mine when I install the new Y Linears when I get them. I have a friend with a very large 5 axis mill ( he paid for it doing outsource work for boeing ) He could probably index it for me pretty easy (and cheap).
Maybe snapmaker will use feedback to improve Quality Control , one can only hope.

Printing Great is the end goal I guess but repeatability and durability Count for a lot , This isn’t
1950’s farm equipment. You guys are a lot like those farmers though know your equipment and make it work.
Thanks for the “scoolin” I need it

1 Like

That was supposed to be definitely not defiantly

Yes 11 points x 11.
It takes longer so currently the bed has time to cool down quite a bit from the initial 60 degrees…

But the bed starts to cool down during the calibration process. As far as I know they have not address this part of it.

I think you would come up against some practical limits. If nothing else it would be pretty slow to do this on the fly.

Yes they have. It’s the very first reply in this thread

It’s this change:

So, not it’s not yet available in the last release, but it will probably included in the next one as the code change is already made.

If you can’t wait, you can compile it yourself. (it is actually quite well documented, even if you don’t have experience with embedded software development). Otherwise, you’ll need to have some patience to wait for the next official release.


Thank you for explaining that. I can wait as I now have my laser to play with for a while!

1 Like

As a data point, after installing the GitHub firmware while playing with @brent113’s inline power setup, I tested cold vs 70c 11x11 bed level: 70c vs cold 11x11 level - Google Sheets.

The summary is at the end, but tl;dr: the max difference between cold and heated (towards the center of the bed) was -.08 mm, with the average difference being only .01. Visually you can see that it deforms much more in the center vs the edges:

(Keep the scale in mind while looking at this)

What is interesting to note is that many people (myself included) find better adhesion at the center than the edges, which seems to indicate that the larger deformation actually helps adhesion instead of hurts (that or I’ve learned to level to the deformation leading to being too far away at the edges).

A bit of an update on my 11x11 experience - I today started a large print that fills ~80% of the bed, and I did a cold 11x11 calibration of the bed before (@20°C bed temperature). This works nearly fine! The bottom ~1 cm of the print bed were too low and the print did not stick there, but adding pieces of cardboard between the bed and the frame for the front three bed screws was a quick fix to that (if not a sustainable one…). Now here’s a bit of first layer p*rn, bed at 60°C:

And this is my bed vizualisation (it is really a bad, bumpy bed!):

My conclusions:
a) 11x11 is good enough for my awfully bumpy bed, and the printer does a good job to compensate
b) Heated bed calibration not needed for me (the problem with the bottom centimetre would not have been fixed by this: this part of the bed is just not properly covered by the calibration).
c) I still need to do something about my bed - the first centimetre problem is real.