Do you see any major effect of a non-symmetrical 3-point leveling positioning for the bed? I am having the situation in the attached image and was wondering if I must bring the third point to the center of the bed:
The BCN3D Technologies Sigma dual extruder uses a three point leveling system, as does the Cube 3rd generation. Any three points define a plane, removing the possibility of adjuster-induced bed warping. You will reduce the vertical movement by having the adjusters at farthest points, which is a desired feature. Having the points in a non-symmetrical configuration will make adjustments “interesting” but not impossible.
Consider a centered adjuster on the left side of your drawing. If the nozzle measurement shows the back side to be higher than the front by 1 mm and your threads are 0.5 mm pitch, you should be able to do “symmetrical” math for adjusting the right side knobs.
My third point is in the rear and the two symmetrical others are at the front, making adjustments easy. I’m not sure I always understand why the leveling algorithm has me turn the left knob to the left on one pass, then to the right on the next and back to the left again, although it certainly has to do with the equally confusing adjustments to the right knob.
The semi-automated bed leveling in the Sigma reads three points using the left nozzle, two in front and one on the left rear, then reads the same two in front with the right nozzle and then the right rear. I’d love to see “back-door” values of this process. I expect that the frequently-lauded pronterface would give me some insight, but I don’t have the brain cells to put that package to work.
What will you be using for calculating a level bed?
I am not sure I completely understood your response :). A plane is defined by 3 points, yes, wherever they might be, but in a real world situation the weight distribution of the bed left and right side of the one-screw-side might affect the slight sagging? of the unsupported corner or might stress the screw “bowing” it sideways towards the heavier half? …or things of this nature, minute as they might be.
I am trying to avoid having the Z axis moving in any ways during the layer printing to compensate for an unlevel bed so I will try to see if I can adjust manually and leave it like that and see how long until it gets out of level again due to less than ideal construction. If it will be bad I will use a probe to test flatness and let the firmware deal with it.
This would work if the bed is perfectly leveled , however one problem I foresee is that the lower left corner would have a larger impact while levelling. E.g. If the top right corner is up by 1mm, the lower left would be down by ~1.5mm
@Dushyant_Ahuja
Umm, yes you are right, I haven’t thought of this. I am trying to envision a way to level this…
@Florian_Ford you will have to extra careful on the top right nut. That’s all.
@Florian_Ford If the bed is going to sag at an unsupported location, the material used in the bed would indicate a design flaw. If the bed material is not stiff enough (glass works!) all bets are off. Then again, if the material is not stiff enough to manage the portions unsupported by the single screw, it’s not strong enough to support the entire bed between the two screws and the single screw.
@Dushyant_Ahuja What you suggest is correct as a starting point, but with three adjusters all points would eventually come out level. Asymmetric changes from one screw to the other are to be expected, but can be calculated as a proportion of the spacing of the single screw to one edge and the spacing of the single screw to the other edge. Exaggerated example: the single screw is 25 percent of the way to the center. An adjustment of one of the two side screws would indeed increase one corner more than the other. Also a factor is the distance from the single screw to one of the two side screws. The distance difference between that measurement for each side screw is going to be small, destroying my earlier belief that the single screw side will have severe changes when measured at each unsupported corner.
A solidworks or fusion 360 dynamic model is needed and I lack the skills to create such a thing.
I would expect that small incremental adjustments would level out the bed in a few passes. I remember my first delta-style printer manual calibration when you would rotate between towers for many times until the leveling would be done. I kind of like the delta-style for auto-leveling, because the carriages all move on the same layer while for cartesians the Z should only move on layer shift.
@Florian_Ford I think your expectations are accurate.
@Fred_U Actually the bed would be ALU heat spreader+glass on top, which should provide enough stiffness for sag-free in the unsupported corners but I asked this question just to make sure I wasn’t leaving anything obvious (to others) out. And it turns out I kind of left out @Dushyant_Ahuja 's observation. Luckily it is not something too bad.
Actually, having unique values for X and Y at each levelling point helps the process.
I have configured rear-left, mid-right and front-centre and it works very well.
@Neil_Darlow Could you please elaborate a bit? Are you saying that this asymmetrical configuration helps with leveling the bed? In what way?
what do you mean by “I have configured rear-left, mid-right and front-centre”?
You need to have a heatbed and glass surface that are flat and thick enough to support their own weight and a print, if they are not then no number of screws is going to fix that huge issue. that is the starting point, will this bed support itself and a print of max build area and not sag, if it wont its not appropriate for the task and needs to be redesigned with stronger materials. If it is, continue to ‘how do I then attach it so I can adjust it to level?’.
on my i3 I use 3 corners with one cantilevered unsupported with no issues at all, the glass and aluminum are plenty to support a print that is usually center of mass near the center of bed anyway. for me with that setup the 2 diagonally opposing corners act as an axis through the center, and the one corner with no opposing screw rolls the bed plane level. It works extremely well and is super quick to level. one screw acts as a master height, one I level occasionally to the master, the third adjusts the roll of the bed on the axis between the other 2 and is the main adjustment screw. In RL for my machine that short workflow is: set screw height on back left corner with calibration paper as master, come forward to opposing corner and set to level the axis across center, move to roll adjustment corner and set the roll of the plane to level. check at cantilevered corner occasionally, but it works well and I have come to trust just quickly setting the roll at the adjustment screw and not worrying about a double check. Its reliable.
On you machine I am not sure what is front and what is back from the image, but you want to use the axis and roll method I describe as if you try to level 2 corner on one side first then bring the roll of the plane up to that, you have to deal with issues if the nozzle cannot get to the exact axis between the first 2 screws you level, if its not then you will be changing that height as you change the plane of the bed. Nozzle has to be able to sit directly over the axis between the first 2 screws being leveled, at each end of that axis just inside the screw, or it’s a tedious pain in the butt instead of a quick painless process.
looks like diagonal will work for you, they dont need to be at the corners if the material is appropriate for a printbed. Good luck post pics :o)
My setup is Tilt with the two–screws-side points and Roll with the single-screw-side point, hence the issue on the Roll side of adjusting of the point not being centered.
So I believe I would do a classic leveling with the two Roll and Tilt axes making a cross (off-center on the Roll axis). I’ll have to take it from there, when I get there 
Will post pics.
I hope we all are not missing something extraordinarily simple in this discussion. If the glass at each adjustment point is the same distance from the nozzle, and allowing that the glass is flat and not warping, then the glass is level at all other points. If it’s not the same distance from the nozzle, each screw is adjusted in turn. After each adjustment, nozzle clearance is checked and certainly can be expected to change at the adjusted point, but also slightly at the other points, but not by much.
If one takes the nozzle clearance measurements at some distance from the adjuster, one should use a radial of the center of the three points to the adjuster and extend from there. That is to say, draw a triangle with lines between all three adjusters. Find the center of the triangle, then draw lines from that center back to each adjuster and beyond. If the measurment points are on each line, as far as possible on the glass bed, it will be representative of the adjustment in a linear manner independent of the other adjusters, generally speaking.
I would like to see the formulae and algorithms used by the software that determines how much to turn an adjuster and in what direction, as well as how the input information is collected. My Sigma has microswitches on each head. Click, there’s a z-measurement.
I hope you finish
@Itzel_Ortuno
I hope that too, life really stands in the way of my design getting built
@Florian_Ford In words I meant rear-left corner, right-side middle and front-edge centre of the bed e.g. 0,200 200,100 and 100,0.
To correctly specify a plane using three coordinates the X and Y components of those coordinates must all differ.
If you used e.g. 0,0 0,100 and 0,200 (identical X values) that would not define a plane just a line on a plane.