Impossible calibration due to extruder weight

Been plenty of time figuring out why could I not calibrate my bed. I own a Prusa i3 and the bed is made out of aluminium heated bed and a thick glass. I've checked that they are both perfectly flat. The matter is that I can manually calibrate the bed on each edge of it, however, when the x axis is in the midle, it touches the bed (as if the glass was higher in the midle). I think the problem is that the extruder weight makes the nozzle go a but down when it is in the midle of the two aluminium bars (because it does not happen when the extruder is at the beginning or end of the x axis and this problem does not apear in the y axis (this one is always well calibrated).

I'll try adding a counterweight but it is not ellegant at all. Any idea?

PD: sorry for my bad english, hope I have managed to explain myself, thanks.

A counterweight won't do it, it is the axis bending in the middle. Upgrade to thicker rods, Or try and lower the weight of the gantry, but obviously this does not address the root of the problem

Any printer with 8mm rods and direct drive will show that artifact, either switch to a thicker rod, linear rail bowden system or do as most of the people do use ABL to compensate for it.

Quote
ggherbaz
Any printer with 8mm rods and direct drive will show that artifact, either switch to a thicker rod, linear rail bowden system or do as most of the people do use ABL to compensate for it.

Does ABL compensate for sagging rails?

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Not for rails but for rods they do. Haven't seen a saggy rail yet.

They do not bend vertically; the rods rotate, because if I push the rods straight down they do not bend at all. I'll try with the counterweight, there's nothing to lose. I'll tell u how it turned out.


Thank you all

By the way: how works the abl?

The problem here is a twisting force in the vertically stacked 8mm rods. I see this on my i3. I do not see it on my corexy despite it having the same length vertically stacked 8mm rods, the reason is the x carriage is extremely light on the corexy as it has dc42 sensor and a flex3drive extruder which are very lightweight units, my i3 has a wades extruder with motor attached, a large ABL sensor, a part cooling fan etc.. so its total mass is maybe 4 or 5 times greater. The rods do twist if you apply a force to the nozzle with a finger, but this is outside of normal printing forces so it does not manifest in the prints.

A horizontal arrangement of the rods does not have this problem but that isn't feasible with an i3. Stiffer rods might work also but none of the steels available are really stiffer than any other. Carbon fibre might be stiffer but the bearings will not run well on it and its lifespan might be quite short. A pair of linear rails would be wonderful, but expensive and quite a bit of redesigning of the z slides to achieve it. Thicker rods is the way to go, you can go up to 10mm, there must be stl's available for them. You can use a rod deflector calculator to work out the effect of changing to thicker rods. [home.huntvac.com]

ABL as it is implemented in Marlin will not solve this problem at the present time. It merely applies a correction plane to the coordinates as it prints, so it assumes a flat bed that is on an angle rather than a bed with varying heights across the surface. However the Marlin dev team especially Roxy-3D is writing unified bed levelling code, which will allow for probing at multiple points and a map of an uneven bed (or in this case drooping/bending 8mm rods) to be made when auto probing. Then printing will be possible on an uneven bed. Purists will point out this will make uneven objects and I agree with them, but it might be a level of unevenness that is acceptable in many cases i.e. printing larger "art" objects rather than larger machine parts. For smaller objects it is likely to have less effect. If a slightly uneven object is a big problem then a better printer design/better printer is the way to go. This auto-mesh-bed levelling feature will be a lifesaver for those with a very large printer where a truly flat bed is extremely difficult to achieve, and the effect of 0.1mm unevenness in an object of large size will be fairly unimportant.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I guess I don't understand the difference between a rail and a rod...

As ABL works now, it effectively tilts the print to match the tilt of the assumed flat but unlevel bed surface. It assumes that the axes are orthogonal. If it starts compensating for multiple points -i.e. a non flat bed surface- does that mean every layer and the top of the print will match the nonflatness of the bed surface? Is that desirable? Assuming that a nonflat, nonlevel bed is an undersirable condition, shouldn't the top of the print (at least) be flat? If the printer can't keep the bed level or flat, is it reasonable to assume that the axes are orthogonal and that they stay that way? If the printer does maintain orthogonal axes, why can't it maintain a level bed?

Never have so many worked so hard to avoid something as simple as building a rigid frame and using a flat bed surface...

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

When I say rails I mean linear rails - these:


I agree non-orthogonal axes and uneven beds are not desirable, but they are more affordable in terms of time and/or money. If all axes were orthogonal and all beds even and mechanically level we wouldn't need ABL but they aren't and a lot of people do need it.

If we can map bed height in a mesh we can make substandard printers like i3's work well enough for most. A $40 2d printer works well enough mainly as software compensates for the cheap hardware.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
alBarras
They do not bend vertically; the rods rotate, because if I push the rods straight down they do not bend at all. I'll try with the counterweight, there's nothing to lose. I'll tell u how it turned out.


Thank you all

By the way: how works the abl?

You are thinking of a counterweight behind the carriage? It would possibly solve the "hump in the middle" problem but will make your carriage even heavier and therefore reduce your printing speed, which may or may not be a problem.

Maybe an extruder design which places the motor on the back side of the carriage maybe with a belt? Trying to redistribute the mass you've already got makes more sense than adding "dead weight".

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Not too many people can afford a 2000 dollars printer or are willing to put that much into something that they may not keep, that's where the cheap printers are effective, it allows the introduction to the hobby and experimentation, that's why even though isn't desirable to have saggy rods and flimsy frames it will keep happening untill new production methods improve current designs.