tilted mesh for multiple toolheads on kinematic beds

oliof

@o_lampe at the minimal changes we are talking about and with mesh fading, it should be minimal.

MaxGyver

I strongly support this idea!

I have almost given up on the whole IDEX Idea because of all the things you have to consider to make this work.

On paper IDEX looks like a neat feature, but in reality you need the X and Y axis to run absolutely parallel to the bed at every point. Something that I have not been able to achieve even with high grade linear rails mounted to a heavy CNC-maschined baseplate. Furthermore, both nozzles need to be at the exact same Z-Heigt and the Printbed also has to be absolutely flat and stay flat even when heated...

o_lampe

@oliof said in tilted mesh for multiple toolheads on kinematic beds:

at the minimal changes we are talking

I had the idea of using a standard E3D- and a Volcano hotend on the other tool. The difference in length would be huge and thereby the tilt angles.
I had it running on my "hitchhiker IDEX", but not at the same time (no dual gcode)

...and let's not forget the gantry sag. With two different tools (weight), it will be different too. You can't compensate that with a single heightmap from a single Z-probe.

oliof

@o_lampe I am sure there are uncounted setups where this may not be useful. That doesnt mean its not useful for the most common IDEX setup -- identical tool heads with identical depth, either printing the same thing cooperatively (where thus feature wouldnt be needed) or doing what only IDEX (or hash style) can -- mirror and copy mode.

I dont know why you keep dumping on this feature request. Its not taking anything away from you.

o_lampe

@oliof said in tilted mesh for multiple toolheads on kinematic beds:

I dont know why you keep dumping on this feature request.

It's not my intention to dump it.
I didn't understand, what it was meant for in the first place.
At first I thought you want to tilt the bed once after loading the heightmap in a way, that the highest and lowest probe spots build a line parallel to the gantry.
But it seems, tilting would be a continuous thing depending on the momentary position of both toolheads. (both in work-mode...)
That's probably easier to implement as the "individual z-axis per tool" I'm trying to propagate for 1-2 years (mini z-axis), so I'm all for it!

I'm concerned about bossing around a heavy bed. It may require stronger z-steppers, but it's only for a few layers.

o_lampe

I thought about this idea for a while and it looks similar to the "least square" algorhythm used for delta calibration.
Only difference is, there are three leadscrews (building a plane) but only two nozzles. Maybe a third virtual fixed point can be used to build a second plane? Like the front/left edge of the bed...( current z-height needs to be added)
Isn't there a way to modify the least square method and let it run continously?
Maybe THE delta-expert @dc42 can tell more?

dc42

@oliof is the problem that your Z probe can't reach enough of the bed width to detect the hump and the dip? Because it it can then you should already be able to do this, by choosing the probe points to be closer to the hump and the dip instead of probing at the extremes of the bed, which is probably what you are doing now.

oliof

@dc42 no, the issue is that in Copy/Mirror mode, I may need to have a different adjustment for each head in IDEX setups. And machines that can tilt their gantry or bed could do that to adjust for both heads at the same time by doing a coordinated tilt instead of a lift and lower of Z as we do with single head machines.

oliof

@dc42 maybe my example was a little bit too simple. Lets assume a mesh like this:

  |   0    1      2     3
  |====================
 0| 0.0   -0.1   0.0   +0.1
 1|+0.1   -0.1   0.0   -0.1
 2|-0.1   -0.1  +0.1    0.0

Assuming copy mode, and X is at 0.0 and U at 2.0. Both heads have the same correction factor, mesh bed is applied normally

Lets assume a move to 0.1 for X, U moves to 2.1. with a non tilted bed or gantry, we could adjust either for X (+0.1) or U (0.0).

moving on to 0.2 and 2.2 we could either adjust for X (-0.1) or U (+0.1)

In both cases, we cannot adjust the height of the heads independently. But if we would tilt the bed just so that the bed is 0.1 lower for X in the first example (Z offset +0.1 ->lower bed), or that the bed is tilted so that the bed is higher for X and lower for U in the second example, then we would have accounted for the bed imperfections.

I would assume that there shoud be limits to this adjustment, but I hope that makes the idea clearer.

dc42

@oliof thanks, I understand what you are asking for now.