Add feature to avoid bed clips while probing

[apulver]

Description

Allows user to define four square areas in Configuration.h that probe.h's "can_reach( ... )" function will reject.

Requirements

None

Benefits

Using mesh_inset or probing_margin to avoid probing clips either shrinks the mesh or needlessly excludes many points that can be probed. This will avoid just the areas of the clips and the areas in which the nozzle is above the clip.

Configurations

Relevant #defines have been added to Configuration.h

Related Issues

None?

[EvilGremlin]

Does this 1cm of extrapolated unprobed area make any difference at all? Does it make visible quantifiable difference in print quality? I'm sure it doesn't, unless your bed literally looks like this

In which cale probing area is not a concern at all

[apulver]

Does this 1cm of extrapolated unprobed area make any difference at all? Does it make visible quantifiable difference in print quality?

This feature does not extrapolate unprobed areas. Rather, it avoids probing just those areas in which clips would interfere with the probe or nozzle, so that the user does not have to use mesh_inset or probing_margin. Depending on the probe offset, using mesh_inset for this purpose may reduce the size of the mesh significantly. If probing_margin were used, then the user would have to manually probe many extra points that the probe could have reached just fine without any interference from the clips.

[apulver]

Incidentally, if G29Jx is just computing a linear transform to apply to the current mesh, then it would be better to have its own separate "margin", set in farther inward than the margin (if any) used for building the mesh.

Also, I wrote the feature with UBL in mind to help the user maximize the mesh area and number of points probed. It does not rearrange the grid in bilinear leveling or anything like that (which is perhaps what EvilGremlin thought.) It's my understanding that with UBL, there is no extrapolation to outside the mesh area and that it's important that mesh_inset be as low as possible.

[EvilGremlin]

...probing_margin were used, then the user would have to manually probe many extra points

Nope, i mean both bigger mesh_insert or autofill with extrapolated values are perfectly fine when we're talking about ~1cm.

I think feature is ok in general and might find some use, just not for bed clips specifically. It drastically complicate geometry setup for no real benefit over simple mesh_inset.

[apulver]

Nope, i mean both bigger mesh_insert or autofill with extrapolated values are perfectly fine when we're talking about ~1cm.

Don't forget you have to combine the clip dimensions with the x/y probe offsets when you're looking to set a margin or inset. You could lose several centimeters to make both the nozzle and probe clear the clips. It matters when you're trying to print a large item.

I think feature is ok in general and might find some use, just not for bed clips specifically. It drastically complicate geometry setup for no real benefit over simple mesh_inset.

What exactly does it complicate? It's only a few comparison operations to avoid four areas adjacent to the bed perimeter with the probe and nozzle. It's probably optimized out of the compiled firmware entirely when it's disabled in Configuration.h. Many inexpensive printers come with a build surface that attaches with clips to a not-so-flat bed. I think it's very well-justified.

[EvilGremlin]

What? No, not several, just clip size. Probe offset vs travel limits is main loss in auto probing. In fact, in most cases it's preferable to set bigger mesh_inset (20-30mm) so all points will be probed automatically.

What exactly does it complicate

Support overhead. Soooo many people can't figure out geometry setup even as it is... Ultimately it's up to Scott, but even accepting PR i wouldn't include all this in configs.

[apulver]

What? No, not several, just clip size. Probe offset vs travel limits is main loss in auto probing. In fact, in most cases it's preferable to set bigger mesh_inset (20-30mm) so all points will be probed automatically.

Suppose your clip extends 10mm into the bed. If the nozzle is 15mm closer to that side of the bed than the probe, then the closest point with mesh_inset would be 25mm. The opposite side still has to be inset 10mm for its clips. This is not preferable because, if I understand, UBL does not extrapolate the mesh outside its perimeter. So on a typical printer you might have 35x235mm of bed area that UBL does not apply any z correction for. This is a problem when you want to print a big item.

What exactly does it complicate

Support overhead. Soooo many people can't figure out geometry setup even as it is... Ultimately it's up to Scott, but even accepting PR i wouldn't include all this in configs.

What do you mean by "geometry setup"? This feature would in many cases obviate the need to adjust MESH_INSET and PROBING_MARGIN in the first place, which are relatively opaque. If the user does not want it, they can leave it disabled, just like other geometric features such as skew compensation.

[EvilGremlin]

Oh well, we have very different views on leveling setup logic, that's fine.

[ellensp]

One issue I see is most systems that use clips, the position is not fixed, the clip will move position every time you clip and unclip.
Sometimes deliberately, if your trying to print a large part and the clip is in the way, you move the clip..

[thisiskeithb]

One issue I see is most systems that use clips, the position is not fixed, the clip will move position every time you clip and unclip.

Yep. PROBING_MARGIN_FRONT/BACK already exists and negates the need for these changes in most cases since that is more universal / simplifies configuration setup.

[apulver]

One issue I see is most systems that use clips, the position is not fixed, the clip will move position every time you clip and unclip.

Yep. PROBING_MARGIN_FRONT/BACK already exists and negates the need for these changes in most cases since that is more universal / simplifies configuration setup.

Personally I always took care to position the clips the same way every time when I was using bilinear leveling in order to avoid collisions and bad probe values. It's not possible to do this with a finer grid as is typical for UBL. Suit yourselves though.

[thinkyhead]

I'm not entirely disinclined to add this since it only affects UBL, which is already quite a meticulous beast and tends to be used by meticulous humans. But it does add a lot of new defines. So, I will at least reduce the number of defines and make them into 4-element arrays, one per obstacle.

One difficulty I see with this feature, as implemented, is that you might be probing a point that is near a clip, and even though the probe is taking care to avoid the clip, the nozzle might still run into it.