Introducing The Experimental Bridging Settings

[burtoogle 513]

Introduction

The 3.3 beta introduces new functionality that is intended to improve Cura's ability to print bridges and overhung areas.  The current Cura code base has long had the ability to detect when a skin region spans across islands of support and those skins that Cura thinks are bridges will have the direction of the lines aligned with the detected bridge detection. But nothing was done to detect walls that spanned unsupported areas and no changes were made to the wall or skin print speeds, flows or fan speed.

 

The new bridging functionality does now detect when walls cross unsupported regions and it does now modify the print settings for those walls and skin that are determined to be unsupported. This topic introduces the new settings that are grouped together in the experimental section and are enabled using the Enable Bridge Settings checkbox.

 

The settings Explained

 

You will immediately notice that there are quite a few settings. Why so many? Well, I realised pretty early on in my experimentation that modifying the skin settings for the first bridge skin and then using the normal skin settings for the skins above the first does not always lead to a good result. So there are settings to not only modify the print speed, percentage flow, density and fan for the first bridge layer (bottommost layer) but also for the two layers above. If you want to keep things simple you can un-check the Bridge Has Multiple Layers checkbox and then only the first bridge skin layer settings will be modified.

 

The settings used for the bridge walls are Bridge Wall Speed,  Bridge Wall Flow (aka line diameter) and Bridge Fan Speed. Also relevant for walls are Minimum Bridge Wall Length which only considers wall line segments that are longer than this length as bridges. Shorter wall line segments are just printed using the normal settings. Bridge Wall Coasting controls a feature that reduces the pressure in the nozzle as it approaches the start of the bridge. This is necessary because the extrusion rate will need to be very much reduced from the normal rate as the speed and flow used on the bridge is likely to be less. If this is not done, the wall line tends to droop very badly at the start of the bridge. Finally, there is Bridge Wall Max Overhang which controls how much a wall line can overhang the layer below before it is considered to be a bridge wall line. By default it is 100% of the wall line width so the line has to be completely over air with no overlap whatsoever with the layer below. Reducing the value of Bridge Wall Max Overhang means that the line doesn't have to overhang the layer below so much for it to be printed using the bridge wall settings. See below for an example of its use.

 

For a given material and print temperature, there will be some combination of the bridge wall settings mentioned above along with Bridge Skin Speed,  Bridge Skin Flow (aka line diameter), Bridge Skin Density (line spacing) and Bridge Fan Speed (and the similar settings for the 2 other skin layers)  that produces the best results. This is where the fun starts because it's going to require a lot of experimentation to come up with suitable settings. The default values are tested with PLA at 200deg, 0.4mm nozzle and 0.2mm layers and may work OK, or not (YMMV).

 

One other setting is Bridge Skin Support Threshold which is the percentage of a skin area that is supported for it not to be considered a bridge. i.e. skin areas that are supported for less than this percentage (default 50%) are considered to be bridges and will be printed using the bridge settings. Skins that are supported for more than the threshold value will be printed just as normal.

 

Here's some picture that, hopefully, will make things clearer!

 

First, here's the settings showing some typical values:

 

 

Here is the first bridge layer, I am showing the layerview as feedrate so you can see the speed changes on the bridge. Notice also the coasting just before the bridge walls start (the walls are printed anti-clockwise):

 

 

Here's the second and third skin layers:

 

 

Finally, here are a couple of images of a model (thanks to the community member that sent it to me) that has some overhung regions. The first image is using Bridge Wall Max Overhang set to 100% and the second is using 50% and you can see that more of the lines are going to be printed using the bridge settings. For overhangs, I recommend setting Minimum Bridge Wall Length to zero as the wall line segments are likely to be very small.

 

 

 

Feedback Required From You

 

Please give it a go and share your experiences, all feedback is welcome (good and bad). You can either reply to this topic or for really bad news, please submit a github issue.

Edited March 27, 2018 by burtoogle
Removed unwanted image.

[neotko 1,417]

2 minutes ago, smartavionics said:

but also for the two layers above.

 

That’s brilliant ??????

[SandervG 1,521]

19 minutes ago, smartavionics said:

@SandervG, help! there's this blasted image stuck at the end of my posting and I can't delete it! Can you?

 

My first attempt of 'backspace' didn't work, I'm sure you tried it as well  Let me take a deeper look. 

[gr5 2,172]

I've had this problem before.  I forget how I fixed it.  I think I deleted all images and then re-inserted them a different way.

[burtoogle 513]

I cheated and made it very small!

 

[catohagen 41]

I've just tested a print with this setting on (brigde settings) but the result was....spaghetti  

 

I printed this : https://www.youmagine.com/designs/quick-temperature-fillament-test in greentec, with basicly default PLA settings (210°c)

 

Why are default feedrate set to 15mm/s ? Its way too slow, the string have plenty of time to sag, it doesnt help when flow is reduced to 75% either...

 

I edited settings with 60mm/s feedrate and flow back to 100% and got very nice bridging  

[catohagen 41]

here's the print with default bridge settings:

 

[burtoogle 513]

Hi @catohagen, thanks for the feedback. Well, all I can say with regard to the default settings is YMMV. They work for me just fine but I don't for one moment expect them to be optimal for other people. So, you have to experiment and find what works best for you. You can then share that info with others and when enough people have tried it out and got good results we will have an idea what are the default values that are most likely to be successful.

Edited March 29, 2018 by burtoogle

[geert_2 557]

Last week I designed a little test model to demonstrate the limitations of an FDM-printer to people who have no experience with it, but who want something designed and 3D-printed. Purpose was to show big overhangs and tiny extensions that get not enough cooling (the little cones at the top). I also used it to experiment with temperature and speed settings, for best results.

 

But the model might also be usefull for this new bridging function. So, feel free to experiment with it. Size is fairly small: ca. 40mm x 30mm x 30mm, about 1.5h printing time at 50mm/s.

 

The item at the back is a real bridge, with at one side a small horizontal overhanging plate, and at the other side a 45° overhang.

 

The bow at the front only has support on one side. So the goal here is to find settings that do produce the minimum amount of spaghetti, and the minimum of curling upwards while printing.

 

STL-file: overhangtest3d.stl

 

 

 

Edited March 29, 2018 by geert_2

[Brulti 177]

Torture print... I like! ^^

 

Except I can't: 'You cannot add more reactions today' ?

[SandervG 1,521]

10 minutes ago, Brulti said:

Torture print... I like! ^^

 

Except I can't: 'You cannot add more reactions today' ?

 

Hi Brulti, did the forum software tell you this? There are limitations to new users, to protect us from being spammed into oblivion but you should be in the clear. Could you try again, and if you can not, could you send me a DM? This should still work. 

[darkdvd 975]

@Brulti is too expressive....

[Brulti 177]

DM sent without a hitch, still unable to react to any message.

[burtoogle 513]

Hi @geert_2, thanks for the torture model. However, I cannot see how any FDM printer would be able to print that satisfactorily without the use of support. Bridging is only appropriate when there are multiple pillars that the bridge can be spanned across. Bridging cannot help when only one side of the area has support under it.

[SandervG 1,521]

Ah, it is set to a max of 10 reactions/likes a day. I guess I didn't expect us to be such a likeable bunch of people when I set it up initially

I'll change it to unlimited, so you should be able to express your appreciation to your fellow forum members again! 

 

Apologies for the inconvenience. 

[Brulti 177]

16 minutes ago, SandervG said:

Ah, it is set to a max of 10 reactions/likes a day. I guess I didn't expect us to be such a likeable bunch of people when I set it up initially

I'll change it to unlimited, so you should be able to express your appreciation to your fellow forum members again! 

 

Apologies for the inconvenience. 

 

We're all very likeable here! Thanks for the quick reaction!

[geert_2 557]

1 minute ago, smartavionics said:

...

Bridging cannot help when only one side of the area has support under it.

Yes, I understand. But a lot of other people don't know. 

So that is why I needed a demo-model to show them the effect with and without supports.

 

But for such a single-sided overhang like the bow above, if printed without supports, I have been wondering if the results would be better (=less spaghetti, but of course still some spaghetti) when the first layer is printed length-wise? Or if it is printed transversal, line by line, maybe with a reduced line-spacing (=thus with some overlapping), so that each line still glues a bit to the previous line? Very hard to predict.

 

In real life, printing small horizontal overhangs might be usefull for an architect who wants to print a house model, and the roof is overhanging a little bit from the walls. Or if you want to print a vertical pipe with a little flange.

 

For my real designs, usually I include the supports in the CAD design itself, like this (=the 2 purple items with ribs), for printing on my UM2. The hollow pink block is a dummy to provide enough cooling while printing the tiny top area of the model.

 

[burtoogle 513]

Yes, @geert_2, I agree with all that you say.

 

One other thing that I was discussing with a friend recently was the idea of using a bridge as support below the real bridge which is printed just like it would be done above normal support. Once the print is finished the bridge-support is removed to leave a very good quality bridge on the model. That could save a lot of time and support material.

[geert_2 557]

1 hour ago, smartavionics said:

Yes, @geert_2, I agree with all that you say.

 

One other thing that I was discussing with a friend recently was the idea of using a bridge as support below the real bridge which is printed just like it would be done above normal support. Once the print is finished the bridge-support is removed to leave a very good quality bridge on the model. That could save a lot of time and support material.

Yes that seems like an excellent idea, if it can be done without too much damage to the side walls (in single nozzle printers with only one material).

 

The concept might also work for small overhangs like roofs, if you would build a sort of triangular support that is only sticking lightly to the side walls, just enough to keep the roof in place, but still easy to remove.

 

I have quickly modeled such a thing in DesignSpark Mechanical. It's a very crude model, just to try the concept. The supports (yellow and red) are totally separated from the model (cyan). For adhesion, they would rely on the spaghetti and nozzle leaking that is produced in the first layers of these supports. Currently the horizontal gap is 0.5mm, and the vertical gap between the ribs and roof is 0.3mm, but these might need adjustment. I am curious if it is going to work? Next week I will print it (I am not here tomorrow).

 

 

 

Edit: dimensions: center overhang = 30mm, side overhangs = 5mm, width = 15mm. Little ribs on supports = 0.5mm wide, with gaps of 1.0mm in-between.

 

Edited March 29, 2018 by geert_2

[burtoogle 513]

Hi @geert_2, I made a test model to try out the idea of "flying support", i.e. printing support on top of a bridge. It worked very well and I think could be a useful technique in those situations where you need support but it's either a long way down to the bed or you really don't want to print support on top of the model that is below. Here's a picture of my test piece. You can see the bridge (which is cut off and thrown away along with the flying support) is only one layer thick at its first layer. Sure, it does leave a visible mark on the vertical pillars when it is cut away but it's not horrendous.

 

 

BTW - the 3.3 beta bridge code doesn't have a few changes that stop bridges being printed over support so if you try to recreate this using 3.3 beta you will find that it will print the skin layers above the support as bridge layers which isn't what one would want. I have made the required changes but they haven't been merged into the Cura release branches yet.

 

Update: You don't even need to make the first bridge layer the whole width, so in my test model I reduced the first layer to two 3mm wide "spars" that span the gap between the pillars. Certainly with PLA, the sheer strength is such that you don't need very much width to support the bridge.

 

Edited March 30, 2018 by burtoogle

[burtoogle 513]

Oh yes, @ahoeben, I used the support blocker to remove the support between the build plate and the bridge in my example above. Thanks!

Edited March 30, 2018 by burtoogle

[geert_2 557]

Here the print results of my free hanging support tests shown above. (To avoid confusion, keep in mind that this is a manually created test model, not automatic support generation.)

 

The basic idea of a support bridge to support the real bridge works great. The underside of the real bridge is as good as it can get with a single nozzle printer (UM2). And the support is easy to remove. So I am definitely going to include this concept in my inventory of support methods. Thanks to smartavionics for the great idea. 

 

But my model still needs refinements. The triangle support shapes do need modifying. This will take some trial and error. Now the edges curl up way too much at the outside (more than 1mm, at 0.1mm layer height). Then the nozzle crashes into these curls violently, and tends to knock the model off the build plate, or to separate the supports from the model. As you can see in the light green model (PLA, 210°C, 50mm/s). The dark green model survived, but was printed much slower (25mm/s), so the nozzle crashes were not as severe.

 

Also, for optimal stability the free hanging supports would need a few little strands to connect them to the main model. Relying on the spaghetti works well for the first layers, but after a couple of layers there is no more spaghetti. So there it needs a few connecting strands. Using strands in well defined places might also reduce the amount of damage to the side walls of the real model. This also will take some testing.

 

But the basic concept definitely can be used for printing small overhanging roofs and flanges.

 

 

 

 

 

 

[mastory 44]

I have wondered if it would be beneficial on bottom bridge layers to lower Z slightly in the first 1/2 -2/3 of the bridge span while stringing across, then in the last portion return to the normal layer height.  The purpose would be to fight the natural sagging that occurs while the line is still flexible, then connect it to the support at the other side.  Also, I have seen where when printing adjacent bridge lines ( especially bottom bridge layer) where the nozzle dissturbs the previous line and breaks it, sticking to the nozzle.  Lifting slightly in mid bridge would defeat that I think.

[mastory 44]

The Z motion mihht likely be a parabola relative to the bridge length...

[burtoogle 513]

Hi @mastory, in my testing with PLA, I have found that with suitable speed/flow/fan settings, the first layer strands do not sag on small to medium length bridges (say up to 50mm). Maybe for longer bridges or difficult materials the parabola would be worth trying. It would be straightforward to implement in Cura.