Share:
Share
Notifications
Clear all

Supports and rafts in Prusa Slicer

SusanOKC
(@susanokc)
Famed Member

Hi, Luke:

I have one part I'm trying to print (the lid on a box) but am having trouble with the Prusa slicer support settings. There's a section of the lid that sits down into the box and then the top part of the lid rests on top of the box walls so the top of the box is completely solid. That means we need support under the lip of the lid all the way around. When we migrated away from MakerBot we cut down the box and lid to get them to fix on the Prusa bed, but now we're having an issue with the Prusa slicer. It seems the slicer mandates a raft outside the area of the supports. So while the lid and supports themselves fit on the bed, that raft doesn't.  I've tried painting a narrower strip of support to see if I could trick the slicer into bringing in that raft but that didn't work.  Is there anything else I can do to get it to fit without reducing the size of the box even further? A way to remove the raft from the G-code maybe?  Other ideas?

I would note we also tried simply printing the lid upside down but it isn't quite perfect that away, and the lid is a real area of focus in the finished product so that's not our best solution. 

Thanks so much! 

Quote
Topic starter Posted : 27/03/2021 4:12 pm
Luke
 Luke
(@lmf5000)
Expert Moderator

Hi Susan,

Good to hear from you again :). Thanks for including the 3mf file, that makes troubleshooting infinitely easier since I have the 3D model as well as the profile you're using so I can make changes and upload an amended file that you could theoretically put straight into production 😀.

So, firstly (this isn't what you asked, but you'll find this very helpful), your lid is a very large, solid flat area. This is the perfect use case for adaptive cubic infill, since it concentrates the infill only near the perimeters, top and bottom layers of the print, making the centers more sparse. It results in huge time and material savings, because the very center of the print (the least important part, since it's buried deep within the print where it doesn't contribute much strength or support) contains almost no infill.

The print you sent me used 105.11g of filament and 7h 51m of printing with 20% gyroid infill

With 20% adaptive cubic infill, that single setting change cuts it down to 99.6g of material (a 5% savings) and 6h 27m of printing (almost 1.5 hours saved on each print, so an entire day's worth of production saved for every 16 units you print).

---

As for the fix for your support issue, it's actually very easy, but not intuitive at all (you have to dig deep into the support generation algorithm to understand why it works).

Basically, PrusaSlicer divides the build plate into a grid of small squares, then decides whether each piece of that grid needs supports or not. So the size of these squares determines the "resolution" of the support (i.e. the size of the section of support that can be "on" or "off" - a lot like how a screen makes up a picture using individual pixels. Naturally, the smaller the squares the closer the support will conform to the actual shape of your part.

In Simplify3D this setting was intuitively named "support pillar resolution":

image

And that determined the size of the "columns" of support you placed on the plate with your mouse. So setting that to 2mm gave a 2mm resolution, meaning you would be placing 2mm columns of support, and the support confirmed to within 2mm of the shape of your print. Likewise with automatic supports. So here's 2mm resolution:

image

And here's 4mm resolution:

image

---

In prusaslicer, the resolution is governed by the "pattern spacing" setting -

image

But this not only affects the resolution of the grid it uses to determine whether that block gets support or not, but also the spacing between lines of support (which in Simplify3D was a separate setting called support infill percentage).

So in a nutshell, S3D let you independently set a) the resolution of its internal support-determination-grid (which areas get support) and b) the density of said support. With PrusaSlicer, you can't set them individually, you have one number that changes both at the same time.

In a way it makes sense for the two numbers to be equal (as with PrusaSlicer) because if your support lines will be 2mm apart, there's no point to having the slicer determine which areas need support with a greater accuracy than 2mm (since it will just be rounded up to the nearest 2mm once the support lines are actually drawn). But it does lead to the quirk that setting a smaller pattern spacing number (to get higher resolution, i.e. a support that confirms more closely to the shape of your part) also makes support lines closer, which in effect increases the support density, thus using more filament and taking more time. Another limitation is that you can't set the size of the raft under the support in PrusaSlicer (it's a fixed number that's baked in).

---

So, quick answer is, you reduce the "pattern spacing" in small increments until you stop getting the error. Your setting was 2mm, I tried decreasing in 0.1mm increments and fortunately stopped getting the error at 1.9mm, so in this case the impact is minimal and is more than made up for by the time saved switching to adaptive cubic infill (final count is 99.71g and 6h27m).

Here's the file with the above implemented:

---

There are other possible enhancements you can make for more speed/quality (these are a little more extreme):

  • Increase layer height to 0.25mm (this is a straight-sided print with most of the detail on the top, so there's no benefit to fine layer heights).
  • Reduce bottom/top to 4 layers (same thickness as 5 layers at 0.2mm)
  • Turn on avoid crossing perimeters (better quality, less chances of undesirable stringing)

This is with avoid crossing perimeters switched off (travel moves in blue):

image

Each blue line is a potential area where a string will form (if there's a drop of ooze it will be stretched out over that path)

This is with avoid crossing perimeters switched on:

image

Basically only one possible stringing line instead of dozens (and no impact on print time)

  • Increased speed of perimeters but reduced perimeter acceleration. This will improve speed since it's such a large print, while at the same time the reduced perimeter acceleration will improve quality and reduce ghosting.
  • Reduced lift Z to 0.3mm (slightly faster, slightly less wear-and-tear on the leadscrew).
  • Increased retraction speed to 70mm/s (faster, less oozing)
  • Had to reduce support pattern spacing further, to 1.8mm. In theory none of the above should have made this necessary, but might be just another unexplained quirk 🤔 

The results of this are a further reduction of print time to 5h22m (a savings of another hour over the previous enhanced version). Here's the .3mf:

You can leave it at that, or you could iron all top surfaces to make the part look much better (almost injection-molding quality) since the top is a giant flat surface that would really benefit from it. Beware though that ironing isn't perfect and in some cases might give a worse surface quality and there's a small chance of nozzle clogging due to the slow feedrate used. I found that the default 15% extrusion multiplier wasn't enough to totally fill in the pores of the surface, so I've increased it to 20%. I've also increased the pattern spacing to 0.2mm (up from 0.1mm) because with the default settings it brought the print time up to a whopping 9 hours. With the 0.2mm spacing it only brings print time up to 7h10m which is close to your original print time:

---

That about concludes it. I would try the files in the order above. If the slightly optimised one works well, try the highly optimised one (which is more cutting-edge so more chance of unforseen quality mishaps). Then if you want more quality at the expense of longer print time, try the ironing one.

Looking forward for your feedback 🙂

Regards,
Luke

ReplyQuote
Posted : 27/03/2021 10:39 pm
SusanOKC
(@susanokc)
Famed Member

Amazing, Luke! Thank you SSSOOOOO much! I will give these a shot and let you know how it goes.  Thanks again! 

ReplyQuote
Topic starter Posted : 27/03/2021 10:56 pm
SusanOKC
(@susanokc)
Famed Member

Okay, I used the settings of the final file you provided (highly optimized plus ironing) to run a small test print with just the logo. That's the print in red. You can see it's a little wobbled around the letters - not terrible but not perfect.  The purple is from a previous print - also not perfect but more crisp around the letters. Is there a way to dial that in on the ironed settings? Thanks! 

IMG 9477
IMG 9476
ReplyQuote
Topic starter Posted : 28/03/2021 2:56 am
Luke
 Luke
(@lmf5000)
Expert Moderator

Hi Susan,

Ironing will always smear the top surface a bit and make outer perimeters fuzzy. What might help to decrease the degree of fuzziness in your case is totone down the ironing flowrate (infill menu) back to the default of 15%. I had increased it to 20% because it gave better results on my printers but as every printer is calibrated differently some fine tuning will be needed. You might try an even smaller sample (like the top of a calib cube) at 10% and see which looks best. If unsure where the optimum lies, better to err on the side of slightly over than slightly under, because if it's under then some areas of a large print won't get ironed (not enough plastic) and the result will look terrible, a combination of smooth and rough patches.

Regards,
Luke

ReplyQuote
Posted : 28/03/2021 3:06 am
Luke
 Luke
(@lmf5000)
Expert Moderator

P.S. your purple print shows gaps between the top layers and the perimeters as well as between top layer lines. Try bumping up the perimeter overlap by 5% from its present value to improve that a bit. You also need to increase your extrusion multiplier, start by increasing it by 2% (so +0.02 in the PrusaSlicer box) and tune from there. If you still get gaps between top layer lines increase it further, and if it starts to look blobby, undefined or has lots of nozzle scratches, turn it back down slightly.

Regards,
Luke

ReplyQuote
Posted : 28/03/2021 3:10 am
SusanOKC
(@susanokc)
Famed Member

I'm running multiple tests now on the extrusion multiplier. Our calibration cube is exactly 20x20mm so I thought that meant our extrusion was good and we wouldn't want to increase it further because it could throw of the size of our prints?

Also, and I'm sorry to ask (another) silly question, but can you please tell me where to find the perimeter overlap setting in the Prusa slicer? I'm still finding my way around after switching from Simplify 3d.  I'm probably looking right at it without seeing it. 

Thanks!

ReplyQuote
Topic starter Posted : 29/03/2021 1:32 pm
Luke
 Luke
(@lmf5000)
Expert Moderator

Hi Susan,

That's a very good question. And the answer is yes, increasing extrusion multiplier will increase how much the extruder turns for a given amount of X/Y movement, which means the printing line will get slightly thicker throughout the print (that means top, bottom layers, infill and even perimeters). So your print will get slightly bigger on all sides.

If your calibration cube is 20.00mm at this point, increasing extrusion multiplier by 1% will dispense 1% more filament. Default extrusion width is 1.125x nozzle diameter = 0.45mm, so with 1% extra filament this grows to 0.45 x 1.01 = 0.4545mm. That's an expansion of 0.0045mm. Since it happens on the perimeters on both sides of the cube, your calliper will therefore measure 20+0.0045+0.0045= 20.009mm.

Worst case, the inner perimeter expands as well and squeezes the outer perimeter even further, so double it again to get 20.018mm. So as you can see the difference is quite slight for a 1% change in extrusion multiplier. You'll need a pretty significant increase to really start to affect dimensional accuracy, but you only need 3% or so to close the gaps in the top layer, so it should be possible to find a setting that just closes the gaps without making your prints expand enough to where things don't fit any more.

Now, the best approach depends on what you prioritise in a print 🙃 . If you absolutely need perfect dimensional accuracy then slight under-extrusion will be necessary because any slight areas of over-extrusion (which will happen randomly because 3D printing is an imperfect process) will cause oversize parts that won't mate properly. In my case my main priority is no gaps in the top layers as I feel they detract from the appearance, so I raise the extrusion multiplier in 1% increments until the gaps in the top layer just disappear. Then I add a slight tolerance of at least one nozzle diameter to all my designs that are supposed to be a press-fit. For example if a 5mm peg is supposed to be held by friction into a 5mm hole, I'd make the peg 4.6mm in diameter and the hole 5.4mm in diameter. If I want a sliding fit (like a hinge mechanism that moves freely), then I increase the clearance even further. If I were you I'd close the gaps then test how badly it affects precision and take it from there.

---

The overlap setting is under print settings -> advanced (make sure you're in expert mode):

image

Hope this helps! 🙂

Regards,
Luke

ReplyQuote
Posted : 30/03/2021 12:14 am
SusanOKC
(@susanokc)
Famed Member

Thank you, thank you! First the good news.

I found the perimeter overlap - what a relief, thanks!

The ironing at 15% is very beautiful - see the peach print Z. 

I can't tell that increasing the extrusion multiplier is really helping.  See the three red Zs - from left to right, 1.02, 1.04, 1.06.

The support settings for the first lid failed at two corners - see pink photos. Also, the top wasn't great but we're going to apply the ironing next and hopefully that will help. Any suggestions for which things to adjust next to get more support to the corners without extending beyond the size of the build plate?  (I printed your "Lid - slightly optimized" file from above, no changes.)

Thanks again! 

image4 2
IMG 9784
image0 2
image2 2
image1 2

 

ReplyQuote
Topic starter Posted : 30/03/2021 3:04 am
Luke
 Luke
(@lmf5000)
Expert Moderator

Hi Susan,

Ironing does indeed make for beautiful surfaces :). The peach print shows a slight hint of overextrusion, and your red cubes are all overextruded, that's why you're not seeing much changes as there are no gaps, the extra plastic is jut making the surface rougher where the nozzle digs into them. I think this is a filament consistency problem because your purple print from earlier was underextruded:

image

But your red one was fine. I might have missed it but was the purple one printed on the Prusa? If not then ignore my recommendation to increase EM 😆. If yes, I suggest is increasing extrusion multiplier only for the purple filament and using around 1.00 for the red one, as even your 1.02 in red looks a little over-extruded

image

---

With regards to the print, it's at the very edge of the bed so there's a chance the heat isn't quite reaching that corner. I didn't amend your cooling settings too much because I figured they might work, but seeing that you have lifting problems, I recommend this set of settings for your cooling tab:

image

Notice how minimum fan speed is now 50%, and more importantly there is zero fan for the first three layers. That will stop warping from the breeze caused by the cooling fan.

Also, you weren't getting supports all the way to the corners because of the region that was defined. For this I tried lots of different ways and found this one that worked - firstly I removed the paint-on support and defined the region by means of a support-enforcing block:

image

Then I played with the pattern spacing settings until the error of exceeding the printable area disappeared. Lower settings weren't working, so I went higher (I figured that at some point the spacing would get high enough that it would miss an entire line as the edge of the print would be beyond the threshhold for cramming another line of support there). The ultimate answer was 5mm. Here's the support settings:

image

I added an extra interface layer since the extra spacing is going to make the interface sag more. The extra layer should offset that a bit.

Here's the print:

image

There's support coverage all the way to the four corners and a bit of excess on the top and bottom which should make it resist warping a bit better too.

Here's the .3mf -

Hope this helps 🙂

If this goes well you can then recreate the highly-optimised settings by retracing the steps in my earlier description (keeping support and cooling settings from this one).

Regards,
Luke

ReplyQuote
Posted : 30/03/2021 1:16 pm
Luke
 Luke
(@lmf5000)
Expert Moderator

Edit: Here's a slightly updated one with additional support at the top of the lid (I enlarged the support box a bit so both top and bottom have 5mm of extra support) (one of the corners didn't have the extra support in the earlier version). Use this one instead of the above.

Before:

image

After:

image

And for even stronger supports, here's a version where I ticked "with sheath around the support":

image

Note that this makes support removal harder, so only use this one if the previous one still lifts off:

image

Regards,
Luke

ReplyQuote
Posted : 30/03/2021 1:25 pm
SusanOKC
(@susanokc)
Famed Member

Got it, thanks!  The purple was the Prusa but it was actually printed upside down (not sure that matters to you.) It was a crazy experiment, although actually turned out pretty well.

I will try the lid with Extra Support and will keep you posted.  Thanks again! 

ReplyQuote
Topic starter Posted : 01/04/2021 2:06 am
Share:
Share

About us

We’re a bunch of 3D printing and design nuts. We just want to make great 3D design available to everyone. 

Our mission

Good design takes time, great design takes a process. You can learn this with us, to build your skill set in this rapidly expanding market. 

Learn with us:

© Copyright: 2019 io3dp.com
Close Menu