Slicer scaling issues

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ddmckee54

Well-Known Member
HMEM Supporting Member
Joined
Jul 8, 2009
Messages
760
Reaction score
234
I've been using Simplify3D as my slicer for a number of years now, mostly with good results - until recently.... I started having issues with internal bores and pockets not being dimensionally accurate - they all wind up too small.

When I first started 3D printing, 8-10 years ago (Or more?), I was using a kit-built printer and learning as I went - by trial and error. (Mostly error.) I started out printing junk off Thingiverse and very quickly discovered that I couldn't find EXACTLY what I wanted. At this time I was using Slic3r for my slicing software - the price was right. But I NEEDED to learn 3D CAD. Talk about going down one rabbit hole only to find you need to go down another even deeper rabbit hole to find your way out. I tried Fusion 360, back when is really was free AND powerful - and absolutely hated it. I just could not wrap my head around the parametric modelling. Then I found Designspark3D with its' direct modelling. With enough poking and prodding I could get it to follow my commands/wishes. What was I designing? Upgrade parts for my 3D printer of course. In the course of these upgrades I decided that I could no longer tolerate the rat's nest of a wiring job that I had done in my haste to see my printer move under it's own power. One of these upgrades was an enclosure for all the electronics, gotta keep the angry pixies confined ya know.

In the process of printing the enclosure, let's just say things went - awry? Part of the enclosure was an opening for a 40mm cooling fan. To say the opening was a little wonky is an under-statement. It was visually an ellipse. NOTHING matched the model's dimensions. Not knowing any better I assumed that my cheap Chinese clone of an I3's steps/mm settings were wrong. I found a way that I could, by brute-force, change the steps/mm settings through the g-code. I starting printing 20mm test cubes - LOTS of 20mm cubes. I eventually got to the point where I could print a 20mm cube that was dimensionally accurate to within a few hundredths of a mm. HUZZAAH!

Fast-forward a couple of years and I was watching a video by James, of Clough42 fame, about issues he was having with 3D prints not being dimensionally accurate. He remarked that this was caused by the thermal expansion. This is the same reason that pattern makers in foundries have used rulers that are scaled over-sized for many-many years, to compensate for the shrinkage. Talk about a DOH moment, what I did with my brute-force method was doing the exact same thing - compensating for shrinkage. He was also using Simplify3D and showed how to use the scaling in S3D to scale the model for printing a dimensionally accurate part. For several years that has served me fairly well. I had problems with hole sizes, but I always wrote that off as part of the inaccuracies of FDM printing.

I'm working on parts for RC truck models now and I'm trying to print bearing pockets - so I NEED an accurate fit. I scaled the model to 103% in the slicer and the external dimensions was about as spot-on as you are going to get with an FDM printer. BUT, all of the internal bores and pockets that I could reasonably measure were about 97% of the model's size. I think what S3D does when it scales the STL is to offset to the outside of the outer perimeter by that scaling factor. That would explain why the external dimensions were OK, but the internal dimensions were all undersized. This leads me to the conclusion that I can't use S3D to scale model. I don't know about Slic3r, or the Prusa Slic3r, or Cura, but I would guess that they all behave similarly.

So I tried to scale a part in my 3D CAD, just to see how it behaved. I just whipped up a simple cube with a couple of holes going through it, and scaled it to 103%. Everything worked as expected, what was initially a 10mm bore measured 10.3mm, and the 20mm cube measured 20.6mm.

The moral of this rather long-winded story? For a dimensionally accurate 3D FDM printed part... scale your model in CAD, not your slicer.

Don
 
I would probably just adjust my tolerances to match the printer. ;)
For sure if I scale a part with inside bores to 2-times the size, everything becomes bigger including bores.

scaledblock.jpg
Scaling in Cura scales everything, but holes tend to end up undersize. I would just re-work the holes in the print with the usual methods.
 
Not everything scales, and if you have any compenstation for holes in the unscaled part, that might well change when you scale. The nozzle does not scale, so you need a different compensation when you scale the part. A smaller part will retain more heat that interacts with the next extrusion pass differently that on a larger part which has more time to cool before the next pass.
 
I have not heard of this issue before.
I will have to do some experimentation.
Generally I close holes in patterns when I 3D print them.
.
 
Timo & kf2qd:
I agree with what you're saying. Simplify also scales uniformly, the dimensions of all the external perimeters are dimensionally accurate after printing. However, all the internal pockets and any bores are too small, which you both have seen. And, as close as I can measure, they are too small by the shrinkage factor. Which is what my scaling to 103% should have corrected. I think this is a slicer issue, and from what you said Timo it's an issue with Cura also. I tried an experiment last night with my CAD, similar to the example you showed Timo. I tried a couple of the same nominal ID bores as I originally was using on other parts. I then scaled the part in CAD to 103%, the dimensions of the bores matched the tweaked values that I had to manually enter. I need to print another dual wheel for the RC truck conversion project I'm working on, so tonight I'm going to change the dimensions in CAD back to their original values - there's only a couple in this part that are critical. Then I'll scale the part to 103% in CAD and export it to STL. If I print that STL at 100% that should tell me if my theory holds water. Or whether, like the 6 million dollar man, my brain's just full of cotton and sand and squirrel-stuff.

GreenTwin:
For years I've been lurking on the casting forums, and collecting the equipment to allow me to cast model parts. I'm not quite there yet, but I'm close. Until then I'll just use the output of my plastic-pooping robot directly. That way I can de-bug my designs with a lower investment of my time. On the plus side, this is also pointing out potential gotchyas and trouble spots that I need to look for when I do start casting.

Don
 
For 3D printing patterns, I have been scaling the part in the Prusa slicer, and have not yet noticed any problems.
I add the appropriate shrinkage factor (a scale factor of about 1.015), add machining allowances to the surfaces that will be machined, and make sure I have draft angle in the direction that the pattern will be pulled from the sand.

And I close all small holes in the pattern, since those would only show up if I used a lost wax or lost PLA process.
I use a resin-bound sand process, and so the various holes for bolts and such are drilled after casting.
Since one does not know the exact amount that any give casting will srhink, it is best to accurately lay out holes after casting the part.

I try to gate into the major part of the casting, and in a location that has to be machined, so there is no trace of the gate(s) after machining.

Making my own castings has been one of the more interesting things I have ever done.
It can be challenging, especially if you don't have anyone to give guidance when things go wrong.
I have studied backyard casting methods and made backyard castings for the last 12 years, and I have a good feel for how to make aluminum and gray iron castings. I am glad to share any casting info that I have.
.
 
Last edited:
GreenTwin:

The current plan is to use investment casting for the jewelry sized stuff, and some type of bound sand for the larger stuff. Athough.... Foketry seems to have a reasonable amount of success with his sand/casting-plaster Lost PLA method. I've often wondered if using an actual investment plaster instead of the sand/casting-plaster would improve his results? When I get to that point I may give it a try.

Don
 
When I started to learn how to make castings, I tried Petrobond (tm), which is an oil-based foundry sand, and had some success with it, but I found it did not stand up to iron temperatures very well.
I also tried traditional water-based greensand, and it resulted in the worst surface finish I have ever seen.

I also tried sodium silicate bound OK85 foundry sand, that that actually works pretty well, but required CO2 to cure, and getting the entire mold cured evenly could be tricky.
And finally I found resin binder, and I use that with OK85 foundry sand, and a sprayed on ceramic mold coat, and I can get commercial-grade quality results with these materials.

Finding and correctly molding/gating a good sand for molds is probably the most difficult thing to resolve in the backyard casting world.
Melting metal is relatively easy compared to making a good sand mold.

My philosophy is to use the most simple and least expensive method that gives consistent high-quality results.
I don't like any defects in my castings, and I use ferrosilicon to avoid hard spots in gray iron castings.
For me, I don't save any money by having to cast the same part multiple times trying to get a good casting.
I carefully lay out the sprue/runners/gates/risers, etc. to be sure that every casting is a high quality casting on the first try.

On a small scale, you could try regular plaster vs investment plaster, and see what sort of results you get.

.
 
Last edited:
Jumping into the 3D printing world was a total learning curve for me. I am proficient with Solidworks so creating the models wasn't a problem. It was learning the intricacies of printing itself. I use Cura for my slicing and it seems to work fine even when scaling. That's not to say it would make a hole within. 002 or .003 but pretty accurate. The biggest problem I have with some holes is the Z step. The slicer moves Z at the same spot for each layer and this creates a small ridge.
I modeled my Ford 302 at 30% of full size but didn't want to print it that size but rather at 25% of full size. This gave me a scaling factor of .833, so in Cura I used the number 83 and I must say that with only a little filing here and there the whole thing fit together beautifully. I have 2 Creality printers, a CR 10 V2 and an Endeŕ 3 Neo. Not the best in the printing world but then they are my side line hobby.
 
Jumping into the 3D printing world was a total learning curve for me. I am proficient with Solidworks so creating the models wasn't a problem. It was learning the intricacies of printing itself. I use Cura for my slicing and it seems to work fine even when scaling. That's not to say it would make a hole within. 002 or .003 but pretty accurate. The biggest problem I have with some holes is the Z step. The slicer moves Z at the same spot for each layer and this creates a small ridge.
I modeled my Ford 302 at 30% of full size but didn't want to print it that size but rather at 25% of full size. This gave me a scaling factor of .833, so in Cura I used the number 83 and I must say that with only a little filing here and there the whole thing fit together beautifully. I have 2 Creality printers, a CR 10 V2 and an Endeŕ 3 Neo. Not the best in the printing world but then they are my side line hobby.
On cura there is a paramter z Seam Alignment, if you change it to random the vertical seam in the bore should go away. If that creates other issues I do not know. :cool:
startpoint.jpg
 
Last night I re-printed the rear wheel, with the bores corrected back to their original values, and the part scaled to 103% in CAD. It ALMOST worked like a charm. I corrected the 12mm hex pocket back to 12mm, but I forgot to add any clearance Clarence! What I've got now is very much an interference fit with the 12mm hex driver, that's been fixed and it's printing again now.
 
Back
Top