negative spaces

[JFettig]

I like to design and build functional parts, parts that fit together with printed parts and non printed parts.

I tend to find that holes and internal features are always small. How do you guys like to deal with that?

Printer is well calibrated for size, for this part I'm using PETG which doesn't shrink considerably.

My STL files are extremely accurate and have very dense mesh so its not that.

I did a quick print, measure, comp, re-print test - that worked but there is going to be more to it than that.

Anyone have any pointers?

[Polygonhell]

Is it just round holes you are having issues with or all internal dimensions?

[JFettig]

Seems like all internal dimensions, I guess I haven't really taken good measurements of non-round holes.

[Polygonhell]

That's where I would start, round holes will generally come out undersized for a number of reasons, all you can really do is understand how much you need to adjust them by for a given size, filament and printing temperature.
There are several contributing factors, it's not clear which are dominant, the modeling packages tend to do the wrong thing when they tesselate to create the STL file, ie they sample the points and build an inscribed polygon which is already guaranteed to be smaller than the requested hole size. The filament is stretched out of the nozzle and tends to under cut interior corners slightly, and of course there is plastic shrinkage.
Do a search for Nophead poly holes, he discusses the phenomenon pretty extensively in one of his old blog posts. It's been discussed on here several times too.

[JFettig]

That's where I would start, round holes will generally come out undersized for a number of reasons, all you can really do is understand how much you need to adjust them by for a given size, filament and printing temperature.
There are several contributing factors, it's not clear which are dominant, the modeling packages tend to do the wrong thing when they tesselate to create the STL file, ie they sample the points and build an inscribed polygon which is already guaranteed to be smaller than the requested hole size. The filament is stretched out of the nozzle and tends to under cut interior corners slightly, and of course there is plastic shrinkage.
Do a search for Nophead poly holes, he discusses the phenomenon pretty extensively in one of his old blog posts. It's been discussed on here several times too.

Like I said in the first post, the polygon deviation isn't an issue, its set to .0001" in Solidworks so there are thousands of sides per hole.

I'll try printing something with a square hole in the middle to see how it does.

[Polygonhell]

In terms of what I do if I want a particular fit for a mechanical piece, I print test pieces to dial the fit in, but I can usually get pretty close first time.

The prusa mendel printer came with a set of test print pieces, you were supposed to adjust your printer to get the right fit on those before printing the plate for the printer.

[Polygonhell]

That's where I would start, round holes will generally come out undersized for a number of reasons, all you can really do is understand how much you need to adjust them by for a given size, filament and printing temperature.
There are several contributing factors, it's not clear which are dominant, the modeling packages tend to do the wrong thing when they tesselate to create the STL file, ie they sample the points and build an inscribed polygon which is already guaranteed to be smaller than the requested hole size. The filament is stretched out of the nozzle and tends to under cut interior corners slightly, and of course there is plastic shrinkage.
Do a search for Nophead poly holes, he discusses the phenomenon pretty extensively in one of his old blog posts. It's been discussed on here several times too.

Like I said in the first post, the polygon deviation isn't an issue, its set to .0001" in Solidworks so there are thousands of sides per hole.

I'll try printing something with a square hole in the middle to see how it does.

You should be aware that slicers commonly quantize the coordinates and remove degeneracies which will simplify geometry, leading to shrinkage even if you have thousands of sides in the original model.
The only way to verify is to look at the output Gcode and see how many different motions it generated.

Simplify 3D is particularly bad in this regard, it completely removes fine detail that ends up still present with other slicers.

[JFettig]

Good to know - and that wouldn't surprise me. I am using S3D.

Square hole is a couple thou larger than round. Still small.

It measures .466-.467" where the 25mm outside square measures .984" I will now print a 12mm boss on top of that square rather than a hole.

[KAS]

I'm seeing the issues with circles in a screwdriver holder I'm making while the external measures correctly.

An nice feature to have in a slicer would be a choice of Circumscribed or Inscribed polygons. That's if we're stuck with polygonal circles..

[JFettig]

It would also be nice if slicers had the option to offset internal spaces to a given dimension or scale too.

I printed an internal and external square, they actually fit together quite nicely. Same 12mm dim on both.

[Polygonhell]

I'm seeing the issues with circles in a screwdriver holder I'm making while the external measures correctly.

An nice feature to have in a slicer would be a choice of Circumscribed or Inscribed polygons. That's if we're stuck with polygonal circles..

That particular problem is in the conversion to the STL file, and not in the Slicer, to "fix" it a slicer would have to be able to ascertain that the original model had a circle there, if it could do that it should just emit G2 and G3 codes instead of G1's

[Polygonhell]

It would also be nice if slicers had the option to offset internal spaces to a given dimension or scale too.

I printed an internal and external square, they actually fit together quite nicely. Same 12mm dim on both.

KissSlicer has this option, but it affects both inside and outside edges, i.e you specify how much bigger internal holes should be and how much smaller the outside should be.
I've used it occasionally, but for the most part I bake the tolerances into the model.

[Polygonhell]

It would also be nice if slicers had the option to offset internal spaces to a given dimension or scale too.

I printed an internal and external square, they actually fit together quite nicely. Same 12mm dim on both.

KissSlicer has this option, but it affects both inside and outside edges, i.e you specify how much bigger internal holes should be and how much smaller the outside should be.
I've used it occasionally, but for the most part I bake the tolerances into the model.

[JFettig]

S3D also does, but like KISSlicer - it makes the outside smaller as well.

[KAS]

Is it common to just over size the circle till it fits the intended design? I've tried MatterSlice, Simplify3D, Kiss and Cura, they all pretty much make this part the exact same size.

The both rectangles are close at 21.15 but the circle is always under. 21.21 is the goal.

XYZ_nickel_Calibration_2.stl

(38.44 KiB) Downloaded 184 times

[JFettig]

So weird. I just printed identical parts in PETG and ABS, the PETG parts have perfect sized negative spaces, the ABS parts are -.025"! This is on a 1.880" circle. The mating part(OD of 1.875") is spot on on both situations.

[Polygonhell]

So weird. I just printed identical parts in PETG and ABS, the PETG parts have perfect sized negative spaces, the ABS parts are -.025"! This is on a 1.880" circle. The mating part(OD of 1.875") is spot on on both situations.

you should see about 0.6% shrinkage on ABS, I can't remember where I got that number from, it will affect interior edges dramatically more than exterior, because there is nothing inside them to limit the shrinkage.

[JFettig]

That has to be it. Some day hopefully we'll have slicers capable of compensating for such errors!

[Polygonhell]

I think Slicer at one point did this, it was not a well liked feature and was removed.
There was some debate about features like this on the reprap forums a couple of years ago. The problem is there are a lot of factors, different printers and plastics tend to behave differently, and you can't build something that just works, so now you end up with this guessing game where you try and guess what the error in the Slicers correction is going to be. If your going to accomodate error anyway, you may as well just account for it in the original model.

It should be noted a lot of the parts people complain about sizes are using models from thingiverse, they commonly have no tolerances specified, i.e. whoever made them seems to think an 8mm rod will fit in an 8mm hole.

[JFettig]

I agree and disagree with accomodating these issues in the models - A. you can't always accomodate - especially when printing in different materials. B. most of the parts I design and prototype are just that, prototypes for a part that will get machined, cast, or injection molded later. I'd hate to have to go in and make a guess at what each feature will do, then make a design change, only to have to change it back later.

Hopefully one day there will be a professional level slicer that will be as capable as CAM software is for CNC mills. Most basic CAM software has an option to leave stock, positive or negative(cut a hole larger than specified if negative). All slicers treat the whole model(or section of the model) as one unit and all settings are applied across the board as-is.

To date, I think I have printed 3 or 4 things I have downloaded off the internet. All trinkets for my wife. Everything else is an original creation I designed myself.

[Polygonhell]

The way I do prototypes (I use Solidworks) I set tolerances for clearances as variables, and just change them when I do the CAM work.
It might be worth asking Jonathan to make the KissSlicer setting asymmetric, it's clearly designed to accommodate exactly this issue, but it affects both inside and outside edges identically currently.
This is the current KissSlicer forum I think http://www.kisslicertalk.com/