Undersized holes

[upsm]

There is a very simple reason why holes can come out undersized. Over-extrusion.

In the whole thread the assumption is that machine is properly calibrated. Without proper calibration nothing else makes sense

[carveone]

I have a similar issue with slight hole reduction on my Cubicon Single that I'm trying to configure out. In doing this I've outputted two calibration parts with various hole sizes - one sliced with Simplify3D, the other with Cubicreator. Interestingly, the hole size is accurate with Cubicreator - with a 3.2mm hole I can just about put an M3 bolt through; it will not fit under Simplify3D.

I decided to analyse the gcode outputs of both pieces of software and render them over one another (without the infill). Taking the second layer I find no differences at all in path output (picture apppended). The path order is different in that Cubicon does all the inner paths first and then the outer wall paths while Simplify3D does inner-outer-inner-outer...etc. Not sure if that would make a difference. Cooling times perhaps.

Taking a 3.0mm hole as a path example, I can see the diameter of the outer shell is 3.4mm (which I would expect for a 0.4mm extrusion diameter). The sum of the extruded output (the sum of the "E" numbers for that path) is 3532 units (I multiplied by 10000 to get an integer) for Cubicreator and 3531 units for Simplify3D. So the extrusion length is the same for that shell at least. Checking the inner shell, it's 4371 units for both.

So, the paths are the same, the extrusion length appears to be the same yet the output result is different! I haven't read through this entire thread yet but I'll try to see what other settings there are. All I can see so far is fan speed which, for the default Simplify3D profile for the Cubicon, is set to 0. I'm not convinced it is set to 0 on the Cubicreator software. There's also printing speeds and retraction which are different between the two...

[billyd]

There is a very simple reason why holes can come out undersized. Over-extrusion.

In the whole thread the assumption is that machine is properly calibrated. Without proper calibration nothing else makes sense

If your printer is printing out wrong sized holes while properly calibrated then you need a new printer.

[carveone]

I have a similar issue with slight hole reduction on my Cubicon Single that I'm trying to configure out.

First I should have mentioned that I'm printing with ABS. Silly thing to miss.

I adjusted Simplify3Ds settings from the Hyvision Single profile to slow the feed down slightly and turn on the fan to 50% on the second layer. I also had to change the ending script to turn the fan off (Cubicreator uses the older "M107" but "M106 S0" is fine) because Simplify3D decides to leave the fan on!

I think this is a bug in Simplify3D - it's not useful to leave the fan on with the extruder off as the tip cools very quickly and isn't helpful for a second print.

Anyway, for me that fixes the problem. There is a tiny difference because Simplify3D squashes the first layer down by raising the Z by 80% on the first layer. Which I could fix for any particular print but I'm not bothered generally as bed adherence is more important there. Aside from that, the hole sizes are correct. I believe you are supposed to leave the fan off generally for ABS because of bed adherence but this works for me.

[upsm]

@carveone, so with cubicon slicer your holes are ok, with Simplify3D they are undersized? but when you render the g-code it looks same?

@billyd, same printer, different (closed source) slicer - holes are ok. even when you underextrude, the holes are undersized with Simplify3D, same printer, difference in slicer only.

[carveone]

@carveone, so with cubicon slicer your holes are ok, with Simplify3D they are undersized? but when you render the g-code it looks same?.

Yes, exactly. But adding some fan cooling helps in this regard and is what the Cubicon supplied software does by default. It's a marginal difference and one I could compensate for in CAD (assuming I used a variable for hole size which hasn't always been the case, but will be in future!). Where it mainly hurts me is in small parts like gears, so I can use a 'fan on' profile for those where the first layer is also 100% z height.

To be honest I'm not sure why I shouldn't always have the fan at 50% with ABS. I guess with some printers, curling, warping and delamination is a reason not to, but the Cubicon has a heated chamber and build plate so perhaps that makes it ok. A lot of trial and error with these printer things!

Conor.

[MasterFX]

My holes are to small too, while all other mesurements are perfectly on spot.
But this is definitively not a error or fault by S3D. You just have to look in the generated Gcode file.
For example a 10x10mm square with a 4mm hole. The nozzle diameter is set manually to 0.4mm to get even numbers in the coordinates:
Outer perimeter of the square

Code: Select all

; outer perimeter
G1 X4.800 Y-4.800 F4200
G1 X4.800 Y4.800 E0.5365 F743
G1 X-4.800 Y4.800 E0.5365
G1 X-4.800 Y-4.800 E0.5365
G1 X4.800 Y-4.800 E0.5365

And here is the outer perimeter of the circle (only selected the min and max X/Y values for better overview):

Code: Select all

...
G1 X0.084 Y-2.199 E0.0255
...
G1 X-2.195 Y0.150 E0.0174
...
G1 X-0.084 Y2.199 E0.0255
...
G1 X2.195 Y0.158 E0.0257

So the line of the outer perimeter of the circle is drawn with a radius of 2.2mm, so that with an extrusion width of 0.4mm this should result in a very exact 4mm circle. So Slicer-wise everthing is fine as expected.

[dkightley]

I'm not wishing to be controversial .....but I would like to offer a pragmatic view of the whole issue of sizing for holes that I think is fairly unique for fff manufacturing....although this may be similar for some cnc machining.........

When I design a part in a 3D package and need a hole...say 4mm diameter....I boolean subtract an over-long 4mm diameter rod from the part, and I'm assuming that this is the norm for most creators.

Now...this 4mm rod is in fact a polygonal cylinder with a number of straight sides...the number of sides usually being defined by the size of the "hole" being cut so as to ensure the corners are not visible. Make the number of sides super high...say 256..and the hole will be super smooth to the eye. But lets say we have a hundred holes in this part....and making a hundred 256 sided holes will push the poly count through the roof...and in the case of my 3D package, kill it!

So, I drop the number of sides on my 100 rods down to a sensible figure of 8. The holes are for bolts to pass through....so the octagonal shape is not a problem. I 3D print the part.....and get what everyone here is going on about...I can't get a M4 bolt in the holes!

Tish!! I forgot to make the holes slightly bigger to allow a non-interference fit. Up the hole size to 4.1mm and re-print. I do so....and the M4 bolts still won't fit! It must be the slicer software at fault 'cause the holes are too small!! Well.....perhaps not so......

Let's look a little deeper at my 4mm diameter holes. They're octagons.....cut out by a 4mm OD octagon. Take a look at this image:

Ignore the two octagons on the right for a minute. the three shapes on the left are polygons...starting on the left and going clockwise...of 64, 16 and 8 sides. Inside the 8-sided shape is a polygon with 256 sides....as close to a true circle as you could realistically get...its inside the octagon and its OD is 3.65mm, but its almost touching!! A similar polygon fitting inside the 16-sided hole would be 3.88mm diameter.

So...going back to the above example with holes for M4 bolts, the octagonal holes wild need to be cut with a rod of something like 4.2mm OD for the purposes of fff manufacturing to allow a non-interference fit. And that's just the allowance needed at the CAD stage.

It has been proven in theory by previous posters that the centre of the extruded filament around a hole is half a filament away from the edge.....but now look at the upper of the two octagonal shapes to the right of the above image. Let one square be 1mm x 1mm....and you'll see that you'll be able to fit a 4mm shaft in the hole giving a close fit. The bottom shape is exactly the same size as the top one....but rotated round 22.5 degrees. Top to bottom & left to right you can see the OD of the octagonal hole is well over 4mm...which illustrates what I've said above.

But notice how the extrusion forming the sides "fit" against each other. There's a gap on the outside and an overlap on the inside. Its anybody's guess how the plastic will extrude at this interface...but I think the hole would be made smaller at these intersections...not enough in this example to reduce the overall size sufficiently to cause interference....but possibly so where there are more sides to the hole. And then there's the effect of slight over-extrusion...

So...to close out my thoughts....the CAD drawn hole may be not allow as big a shaft to rotate in it as you may think.....and the fff manufacturing may close the hole slightly just due to the fact the hole is a polygon. Put the two together.....and the result is seen ...well...as a bug, a symptom, a deficiency...whatever you want to call it. But the bottom line is that there has to be allowance made to get a shaft, etc to fit freely in a hole of the same nominal size. Whether this is done in the CAD stage or the printing/slicing stage, I make no comment or suggestion.

And for those thinking "Why doesn't this affect cnc machining?".....a 4mm hole would be drilled and not machined out by small cutter following a polygonal path.

[upsm]

I'm not wishing to be controversial

I doubt anyone participating in this thread care about being "controversial" and are just willing to invest time and thought into getting the issue solved. Confronting different opinions is usually best way to get there.

When I design a part in a 3D package and need a hole...say 4mm diameter....I boolean subtract an over-long 4mm diameter rod from the part, and I'm assuming that this is the norm for most creators.

Now...this 4mm rod is in fact a polygonal cylinder with a number of straight sides...the number of sides usually being defined by the size of the "hole" being cut so as to ensure the corners are not visible. Make the number of sides super high...say 256..and the hole will be super smooth to the eye. But lets say we have a hundred holes in this part....and making a hundred 256 sided holes will push the poly count through the roof...and in the case of my 3D package, kill it!

Depends on the CSG your 3D package uses. If you are working on "solids" and not on "mesh" then until the end, when you are exporting your part to a mesh, you have a cylinder there. And while few packages (like openSCAD) give you ability to define "crudeness" of the mesh per surface, most will (like SolidWorks, Creo, proEngineer, onShape) generate mesh following 2 input configuration parameters: maximum surface deviation (how far will mesh surface be from the designed surface), maximum angle tolerance.

Since the "coarse" settings in most app go below 0.1mm (SolidWorks is 0.07mm for coarse for example) the issue that polygons are cutting in and shortening the hole makes very little sense to me.

The research Chris made on this, that I linked, tests this theory pretty thoroughly. Here's the link again: http://hydraraptor.blogspot.rs/2011/02/polyholes.html

But notice how the extrusion forming the sides "fit" against each other. There's a gap on the outside and an overlap on the inside. Its anybody's guess how the plastic will extrude at this interface...but I think the hole would be made smaller at these intersections...not enough in this example to reduce the overall size sufficiently to cause interference....but possibly so where there are more sides to the hole. And then there's the effect of slight over-extrusion...

That is what Adrian is talking about arc compensation. You are depositing more plastic inside the circle then outside. The "finer" the circle is, the more this is pronounced and closer the hole will be. And if you print 4 perimeters, every perimeter is overflowing towards insides. Then, in my opinion, this joins the forces with "corner cutting" and you have holes that are significantly smaller then designed.

What I think is a shame is "Its anybody's guess how the plastic will extrude at this interface" since fluid dynamic is not that impossible to calculate. This is a problem that's around for a long time. There has to be a math explaining how exactly extrudate is behaving in this case. I don't know how, but we can see that big players solved that problem while back (send your part to be printed on Stratasys and without any post processing, drilling etc. your holes will be as you designed them, printed properly)

And for those thinking "Why doesn't this affect cnc machining?".....a 4mm hole would be drilled and not machined out by small cutter following a polygonal path.

cnc machining is affected by many things 3d printing is not, and many of them are way more complicated then stuff we deal with. "small" cnc machine deals with tens or hundreds of kilograms of mass being moved at high speeds, and they solved their problems without deciding "it's impossible, design your part around the problem". I don't think "smaller holes" are in any way a priority for Simplify3D, there's ton of things that are way more important, and easier, and faster, to solve. I just have a serious problem when ppl say "it can't be done".