Frustration: Hole inner dimensions all 0.5mm under size, help!

Hi all

Well I've built a fantastic kit from creative machines and its giving me lovely print quality aside from this one issue: hole size

I've been through all of the posts on the subject and followed all of the steps to calibrate. Including looking at nopheads polyholes thing. All of the external dimensions are dead on (using a manual vernier calliper to measure ). The more I've tuned the calibration, the more dead on 0.5mm under they are. I've tried with all the slicers and settled on Cura giving the best print quality so far. I just can't see any other settings to change that will give me the correct size internal holes. I must add that this is on all hole shapes and sizes and is always .5mm under and the outside dimension as dead on. Printing a plug and socket both 10mm od/id respectively, makes parts that will not fit, so I don't think this is shrinkage.

I really hope someone knows of a setting or something I've missed. when I make my own models I just add 0.5 to all holes but it makes it hard to print things from other sources.

Thanks in advance

Someone wrote a good posting on tolerances a while back but basically you should have SOME gap. If I'm designing a hole for a 3mm smooth rod, the diameter is usually going to be at least 3.4mm for a snug but smoothly sliding fit. If I have something that fits inside another piece I'll add between 0.2 - 0.25mm gap on BOTH sides.

However, 0.5mm sounds like a big reduction. One of the reasons for holes coming out a bit smaller is that the tension of the extruder head moving around in the (not exactly a) circle is that some of the filament gets pulled closer into the center of the circle based on its own surface tension.

I don't think there is any way around this. I put together a short (and relatively quick to print) channel connector piece here: [www.thingiverse.com]

I was trying to make a larger piece (this: [www.thingiverse.com] ) and found I needed a bit of trial and error to get not only the channel connectors to fit (even with filing - I really needed a smooth movement) but the machine screw holes and nut slots as well. My FreeCAD sources are not that pretty but you can see the gaps I ended up with.

I think there are also some calibration pieces specific to hole size. Your Mileage May Vary...

Have you tried the 'Stretch' module of Skeinforge ? I find it to get very close to the correct dimensions, I just activated it with the default parameters. Before activating it holes would be between 0.3 and 0.6mm too small, the error being bigger on the smaller holes. After activating it the holes are within 0.1mm of what they should be.

I do still however have an issue with external diameters of small parts so I have to undersize them in the design. Oddly, the larger the object the smaller the error. So foir instance a 100mm square might be 100.1mm but an 8mm diameter shaft will be about 8.5mm printed.

The 'Stretch' module does not seem to help here, and if I understand correctly it isn't meant to either (apparently it is only for internal perimeters).

Thanks for the feed back. I feel better knowing that this is something that other people are finding, and the designing for it is a correct way to deal with it. I've not tried the stretch module of Skeinforge, in fact I think Skeinforge is the only one I haven't tried at all.

I've been finding it odd that rather than the inner dimensions being out by some factor of the size of the part, they are almost always out by the same 0.5mm whatever the size of the part. My theory (possibly not the case) was that if it was to do with the way the plastic gets pulled around the curve, the a small 10mm plug should be under sized in the same way as a 10mm hole. I tested this and printed the two parts together. The plug was bang on 10mm OD but the hole...? yup, 9.5mm dead on. In my mind at least, this feels like a tool path offset problem from the slicer, but I find it hard to believe that they'd all have the same issue. A coincidence that my nozzle is set to 0.5mm??

I've also seen many people say they can print two parts with close tolerance and that they fit together, so this gives me hope that I've just made some noob error.

I'll carry on trying.

I do not model any clearance any longer. It is all setting dependent and the main thing that seams to cause this is layer height being too high and the combined width of the extrusion. You can try the calculator in my signature to get the settings I use. I would also not go over 0.2mm layer height.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

@sublime Thanks. This is one of the things I've seen mentioned a number of times. I've two reasons to think it isn't this. Firstly my settings follow exactly that. Layer height of 0.2, speed between 35 and 50. Second if that were the issue then I would expect to see the same error on the outer diameter of a 10mm cylinder plug printed at the same time. What do you think? Are the holes you pringt now dead on expected diameter?

@slarti42uk
Something you might want to check is the sequence of perimeters, loops and infill.

The worst sequence IMHO for correct dimensions (and also other reasons) is infill>loops>perimeters. If the perimeter is laid down last you will always get smaller holes and larger outer perimeters as the infill and loops will offset the perimeter away from them.

For best dimesnional accuracy it is best to lay down first the perimeter then infill and finally the loops as this allows the perimter to be laid down without interference from the loops and infill and will also allow most time for the perimeter to cool down and solidify, preventing it from being pushed aside by the loops.

However that can cause problems with the infill not attaching properly the loops so I find the best compromise to be Perimeter>loops>infill. This at least allows the perimeter to be laid undisturbed (it stays exactly where the nozzle puts it) while at the smae time providing the end points for the infill to attach to. Skeinforge always lays the loops such that the one touching the perimeter is the last one. This helps a little to allow the perimeter to solidify somewhat so it will not be pushed too much when the loop touching it is laid down.

The other thing which seems to help (for dimensional accuracy) is slowing down the feed rate of the perimeters (and the flow rate accordingly to maintain the same flow/feed ratio). The slower the filament is laid down the more accurately it will be positioned - it will be pulled around less. However apart from the fact that it takes longer to print I found that slowing down too much introduces new print quality problems. I cannot understand why as logically I would think that the slower the material is deposited the better it should be.

slarti42uk Wrote:
-------------------------------------------------------
> @sublime Thanks. This is one of the things I've
> seen mentioned a number of times. I've two reasons
> to think it isn't this. Firstly my settings follow
> exactly that. Layer height of 0.2, speed between
> 35 and 50. Second if that were the issue then I
> would expect to see the same error on the outer
> diameter of a 10mm cylinder plug printed at the
> same time. What do you think? Are the holes you
> pringt now dead on expected diameter?


This is Nopheads test part from his Mendel 90 repo printed on Tantillus with Kisslicer which Skeinforge users say gives the wrong dimensions but I get them perfect without any special settings. [www.youtube.com] you can also download my profiles for Tantillus from Tantillus.org downloads page if you would like to see all the settings together.

I think what the issue is is that with the larger layers and thick extrusions it stays too molten which causes the plastic to be able to move and shrink into the holes a little. But with the lower layers and lots of cooling you can freeze the filament exactly where it gets laid down.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Ok, I've printed out the polyhole test piece from nophead Thingiverse 6118 and every hole is 0.5mm under size. I really don't know where else to go with this. Trying all of the suggestions just seem to affect print quality but still not improve the internal dimensions of holes. I've tweaked the E steps wile printing but it just ends up with the extruded trace not bonding to the next one.

Think I'm going to design a tool holder to hold a 0.5 propelling pencil and print some layers and see what's going on with the actual tool path. Not sure if that'll help or not.

I wish the outer diameters were off too, then I could see changes to settings working, but they are all spot on. Are there any other settings in firmware (Marlin) that would alter ID but not OD?

Thanks for all the suggestions. I may give Skeinforge a go but I really do want to get this working with Cura, as it's given me the best results for print quality and it's nice to use.

I know you are printing with a layer height of 0.2mm but you never mention the width of the extrusion. So what extrusion width are you printing at? (note Cura has a strange way of doing this and you need to do the math and figure out the width you want to extrude multiplied by the number of perimeters you want and then enter that as the wall thickness. This has nothing to do with flow rate and E steps per mm.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

@Sublime: I have:

* Shell thickness - 1mm (to get two loops)
* Filament: Diameter - 3mm
* Flow: 100%
* Nozzle size - 0.5mm

I can't see anywhere in Cura to set fillament width. I've been working on the basis that if I have an 0.5 nozzle value and shell thickness of 1mm, it will work that out to 2 loops? maybe?
I did try the thin wall calibration thing but when I adjusted that I just ended up with massive holes in the top and infill.

"I did try the thin wall calibration thing but when I adjusted that I just ended up with massive holes in the top and infill."

There should be separate settings for perimeter and infill. You should be able to adjust one without affecting the other. I'm not familiar with Cura though so I can't help much with finding where the approriate settings are.

Yeah Cura has no explicit way of setting the extrusion width other than the shell thickness divided by the nozzle to determine total loops. Then it takes the Shell thickness and divides it by the previous number to come up with the actual extrusion width. None of it makes sense to me and can only result in poor extrusion sometimes as it will let it print as much as 50% of the nozzle diameter in width and as much as 150%. I would set Cura to have a shell thickness that is a multiple of 0.57mm extrusion width (as calculated by the Tantillus Calculator for a 0.5mm non-J-head nozzle and 0.2mm layers). So for two loops you would set it at 1.14mm. You would then set the speed to 50 - 55mm/s and make sure you have a fan cooling the part to freeze the plastic in place.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Quote
slarti42uk
* Filament: Diameter - 3mm

Is that the value you set in Cura? No filament on the market is exactly 3mm in diameter. You need to take multiple measures along some length of your filament and make an average. Otherwise this throws off the real amount of plastic that gets extruded.

My own setting (in Slic3r) is 2.89mm for Lulzbot 3mm filament.

@Sublime, thanks for all the info I'm going to try what you suggest about the wall thickness as a factor of the extruded filament size. Also I'm not sure if I stated already but I'm priting with ABS. Would you still recommend a fan? I found with one initial attempt with a fan that I have problems getting the filament to bond to the next layer.

Also I'm using a genuine Jhead so I'm going to use the settings from the calculator you suggest.

@NormandC, Yes that is the setting in Cura. I've been pretty lucky with the first roll of filament I have. I've repeatedly check and it is 3mm bang on. I am measuring with a manual vernier scale calliper. I don't have a micrometer but this is a pretty good callipers.

Thanks everyone for all the comments. I really want to try and crack this so I can go on to making parts for a delta printer.

slarti42uk Wrote:
-------------------------------------------------------
> @Sublime, thanks for all the info I'm going to try
> what you suggest about the wall thickness as a
> factor of the extruded filament size. Also I'm not
> sure if I stated already but I'm priting with ABS.
> Would you still recommend a fan? I found with one
> initial attempt with a fan that I have problems
> getting the filament to bond to the next layer.

No the fan will most likely cause it to warp and crack. I would print a little slower and it has been suggested that you need to scale up ABS prints by a small factor. I think it is about 1/2 of 1 percent that is recommended to compensate for the ABS shrinkage. Also you may want to measure your die swell (the difference between the diameter of filament that is extruded and your nozzle size) and enter it into the calculator to get a more accurate extrusion width for your material.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Well I did start to feel like I was making possible progress. I measured the filament extruded in free space and it came out it 0.7mm so I set the nozzle and wall thickness to this and did another test. The inner diameter came out nearer 0.2 undersized but now the outer diameter is also about .2 undersized rather than spot on. If this is the case the as @Sublime suggests I could scale the piece up a fraction. The only problem is that in my hurry to warm the bed to print another test (with a heat gun I've been using) this happened...