dimension corruption?

[grNadpa 2]

The object printed on my Monoprice select mini V2 printer has different diameters than my design specs. For example, when I specify an outside diameter of 21.0 mm in Tinkercad and run through Cura, the printed object diameter is 20.1 mm. When I specify 22.0 mm the objects prints as 21.3 mm as shown in the following images.

 

I create a circular Tinkercad object of 22.00 mm (01Tinkercad.png) that I export to "vacuum connection.stl" Using an online viewer, (02GeneratedSTL.png) that shows the dimensions as I specified them. Loading into Cura 4.7.1 Prepare window (03Cura.png) shows the proper dimensions in the lower left AND (red arrow) shows that it is expecting to print using PLA filament. Upon "Save to file" in Cura, an online gcode reader also shows the proper dimensions (04GeneratedGcode.png) BUT (red arrow) shows that it is expecting to print using ABS filament.

 

Could the PLA versus ABS be why my dimensions differ from design to printed object? If so, I gather I must change the settings that Cura uses to generate the gcode (that already shows PLA despite generating ABS).

 

Or are my precision expectations too unrealistic?

 

Regards,

grNadpa

Edited September 8, 2020 by grNadpa

[GregValiant 1,411]

Holes like to come out undersized.  The plastic is viscous enough that it gets dragged towards the center as it extrudes.  @gr5 calls it the "Snot Factor" and the smaller the hole the worse it is.

Under "Shell" you'll find "Hole Horizontal Expansion" and it will get you much closer.  It is an offset setting and so it acts on the radius.  Your diameters are off by about .8 so set the Hole Horizontal Expansion to .4mm.

Another thing - is that image of the part accurate as to the resolution of the STL file?  Check in TinkerCad and see if there is an option for a higher resolution when it generates the STL file.  A part with so few facets around the diameter will have it's own issues as the plastic gets dragged across the facet angles effectively making the hole smaller.

[grNadpa 2]

Terrific response. Just wanted to make sure that my skill deficiency was not the issue in this case. Thank you. Will look for "Shell".

 

Yes, Tinkercad does provide a way to add additional "sides" to circle / cone / etc. objects. I just take the defaults, for now, as I endure the learning curve of this "new-fangled" software [being a retired Mainframe Computer applications guy that made my living in COBOL and CICS].

[GregValiant 1,411]

I won't comment on any skill deficiencies you may or may not have.  I have trouble enough with my own.  I remember COBOL.  I cut my teeth on an IBM360 and Fortran while at GM Fisher Body.

The finer the model, the smoother the print.  I think that's how some of the machine manufacturers spoil their new customers.  They provide them some Gcode files that were sliced by extremely competent people using extremely fine models and the prints come out superb.  Then the user can't duplicate the superb model because the files downloaded from a website aren't near the resolution of the canned files.  No amount of fiddling with settings can get a rough model to look like a fine model.

[geert_2 558]

In DesignSpark Mechanical there are 3 standard quality-options for export to STL: coarse, medium, fine. I use the fine, which gives about 2x to 3x more triangles than in your image (which corresponds to about medium). Custom quality allows for even better, but that is overkill: the printer can't follow too fine details anyway.

 

But there can be lots of other reasons why accuracy is not great: extruded flow too much or not enough, nozzle too hot or too cold, blobs, incorrectly callibrated X- or Y-steps per mm, printing fast so corners get too thick when it has to slow down suddenly and creates overextrusion (due to nozzle pressure not immediately releasing), elephant feet, etc.

 

Maybe try printing a square of 100mm x 100mm, 2mm wide, and 1mm high? That should give you an idea if the X- and Y-steps/mm are callibrated well.

 

As for the "snot factor" causing too small inner holes: try drawing a circle with honey dripping from a spoon, or try to make a circle by stretching a rubber band into a circle (without any inner support). Gives a good idea how the thick molten material is pulled inwards.

 

[grNadpa 2]

@GregValiant Truly enjoyed your comment. GM Fisher Body ... was that the branch on Willow Springs Road in Countryside, Illinois? Fortran was never my forte, but after programming an IBM 1401 and wire-boarding unit record equipment, I moved up to COBOL on a Control Data 3300 octal mainframe.

 

Suspect you are correct on expectations for a fine model. My adventure presumes to make attachments for household appliances. I'm disappointed that the measurements don't transfer accurately, but encouraged that the inaccuracies are consistent from print to print. At least so far. So I end up wasting filament and a great deal of time fiddling the dimensions. But ultimately the objects I've printed do the job.

 

Edited September 10, 2020 by grNadpa

[grNadpa 2]

@geert_2 Thanks for your detailed response. Honey and rubber bands. Don't recall working in those mediums (grin). Very helpful visuals. Thanks.

[GregValiant 1,411]

I was at Fisher Body at the Tech Center in Warren, Michigan while I attended GMI (now Kettering U.).  I learned a little bit of analog programming at GMI too.  Now that took some thought!!

I did some of work later on getting lab machines to write their data to MSExcel via serial ports.  I've written an application in Visual Basic for Apps to talk to my printer.  It works well but it requires Excel, so as a project I'm attempting to port it to VB using Visual Studio.  What I'm learning now is to really hate Visual Studio.