@trevorflowers - I have an old microscope (I am a bit of a collector), but not the micrometer stage.
or as @dougl says - I was just using settings that make the line width hard to control and the 2 and the 10 are just statistical flukes 
ā¦ Oh wellā¦ I thought that I was being so cleverā¦
Shows why scientists should always open up their work to criticism and should never work in isolation.
2 Likes
Today I printed the same part specifying 3 external perimeters, outer perimeter printed first, perimeter width 0.44 mm (=110% of 0.40, the nozzle diameter).
The result is almost exactly the same as reported before. The external dimensions of the calibrator thing are ~ -0.05 of 25x25x8 (i.e. ever so slightly undersize) - the 2mm and 10mm drills pass through the respective holes, the 8mm drill is a tight fit through the 8mm hole and the 4 and 6 mm holes are too small for drills to fit at all. The drills are all about 0.10 mm undersize, according to my Vernier callipers (the fact that I am working at or near the least significant digit of my cheap no-name Vernier callipers has not escaped me - but its the best that I have)
All print speeds were quite slow, unfortunately the small perimeter speed was not as low as Michael suggested (maybe next time 
)
The prints by the way are beautiful (unblemished, even surfaces, etc).
The discrepancies are not large - but enough to cause problems when tolerances are tighter than ~ 0.2-0.3 mm. I have a hunch that if I increased the diameter of the holes further, the sequence will repeat - that is some holes will be under and some will be over nominal size by a small amount that is less than the line width. Unfortunately thatās when I have had problems, as I donāt have drills that are bigger than 12 mm in diameter.
Even though I found no noticeable difference between Cura and Slic3r wrt to print accuracy, I am impressed by Cura in many respects and will now use it exclusively.
1 Like
Iām not sure how Cura handles calculating line paths but hereās a blow by blow if what Slic3r does. Pay particular attention to the very last section, ādefaultsā where it states the default extrusion adjustment for perimeters(only one side of the line path interfaces with another line path.
https://manual.slic3r.org/advanced/flow-math
So how you want your holes is adjustable in Slic3r( and Prusa Slicer ). Some prefer to run a drill bit through holes so prefer some extra extrusion on the perimeter so when the bulging end of the exposed extrusion path is removed by the drill, thereās a near solid wall in the hole. Others, who might use self tapping screws in these holes prefer the undulating inside surface for the threads to cut into and push aside on insertion.
When I first started 3D printing I tried as many slicers as I could find and since I was learning, I wanted as many setting options exposed as I could and that lead me to Slic3r. A few years later a friend with a Prusa printer told me about the Prusa Slic3r and Iād put it off until one day I read how it was Slic3r under the hood with modifications(all provided to Slic3r if they wanted them) so I tried it and found I liked it. Prusa was moving faster than Slic3r with updates like having variable layer thicknesses and better support implementations so Iāve been using Prusa Slic3r and now Prusa Slicer.
Just over a year ago Stratasys purchased Ultimaker who was the corporate entity behind Cura. Stratasys promotes Cura with their Makerbot line of machines but lots of the literature I saw from them promotes their non-free commercial slicer for their FFF machines. If I were a Cura user I would periodically look at the git repository and observer the update frequency to keep confidence in its future.
Great job on doing your own tests and analysis and sharing it. Itās a great way to learn your machine, slicer and FFF processes.
1 Like
And thank for your generosity with your replies - I learned a lot from you.
I started using Slic3r when I was given the Tevo Tarantula for my birthday - The Tarantula is basically a DIY kit - (imagine getting a 3D printer form IKEA) - and they recommended using Repertier Host as the software - Repertier in turn offers Slic3r, Prusa or Cura Engine as the slicers, and I chose Slic3r and never really had any cause to change as I was happy with it for the same reasons as you.
Recently a local hobbyist store was offering the Dobot Mooz 3 (3 colour, single nozzle) delta printer for $149 AUD . I was always intrigued by delta printers and this price could not be passed up (Fully assembled, great cable management, touch screen controller, glass block build plate, 3 steppers for movement control with lead screws, 3 steppers for filament feed control, and 3x250 g of filament as well as spool holders).
As is the case with many of the things that are made in China the instructions and support are very limited and the manual only has recommended settings for Cura. I did not want to spend too much time figuring out how to set it up in Slic3r so I downloaded Ultimaker Cura.
While playing with my new delta printer I noticed that an M10 thread that printed with it took and M10 bolt without tapping. The same happened when I used Cura to slice the same part for the Tevo. So I thought the solution to the previously reported problems was the slicing software. You know the rest of the story.
There is a lot to like about Slic3rā¦ but I like the clean modern interface in the Ultimaker Cura and itās print preview implementation.
Iāll post again if I get any further insights - at the moment I feel that never has so much been said and done over 0.2 mm ;-).
2 Likes
Have you considered your microsteps/mm. could they have anything to do with the inaccuracy. Just a thought.
I wanted to follow up with two possibly relevant things Iāve learned, not for @NotTheMama who has a working solution, but for whoever next finds this in a searchā¦ 
The first is the discovery that my new enclosed printer prints passable threads in ABS without post-processing with a tap, and without any modification to the thread form from the āidealā thread form in the model. I think that ABS shrinks more than PLA, so that was a surprise to me. Thatās an argument for it being reasonable to expect a well-tuned printer profile to print fairly accurate internal threads without modification to the model.
The second is that at least in the current 0.20.1 FreeCAD Iām running now, in the current Fasteners Workbench, there are preferences for modeling for 3d printing that allow you to set scale factors that are then used for modeling thread forms specifically for 3d printing.
You basically subtract the metric or imperial threaded rod from a hole, and if you set it to be 3D optimized, then it uses the adjustment from the configuration.
4 Likes
Well, Michael I thought that it was reasonable to expect that tooā¦ and found that āaccuracyā is more related to the alignment of the planets than how well tuned the printer is. 
Seriously though, Iāve been thinking about this a lot, and there are so many variables involved between drawing a part in FreeCAD and printing it on a 3D printer, that its a wonder that it works at all let alone reproduction to within Ā± 0.4 mm. I wanted parts to be printed EXACTLY as I had drawn them on FreeCAD - thatās not accuracy - thatās fidelity with FreeCAD - and before I can print a FreeCAD model I have to convert to STL, and slice it. There are a zillion variables some of which change day to day.
I was a bit negligent by not reporting part of the solution. FreeCAD thread creator has several pre-programmed threads and I didnāt want to fiddle with those otherwise I wouldnāt really know what I was creatingā¦ but I discovered that you can add a āCustom Thread Clearanceā and keep every other thread parameter the same. That really got me out of a bind! It means that all you need to do is workout the resolution of your prints and then use that as your tolerance in the thread definition in FreeCAD. (See below, and note that Custom Thread Clearance is set to TRUE in the part definition parameters.)
There are probably similar setting in Fusion 360, etc.
Separately, I also found that the tightness of fit of drill bits in a 3D printed hole depended on the diameter (of the hole)! This led me to the conclusion that 3D print resolution (extrusion diameter, single wall thickness) plays a role, but I could be wrong. Iāve been wanting to experiment with different diameter nozzles, but havenāt gotten around to it.
I hope that all of this is useful to others too.
Wonderful post as always @mcdanlj - thank you!
3 Likes