Printing a screw and nut - a accuracy issue.

I have a downloaded file that has an eyepiece with a screw thread for focusing and a receiver for it. Looking at the file, it has about 0.14mm clearance between the part screw faces. My first attempt at printing it was a disaster. With a lot of sanding and Dremel work, it still won't screw together.
So the question is - what is the trick to printing this so it will fit together. I am using Repetier and Slic3r. It is a home built CoreXY machine and otherwise prints pretty well - except that my dimensions are never exactly what the model is. With 0.14mm tolerance between the parts, tolerances need to get better. I am printing - 20mm/sec, 0.1mm layer, first layer is 0.2mm, material is PLA, temp is 205, 200°C, external perimeter first, no hotbed, hairspray, Marlin f/w, full cooling past layer 2, infill 15%

What can I do to get better tolerances, or is the clearance just too tight?

Edited 2 time(s). Last edit at 02/18/2017 08:55PM by ruggb.

First of all, the X and Y steps/mm have to be calibrated very accurately. If they are not accurate, circles will print oval.
Second, X and Y have to be 90 degrees apart or circles will print oval.
Third, if the extruder isn't accurately calibrated your tolerances will be off.

In order to print tight tolerance parts that screw together you have to align and calibrate the crap out of the printer.

0.14 mm is pretty tight. Even after careful calibration and alignment I'd bump that up to about 0.2 to 0.25 mm. As you are aware by now, holes tend to print smaller than designed. That means you have to allow more spacing between the two parts that have to thread together.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Thx Doc
no problem with ovals. X&Y are mechanical things based on 6400 steps/rev on the steppers (200 x 32 microstepping).
It is a CoreXY, so by design it is pretty much square
extruder calibrated about as well as I can measuring 400mm - but then it wasn't hot. I may need to try that again. Problem is, it probably varies with temp and layer height. Any info RE that?

It wasn't my model. I have no idea what kind of printer the author has. I modded the files to give me about 0.5mm clearance. So taking into account the printer anomalies, I figure I have about 0.4mm when printed. After the first screw thru, they are perfect.

Edited 1 time(s). Last edit at 02/20/2017 10:11AM by ruggb.

holes shrink, pla figures:

where X mm is the required diameter of the hole then Y mm is the dimension (in model) needed to achieve it and that:

a) for vertical holes Y = 1.0155X + 0.2795 mm
vertical meaning your circle is standing on its side, like a coin would never

b) for horizontal holes Y = 0.9927X + 0.3602 mm
how coins lie flat

I am afraid that Hedgehog has mistaken my original post slightly. The formulae are correct against the descriptions of vertical and horizontal holes

Vertical holes refers to holes whose circular axis is vertical (parallel to the Z axis like a coin lying on its face). Horizontal holes refers to holes whose circular axis lies in the XY plane (like a coin standing on its edge).

Apart from shrinkage there are other factors involved. No STL file contains a true circle, since any line or edge other than a straight one will be approximated to by a series of straight lines. Then net result of this is that all circular or spherical objects will have a diameter that in part is smaller than the true diameter.

When we are talking about a hole in a solid object, the effect of all the factors is to reduce the diameter of the hole when compared with the designed diameter. When we are talking about a solid object such as a rod or peg, then the effect of all the factors is to very slightly increase the diameter. Because the factors involved operate in different directions, the reduction in size of a hole of a given diameter is more pronounced than the increase in size of a round rod or peg of the same diameter.

No amount of calibration of your axes, which is generally carried out using square or rectangular objects, will give correct sizing of round holes or rods since the STL circular approximation is not present in those models. Nor will calibration with a given cube size guarantee that cubes of a different size will print correctly, as some of the factors involved are not linear.

If you must have absolute precision of your printed parts, then you need to understand the factors involved and make the necessary adjustments to the design of your model. The two formulae will help you to do this for round holes.

Lastly in your chase for absolute precision do not forget that stepper motors operate in discrete steps, and have a look in the specification for your machine at the figures given for positional accuracy. Chasing ever increasing refinement in the machine will not produce the results that you seek. Now that I have bored you all to death I will leave you to it.

Edited 1 time(s). Last edit at 02/20/2017 06:15PM by Supermec.

Thx hedgehog, I will experiment with that. Is that written someplace that may give more info. It would seem like one would have to take into consideration temp, layer height, first layer height, squash level, etc. I have previously found that a horiz hole comes out about a layer height smaller than design.

My object was not a plain hole, it was a very large (40mm) screw and nut. So making the clearance 0.5 and subtracting 0.1-0.22mm for a 0.2mm layer yeild 0.2-0.4mm for a final clearance. I didn't measure it, but it feels about that.

ruggb,

it doesn't matter if your object is a threaded hole the formulae will still give good results. Please note that Hedgehog got his interpretation of vertical and horizontal holes the wrong way round.

A vertical hole is a hole that looks round in plan (when looking down at it). A horizontal hole looks like a circle in elevation. The formulae are based on a number of factors that are lengthy to explain in detail. I posted them originally having obtained them from a professional 3D printing firm.

They should give good results when printing PLA as I produced some gears that were a close push fit on a metal axle rod. They would need modifying for ABS printing because of the higher shrinkage of ABS. You would need to do some test prints to compare the results and change the formula depending on your results.

If you measure the outside of a printed screw (you could print a short test rod the same design outside diameter) you should find that the diameter increases about 0.08mm when using a 0.2mm layer thickness.

I think that formula applies to a specific manufacturer's (probably) proprietary slicing software that uses a specific/minimum number of segments to represent a hole. Just thinking about it for a moment, if the hole were represented by 12 segments you'd have to use a different fudge factor than the same design size hole represented by 100 segments. Likewise, that formula may apply to a specific formulation of the plastic that shrinks a specific amount at a specific print temperature. Change the plastic, the temperature, or the resolution of your stl file (all of which are usually tightly controlled by proprietary filament and slicing software on a lot of industrial printers) and you're back to guessing/test printing at what will be needed. If your extruder calibration is changed the amount of plastic it deposits will change so the fudge factor has to change.

To me all that formula says is that the design size hole has to be a little bigger than what you need because the printed hole is going to be a little smaller than you expect. But we already knew that...

Edited 2 time(s). Last edit at 02/21/2017 01:28PM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

And they say that additive manufacturing saves material..................
By the time one prints enough test prints to get this right, it would be better to have a lot of drills and sandpaper.

Edited 1 time(s). Last edit at 02/21/2017 09:09AM by ruggb.

digital_dentist

I suggested the formulae as starting points, rather than start from nowhere, which is what you advocate.

It is not the slicer that decides whether the hole is represented by only 12 segments, but rather the 3D design software that has been used design the model and then to convert that model into an STL file. As I said, they apply to PLA, and I have used them successfully to print accurate holes with PLA that are a close push fit on round rods. Yes the adjustments will change if you switch from PLA to ABS due to the different shrinkage, I said that, and I said that the formulae would need adjustment for use with ABS, and the reasons why.

Why should a change of extruder calibration change the adjustment factors (or fudge factors as you refer to them)? The extruder is either calibrated to extrude correctly, or it is not calibrated to extrude correctly. I have not suggested these formulae as a means to correct for this aspect.

It is easy to stand on high and criticise and state that the formulae are a waste of time. What I offered was some constructive advice to the inexperienced in order to give them a starting point. Furthermore there is nothing in what I have said that is incorrect.

Edited 1 time(s). Last edit at 02/21/2017 09:24AM by Supermec.

@Supermec - As OP, IMHO, all of the relies are good information. I think that you may be over reacting to the Doc's statement. The key would be the part that starts "To me.....", which is not unlike my drills & sandpaper take on it. Unfortunately, a screw required drastic measures like modifying the design. Drills and sandpaper don't work.
I took a SWAG at it before your reply and it could be a little tighter but not enough to re-mod the file and reprint it. It works fine. Your info appears like it might get it a little closer - instead of 0.4mm, maybe 0.25mm or 0.3mm. BUT, everything affects it so after rigorous trials one might get it "right" for a particular machine, with certain filament, at specific settings of layer height, first layer, temp, ext perimeters first, for a particular slicer, and bed adjustment on a certain day with a 50% RH - and it will be "wrong" tomorrow when any one of these change. The question comes down to, with this formula can I get close enough that the rest of the variables do not matter. If not, I need a different process.
The "S" in SWAG is key. All science is experimentation and observation. So based on my limited experience and observations, my SWAG wasn't too bad.
thanks people.

Edited 1 time(s). Last edit at 02/21/2017 11:33AM by ruggb.

Quote
Supermec
digital_dentist

I suggested the formulae as starting points, rather than start from nowhere, which is what you advocate.

Actually I advocate starting at about 0.2-0.25 mm size difference between peg and hole and bumping it up or down from there depending on print temperature, print speed, filament material, extruder setting, hole diameter, STL file resolution, print layer thickness, etc. You can often tweak a too small hole with a drill, but if you print too large there isn't much you can do but reprint at a smaller size.

Quote
Supermec
Why should a change of extruder calibration change the adjustment factors (or fudge factors as you refer to them)? The extruder is either calibrated to extrude correctly, or it is not calibrated to extrude correctly. I have not suggested these formulae as a means to correct for this aspect.

It is easy to stand on high and criticise and state that the formulae are a waste of time. What I offered was some constructive advice to the inexperienced in order to give them a starting point. Furthermore there is nothing in what I have said that is incorrect.

People are building printers here- they change things like extruder calibration/flow all the time. There is some useful info to be extracted from your formulas: specifically, the fudge factors of 0.28 mm and 0.36 mm which are right in the ballpark I advocate.

Anyone can pull formulas out of their butt, and they may give the illusion of knowing something, but without actually knowing the circumstances of its derivation and the limitations of the formula (example, to what range of hole sizes, nozzle diameters, and layer thicknesses is it applicable?), having a formula will lead you to think you know more than you actually do.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

an interesting test.
I created this model:holeplanvsactual.min.p.zip
It is 1mm high with holes from 1mm to 9mm sized per the formula.
Slic3r was set to 0.1mm layer, 0.2mm first layer, external perimeters first, 200°C, 205°C 1st layer, PLA filament, no heat, max cooling after layer 2, 0.4mm nozzle
Result: orientation long side = X axis
layer height = 0.25mm ; measured skirt
1st layer width = 0.52mm ; measured skirt
left side of brim = 0.27mm ; measured brim
right side of brim = 0.21mm ; measured brim, 1/4 plate glass bed, not exactly flat
Width = 13.27 ; model 13.00
Length = 55.23 ; model 55.00
hole = actual X, Y measured ; measured with digital calipers on a plastic print 1mm high
1 = 1.07, 1.02
2 = 1.85, 1.79
3 = 2.97, 2.91
4 = 4.18, 4.07
5 = 4.93, 4.93
6 = 6.30, 6.22 ; print anomaly
7 = 7.25, 7.25
8 = 8.30, 8.31
9 = 9.27, 9.32

Measured from the bottom, the 1mm hole is 0.68mm. I assume that is because the hole is small enough that the calipers are not going all the way into the hole from the top.

I guess I should stick with 1, 3, or 5mm holes.
What is your take on the results?

I think 1mm height is too close to the bed to make reliable hole size measurements- there are all sort of first-few-layer problems that can happen that close to the bed. Try printing 10mm high and measure the hole sizes on the top side of the print.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

This is 5mm high with holes from 1mm to 9mm sized per the formula.
all the same setting as above: Slic3r was set to 0.1mm layer, 0.2mm first layer, external perimeters first, 200°C, 205°C 1st layer, PLA filament, no heat, max cooling after layer 2, 0.4mm nozzle

Result: orientation long side = X axis; measured with digital calipers
layer height = 0.24mm ; measured skirt
1st layer width = 0.47mm ; measured skirt
left side of brim = 0.25mm ; measured brim
right side of brim = 0.32mm ; not printed at same spot on bed as last one
Width = 13.52 ; model 13.00
Length = 55.37 ; model 55.00

hole = top, bottom measured ; note above measures were of X and Y direction, these are top & bottom
1 = 0.98, 0.69
2 = 1.79, 1.91
3 = 2.91, 2.94
4 = 4.04, 4.14
5 = 4.85, 4.95
6 = 6.05, 6.16
7 = 7.08, 7.15
8 = 8.09, 8.24
9 = 9.10, 9.00

Strange, most bottoms measure larger, but not all. Me or the bed??