Accuracy?

[SeanFawcett]

What sort of accuracy (precision) should I expect with a 3-D Printer?

I recently purchased a (slightly used) Makerbot replicator Z18.

Generally I am pretty pleased with it but the accuracy seems questionable.

I have a part designed in Solidworks and opened with Rhino. The thickness, measured in Rhino is 0.090"

The thickness of the resulting print is 0.097". As a result two pieces don't fit together?

What kind of accuracy should I expect?

Is there anyway to calibrate this?

Thanks

[Mjolinor]

you have to account for the width fo the plastic you are extruding. I reckon if you add the extrude thickness to the first figure you will get the second figure but it's been a awhile since I worked in inches.

[Ama-fessional Molder]

Technically what you're looking for is precision, not accuracy, but that's just splitting hairs.

Without knowing anything about the software, do you have any shrinkage correction applied?

.007" is a fair amount. When you say thickness, do you mean wall thickness or layer thickness on the z axis?

I'm not the one to answer the question, but I know that there is a difference in the solution depending which case it is.

[SeanFawcett]

Thank you for the reply. Yes, you are correct, we are talking about precision...

I have not applied a shrinkage, not sure if there are rules of thumb for this... I suppose I could calculate some based on the results that I am getting but the error is not the same in all directions. For example the part is nominally a channel shape, with the following section: 1.800" wide, .750" high with a wall thickness of 0.090". The 1.800" dimension measures very close to the model (1.801") but the wall thickness is 0.097" when it should be 0.090"

When the part is printed, it is standing on end so the "Z" dimension is lengthwise along the channel and the cross section is in the X-Y plane.

Also, I have experimented with the Makerbot settings between high and standard resolution. The precision seems about the same but the quality actually seems poorer for the high resolution print.

[curious aardvark]

what's that in real numbers ?
ie millimetres.

Also what your cad program says and what your printer produces are never exactly the same. You have to learn to compensate - and every printer setup is different.

Run some calibration cubes and that will give you a better idea how your machine works.

bear in mind that different filament brands will extrude and cool at different expansion rates. This is determined by a variety of things from the humidity and temperature of your room to the accuracy of your hotend, the speed you print at, any slicer applied extrusion multipliers and a plethora of other things.

getting a consistent and accurate print is as much an art as it is a science. I've spent 2 years learning how my flashforge creator behaves to the extent that I can pretty much get most prints within a few hundreths of a millimetre where I want them.

[SeanFawcett]

Thanks, the dimensions in mm are:

The part is nominally a channel shape, with the following section: 45.72mm wide, 19.05 mm high with a wall thickness of 2.29 mm. The 45.72mm dimension measures very close to the model (45.74mm) but the wall thickness is 2.46 mm" when it should be 2.29 mm.

I will print a calibration cube and see what that tells me.

[SeanFawcett]

So, I downloaded a "calibration model" from thingiverse and printed it:
http://www.thingiverse.com/thing:24238
5mm_Calibration_Steps_preview_featured.jpg

Each step is 5mm cubed. After printing it I measured two rows of steps in each of the X, Y, and Z dimensions (9 measurements in each direction). What I found was:

Z dimension (up and down)
Max error: 1.60%
Min Error: 0.42%
Error Range: 1.19%
Average Error: 0.78%

X dimension (left / right)
Max error: 4.14%
Min Error: -0.69%
Error Range: 4.83%
Average Error: 0.85%

Y dimension (back to front)
Max error: 5.16%
Min Error: 0.59%
Error Range: 4.57%
Average Error: 2.24%

These numbers seem pretty high... What's more alarming is the range in measurements which would indicate lack of predictability.

Is there any way to tighten up the accuracy, or precision of the Makerbot Z18?