Circle diameter discrepancy

[nirfriedman]

Hi,

I am printing something that involve multiple gears. I noticed that the parts do not match up properly, and so started measuring the printed object.

As an example, printing a vertical column with diameter of 5mm results in a print with diameter of 4.5~4.6mm. At the same time printing a square of 5x5 results in the right size (up to .1mm error).

I am using Simplify3D (latest version). The dimensions in STL show correctly in S3D, so the problem is in the printing. I re-calibrated my extrusion multiplier, and this is not the problem (again the square profile is correct). I tried printing both Inside-Out and Outside-In but this does not seem to be the problem.

Any ideas on how to solve this?

Thanks!

Nir

[jimc]

.1mm is slightly less than .004". thats pretty darn close for a printer. you have to remember its not machined, its printed. your not going to get it perfect all the time. there are variables to consider like the thermal expansion and contraction of the plastic you are using. you need to compensate for that by scaling your model in the x and y. every plastic and manufacturer's formula is different. esun petg you need to scale up aprox 0.4%, abs can be 0.8-1.2%, epc 0.85%. usually if i print a 3" dia circle there will be a dia difference depending on where you take your measurement. i will usually see an out of round condition of .003"-.008". its never right on the money. thats just the way it is. if you dont scale your model, typically with petg you would find your part to be undersized .020-.030" for something with a dia of about 3".

[swbluto]

With my ABS, due to contraction of the plastic after cooling, I've found a rule of thumb is to oversize the model by about .1". That's about 2.5mm?

[ednisley]

the problem is in the printing

Cylinders have a particular problem: the nozzle tends to pull the thread toward the center of the cylinder as it travels around the perimeter. Both holes and posts come out smaller than they should, with the error decreasing as the diameter increases.

I found that a Finagle Constant worked reasonably well, after measuring a test piece from Thingiverse:
http://softsolder.com/2013/04/21/makerg ... t-objects/



Larger objects probably won't scale that way, because they're firmly attached to the platform and can't shrink uniformly. The only solution for that will be to fire one for effect, measure the result, and tweak the model to make the answer come out right on the next one...

[jimc]

thats alot but then again it depends on the size of the model. you should really make the model to size then in s3d scale up a % rather than a hard #. a 1/2" cube for instance will shrink less than a 5" cube. scaled at a % takes care of that and keeps it all in line. it will only need to be scaled in x and y. z wont be affected. esun abs is 0.8%. not sure about your brand but its probably a good place to start.

[jimc]

lmao i should have known ed would have done an all out lab experiment on it.

[swbluto]

thats alot but then again it depends on the size of the model. you should really make the model to size then in s3d scale up a % rather than a hard #. a 1/2" cube for instance will shrink less than a 5" cube. scaled at a % takes care of that and keeps it all in line. it will only need to be scaled in x and y. z wont be affected. esun abs is 0.8%. not sure about your brand but its probably a good place to start.

I really doubt 10 inches is going to shrink down to 9 inches. There's a practical hard limit to how much it'll shrink, given that the bottom layer adheres well to the bed; It's not like your injecting an entire 10" object, which shrinks uniformly. It only shrinks by a little bit at the point where the filament is extruded, since the percentage of shrinkage is proportional to the filament, and not the whole object.

[jimc]

that would be 10%. but you are right. it wont be a linear, absolute #. its just a scale to get you close

[nirfriedman]

the problem is in the printing

Cylinders have a particular problem: the nozzle tends to pull the thread toward the center of the cylinder as it travels around the perimeter. Both holes and posts come out smaller than they should, with the error decreasing as the diameter increases.

I found that a Finagle Constant worked reasonably well, after measuring a test piece from Thingiverse:
http://softsolder.com/2013/04/21/makerg ... t-objects/

Thanks for the detailed answer. Just for the record I am using MakerGear ABS.

Thats a real bummer. Does changing the infill and/or printing speed reduces the problem?

[jimc]

im sure printing extremely fast doesnt help but printing in any normal range of speed i dont think will make a difference. alot of it has to do with the thermal contraction of the plastic as it cools. its mother nature and just the way it is. your not going to stop it. they even have to deal with this in injection molding. the molds are made oversize so as the material cools it will shrink back down to spec. the pia is figuring how much and where the part is going to shrink. the other factor is the machine. they are only accurate to a certain amount. machining a part on a lathe is going to be far more accurate and true. thats why these are prototype machines in most cases. it will never be accurate down to .001"