How accurate homemade Prusa I3 can be?

Hi,

I just finished the calibration of my homemade Prusa I3, and everything is fine except that the dimensions are always 0,1/0,2mm out
for exemple:
20mm box
Expected: 10mm x 20mm
Obtained: 9,85mm x 19,88mm
Accuracy: 99,85% x 99,4%

Hollow box
Expected: 12mm x 20mm
Obtained: 11,9mm x 19,88mm
Accuracy: 99,16% x 99,4%

50mm towel
Expected: 50mm x 10mm
Obtained: 49,7mm x 9,84mm
Accuracy: 99,4% x 9,84%

25mm Cube
Expected: 25mm x 25mm
Obtained: 24,75mm x 25,22mm
Accuracy: 99% x 99,12%

So, it seems accurate but my questions are:
-Is it in the standard of accuracy for a RepRap? Is it in the standard of accuracy of other printer such Makerbot,...?
-Is it possible to improve te accuracy?
-At this point, how much work will be needed for which results?

From my point of view the main problem is with the 25mm cube as the combination of the error make it looks slightly rectangular.

https://www.flickr.com/photos/boborz/17089147895/



note: I use slic3r 1.1.7, Pronterface, and a Wade extruder.

Thanks

Hi,

This phenomenon is due to polymers and z wobble and/or xy pulleys.

You wont get less than 0.05mm resolution. At most with allways the same filament (even same color)
You can do a good job of course. Ive had filament which spilled out of the nozzle and
Others wont come out that great. Some.will shrink more some not as much.

Even on one spool i had differences.
Expensive material should work better (if its also better quality).

You need to test some filament and take the more stable ones for fittings
And the cheaper ones for statues or bust or figures or something like that.

I can get my 20X20X10 cube to 20.01X20.01X9.99 or 10.01, then I get to print like now a customer part 50.8X50.8X7.62 and be off by 0.20 in each axis.

It's all in the part design and how forces act on it. Once you get it close enough, then you can play with the scale factor in the slicer to compensate for any errors.

You'll get best calibration results if you have the printer lay down a square that goes almost to the edges of your print bed. If you have a 250 x250 bed and you print a 200x200 square and measure it and calibrate with it, if the error in your measurement is 0.5mm, when you print a 20x20 square/cube, the 0.5 mm error divide down to 0.05mm.

Thank you for yours answers. I will try Play with the scale factor and the size of the calibrating object. That's very good hint I didn't thought off.

the_digital_dentist,

Thanks for that tip, I never thought of that..... Every you learn something new.

Depending on what type of feature you're dimensioning, +/- .1 to .2 mm can be in the neighborhood where you have to start paying significantly extra at traditional manufacturing operations using traditional, truly industrial-grade machining and forming equipment. That's not a bad tolerance in most types of work. To do better will take some significant OCD with accurate diameters for a high-quality consistent filament, a well-tuned machine, and scaling of the STL to account for the shrinkage of the material you're using.

I think your results are not perfect, but can a $500 printer offer perfect prints ? Appreciate the slicing operation introduce errors too. If your dimensions are within 0.2mm it's not that bad obviously. Except if you're into watchery or printing smallest caracters, you will never need a 0.01mm precision anyway. BUT, it's exciting to get the best print possible Calibration is one of the keys. But don't play with steps per millimeter on the XYZ axis. That values are calculated for your hardware. You will introduce errors to your prints, even for a 200mm part (because your adjustment will be good only for a part of that dimension). Obviously, you probably don't need to do that. Just calculate the steps per millimeter to match your components, then apply them. Playing with steps per millimeter on your XYZ axis is the last thing you wanna do.

As it was told above, some of your components and a wrong geometry of your printer can affect drasticaly your print dimensions. The wobble on your guides or your drive systems can introduce easily 0.2mm or more on your print dimensions. Too much temperature on the bed may corrupt the first layers. A print bed not flat enough too. A missaligned axis will do even worst. And so on. First, be sure your components are in great shape, and tune your mecanics and your structure as fine as you can, before to go any further. A good machine tune up is crucial.

Otherwise, be smart with your extruder settings. That will do a lot for your prints. Calibrate your steps per millimeter to extrude a perfect (as possible) 100mm (at last) run. Then, print a thin wall cube, with the same wall thickness than your nozzle diameter. Adjust now the flow on your slicing program to get the right amount of fused filament. Because the plastic expands when it fuses and you want to consider that. Now, if yours steps per mm for XYZ axis are set properly, you're probably as close as your print can be on your printer. You may have too small holes in your parts, but it's not a calibration issue, but a 3d conversion/slicing problem coming from radius interpretation. There's a slicer option to adjust that buy software.

It's a long run job, be patient and good luck !

---

Collective intelligence emerges when a group of people work together effectively. Prusa i3 Folger (A lot of the parts are wrong, boring !)

Quote
Zavashier
I think your results are not perfect, but can a $500 printer offer perfect prints ? Appreciate the slicing operation introduce errors too. If your dimensions are within 0.2mm it's not that bad obviously. Except if you're into watchery or printing smallest caracters, you will never need a 0.01mm precision anyway. BUT, it's exciting to get the best print possible Calibration is one of the keys. But don't play with steps per millimeter on the XYZ axis. That values are calculated for your hardware. You will introduce errors to your prints, even for a 200mm part (because your adjustment will be good only for a part of that dimension). Obviously, you probably don't need to do that. Just calculate the steps per millimeter to match your components, then apply them. Playing with steps per millimeter on your XYZ axis is the last thing you wanna do.

As it was told above, some of your components and a wrong geometry of your printer can affect drasticaly your print dimensions. The wobble on your guides or your drive systems can introduce easily 0.2mm or more on your print dimensions. Too much temperature on the bed may corrupt the first layers. A print bed not flat enough too. A missaligned axis will do even worst. And so on. First, be sure your components are in great shape, and tune your mecanics and your structure as fine as you can, before to go any further. A good machine tune up is crucial.

Otherwise, be smart with your extruder settings. That will do a lot for your prints. Calibrate your steps per millimeter to extrude a perfect (as possible) 100mm (at last) run. Then, print a thin wall cube, with the same wall thickness than your nozzle diameter. Adjust now the flow on your slicing program to get the right amount of fused filament. Because the plastic expands when it fuses and you want to consider that. Now, if yours steps per mm for XYZ axis are set properly, you're probably as close as your print can be on your printer. You may have too small holes in your parts, but it's not a calibration issue, but a 3d conversion/slicing problem coming from radius interpretation. There's a slicer option to adjust that buy software.

It's a long run job, be patient and good luck !

great info! Word reading.
do you know if cura has that option regarding too small holes?