Is there a way to change precision?

[TommyDee 1]

I know the arguments against but you have to trust me that there is a reason for asking...

I have a 4-digit layer height of 0.1925mm which Cura accepts as a layer height.

 

And if it cannot change the precision, how does Cura deal with the cumulative error?...  bummer!

Lets see...  From the output file:

0.442-.25=.192

.634-0.442=.192

.826-.634=.192

Therefore it is obvious the error is cumulative rather than a simple rounding.

 

How do I make the G-code output from Cura provide 4 digits for the Z value rather than 3 for G1 calls?

[ahoeben 1,948]

CuraEngine works in whole microns internally. A micron is a 1000th of a mm. Making CuraEngine more precise than the micron level is going to take a LOT of refactoring and will likely decrease its performance significantly and/or cause rounding issues elsewhere.

[TommyDee 1]

Bummer.

 

That means the layer height field(s) should probably be limited to 3 digits as well.

I wouldn't have known without dissecting the gcode output.

 

 

[nallath 1,120]

14 hours ago, ahoeben said:

CuraEngine works in whole microns internally. A micron is a 1000th of a mm. Making CuraEngine more precise than the micron level is going to take a LOT of refactoring and will likely decrease its performance significantly and/or cause rounding issues elsewhere.

And not to mention a ton of extra memory usage.

[TommyDee 1]

I got it.  Obviously I am trying to mimic a known profile.

 

It does surprise me that I didn't get a warning when I entered 0.1925 as a layer thickness and it allowed the entry.

A simple UI fix would make this clear up front.

 

Where do we submit ongoing bug fixes and enhancements?

 

[ahoeben 1,948]

https://github.com/ultimaker/cura/issues

 

Note that in this case, even after 1000 layers, you are still talking about a discrepancy of half a millimeter. Even “splitting hairs” is many times more than the “bug” you are talking about. I don’t think it will get much attention.

[TommyDee 1]

It is a matter of completeness in the UI.  I have a pet peeve in in that direction coming from that ilk.

Some people care about that 0.5mm.  A 1,000 layers is an every day thing under fine settings.

I could go on but it isn't worth the wasted bits in the cyber universe.

Just leave it to say that it would be a fit-n-finish enhancement request.

[P3D 46]

If something is not correct, it should be fixed. Especially if it is something like this - I suppose Qt (the GUI framework of cura) does have easy options for implementing the right behaviour.

[nallath 1,120]

3 hours ago, P3D said:

If something is not correct, it should be fixed. Especially if it is something like this - I suppose Qt (the GUI framework of cura) does have easy options for implementing the right behaviour.

Unfortunately, there are always more things that are incorrect then there is time/resources to actually fix them with.

[P3D 46]

How to implement this (probably): Use http://doc.qt.io/qt-5/qregexpvalidator.html  (PyQt4 Code snippet - should also work in the PyQt5 used - see answer to this thread: https://stackoverflow.com/questions/34399485/qlineedit-accepts-only-character-in-pyqt4). I have zero Qt/Cura coding experience unfortunately, otherwise I'd probably just try to fix that myself and commit it to github...

Edited September 14, 2018 by P3D

[TommyDee 1]

I've posted the enhancement request at GetHub for an entry field validation. 

I'm perfectly fine with the micron level precision now that I know.