How to fix non round shapes?

that is correct now that I think about it good call

 

the latest firmware is on the forum under documentation

 

Home​

Forums>​

Robo3D Support and Documentation>​

Official Documentation​

Marlin Firmware Update 3/14/14

 

Ok well that last print looks fairly decent
the stepper motor voltages is fine so don't even worry about that any more
The step size in the firmware probably needs to be adjusted to get more accurate results
I would not base it on a singleprint though I would print at least three items and then take averages of the the lengths of the three axis' and the think more about adjusting the step size.

Make sure you have the latest firmware and the Arduino software and you are ready to go

 

not sure how this relates to this post

 

Step sixe is controlled in the firmware so you need "A copy" of some version of the firmware if you have one and you are not making other changes then what you have will probably work . You will also need the arduino software to actually do the update to the firmware.

As far as slicer goes it is not a requirement for the firmware change but it is nice to have the latest one that is stable but it is up to you. You can always go back as long as you have the proper install files.

The two issues are completely separate so decide on what to do with that in mind.

 

Peter did you simply update the firmware or did you make changes to the steps sizes. I didn't see you asking any question on how to figure that part out maybe you found it on line but those are the changes you need to look at in the firmware.
Also I would go away from the circle for now it is great visually but hard to get precise measurements from
Start with the 15mm cube and print it and then see how long each of the side are make sure you mark it before you take it off the bed so you can tell which side is which axis then you can determine the x is off by this much and the y is off by that much. If you have not done anything like this yet then I am not sure how you could have updated the firmware with new settings.

 

This was in response to hie prior attempts at fixing this issue a do agree it should not have to be done but keep this in mind all of these printers are built using motors to make the movements and usually those motors work the same but the is at time an oddball that does not. There are formulas that can calculate the EXACT value for the EACH specific motor and based on what he has tried already it was a logical thing to try. Using the formula you can get to .02 or less instead of the massive .5mm
The characteristic you list could be issues but I am not sure if it would do it to the extent he is seeing in his prints based on the images he has posted. Many many other people would be seeing similar things as they all are affected by those same issues, to my knowledge it is no the case I have read only about two or three cases.

It was simply another logical step in a deductive process.

For a long time the default values used for X and Y were 80.63 if I remember but if the calculations were done based on the motor , step sizes, belts specs, it should have been a standard 80 and recent firmware upgrades have it set that way. Based on many other CNC processess including 3d printing a standard method of fine tuning Each axis is common place I did it on my small hobby CNC machine and on the Robo3d. This process yields a result that is based on each particular printer motor and machine taking into account not only the exact variables of each printer individually but in addition to that how those components are actually working together. My current setup for my unique printer for the X is 79.107 and 78.953 for the Y and 2207.657 for the Z. Far from the default setups we were all given. My resolutions for the 15mm cube are about 15.01, 15 and 15.05 much closer than the .5mm you mention.

I never said it is the absolute end all solution to his problem but it was something that can not hurt it could only help.
according to what he has said he has tried or check the gears and the belts for slippage and found none and his images do not reflect major slippage along either of the axis as you normally see them in significant jumps at definite points but in his case it seems to be a gradual migration that is compensated for beautifully when the printer moves in the opposite direction by having the arcs actually join back up with each other at the end of the arc. So what we are looking to adjust is something that meets that criteria where while going in one direction it moves away from normal but does the EXACT opposite in the opposite direction and comes back to normal at the same point.

This adjustment or at least research into it, is something that would act in that exact way: a linear translation that increases the further away from a given point but is exactly opposite when moving in an opposite direction by decreasing the same amount when it reaches the start point returning it to zero as it reaches the start point.

I am not sure if the variables you describe can be that exact.

The cube will give us definite X Y and Z specs that can be used as a start point and it could be very easily identified as not being the issue but at least it can be analytically ruled out instead of arbitrarily thrown aside

 

Peter the modifications can be done in two places in the firmware for a more permanent spot or in the custom Gcode window of the slic3r config window.

The forumla is actually simple but you need to have some data that you get from the original firmware in the case of this varaible they have made recent changes to it value for the current printers as I said in my previous post it used to be 80.63 I believe and in the latest I believe it is 80 but you should check what yours says in the firmware you have loaded.
The one thing bad is we have no way to idenitify what firmware is loaded on the printer as a default so it is always good to check we have the firmware build number have nothing to associate that and the varaibles in question.

The value is contained in the file called "configuration.h" and its name is:

DEFAULT_AXIS_STEPS_PER_UNIT

The values are an array of four numbers each corresponding to a different item:
first the X axis
second the Y axis
third the Z axis
fourth the Extruder

the formula is as I said simple you can use the dimension of the cube to compare the result against you will have to print the cube several time to fine check the new dimensions but it will have changes
Each axis is calculated separately.


you know you wanted to print something that was 15.00 mm per axis

so taking that as one of the values
and the number of the current steps as the one
and the actual dimension of the axis as the third

You can calculate what the new value for the step size fore that axis should be

current steps new value
========= X ==========
actual distance dersired distance



this is an example

80 unknown
==== X =======
15.5 15.

with simple cross multiplication you get
80 times 15 = unknown times 15.5 OR
1200 = 15.5 times unknown OR
1200 divided by 15.5 = unknown OR
77.419 = unknown

so the new setting would be 77.419 instead of the 80 which make sense because the 80 was actually moving it 15.5mm so reducing it to the 77.419 number will make the 15.5 closer to the 15.0 number


As I said this is only a way to make it more accurate than the settings already are it may not be the full reason you are seeing what you are seeing

As to you r question if I can print circles well I have printed many of them when I make parts that have screw holes and many round objects including spheres as well so the answer would be yes although I do not recall ever measuring them to see how exact the sphere was and currently my printer is shut down due to my move so I can not do any test prints currently..hahah I already have a back log of things I need to print when it comes back up in a few weeks

As far as calibration goes all the printers HAVE been calibrated in that they have all been given the same baseline setup in firmware and component placements and alignments but NONE have gone through a printer specific calibration that would differentiate your printer from mine that is part of ownership and your specific requirements for print characteristics which they can not foresee.

The BASIC calibration that says this machine is a 3d printer with these generic specifications is done on every one of them this is the level where they are all the same.

hope this answers your questions and hopefully may help in your issue, if it doesn't, keep asking about it and we will figure it out.

 

The reason for that statement is the fact that all motors have tolerances and this COULD BE a case where a single motor may be at the extreme end of the tolerence scale this could cause issues.


I agree that it may not be the end all solution but in the images he provides it is a slow migration from a fixed point and shows a smooth linear exaggeration as it moves away from that fixed point and then as the print returns to the original fixed point the linear exaggeration is perfectly reversed and the print returns to the exact point again. This indicates to me something is causing this exaggeration and the fact that it is so smooth in the how it migrates and the exact same reverse migration back indicates it cannot be a simply slipping as the behaviour is is exactly mirrored and is so smooth an arbitrary mechanical slippage is unlikely.

Mike already suggested looking at the voltages from the controllers and that is also a viable consideration the only reason I have avoided that one is that mechanically it may or may not be so easy to perform it is dependent on the controller itself and some knowledge in electronics sometimes it is easy to find the correct points to test from some it is not in my case when I tested mine the controller had a specific point setup so it could be easily tested not all controllers have that and in actuality on one of the controllers the point was accidentally covered by the small heat sink used blocking it, meaning I had to find another way to do it not so easy if you are not familiar with electronics that much.

This analysis has no down side it can only make things more accurate and if the motor is out of specs it could easily be identified.

I think taking an average reading of the sides of the cube can get a pretty good picture and yes I agree it won't be perfect as you can keep doing the process over and over and get slightly different results each time but at a higher level I have see those changes being very very small such as going from 78.406 to 78.593 and then 78.480 but that is a much smaller change than the original one from 80.

 

The attached STL was one I used to get the bed level and it was pretty close to exact even at that size.