Loading

MakerBot Print is our newest print-prepration software, which supports native CAD files and STL assemblies,
allows you to interact with all your printers via the Cloud, and many other exciting new features.

Download Now

lukeskymuh

Calibration cross

by lukeskymuh Aug 15, 2017
Download All Files

Thing Apps Enabled

Please Login to Comment

This is a late request, but is there a version that could be applied to a CTC Bizer printer? I ask because the print as presented overhangs the print plate, and obviously will not print. I have considered reducing the size by 50%, but that has knock-on/iterative effects in error calculation. I can see no other post that has asked this question.

It works with almost any typical fdm printer as 45° overhangs are normally not a problem.
You max scale and there is a field in the excel sheet to concidder a scaled version.

hi , i think there is a small probleme on the sheet for the Y axy.

the point 6 should do 50 and the 10 should do 75 . if you look your picture and the printer i thing you invert the data ..

point 6 it's 75 and point 10 is 50

same bewteen 7 and 9

You are right, i corrected it. Thanks.

Hi, Thanks for creating this.

I'm trying to wrap my head around how this will work in the case of Extrusion multipliers.
the excel sheet has it hard set to 1.00 but my slicer is setup for different multipliers based on tip size, extrusion width etc.

I guess my question is, Say my extrusion multiplier is currently .92 for ABS and it's extruding perfectly. If I bump this to 1.00 the print will be over extruded and unable to be measured correctly, what steps do I need to take to ensure an accurate result?

Oh, the mutiplier is not the extrusion mutiplier.

The advantage of the method used here is that under or over extrusion. If you are over or underextruding you will only see a change in the "mean wall thickness deviation"

I changed the description to " corretion" to avoid missunderstandings)

By the way: ajusting your extrusion multiplier to correct dimensional accuracy is normally not a good idea.

Hey Luke, great Job btw.

One thing is strange. Maybe you can help me. I think that if the part shrinks by the cooling, you have to scale the part larger when slicing. In the example of PLA, the shrinkage is 0.27%. Then the scaling for the part would have to be 100.27% instead of 99.73%.
Maybe I'm wrong. Thanks in advance.

It is correct now.

Oh, right, thanks. I will correct this as soon as possible.

Hey Luke,

Just wanted to point out something that will likely confuse people in the Excel document. For both the X and Y image showing where to measure the image shows the Y-axis pointing up, but you need the X-axis pointing up to measure the Y values. Not a big deal really, was easy to figure out, but just thought I'd mention it.

I have to say this system works really well! I was having tolerance issues with PETG that weren't present with PLA and this helped immensely. Prints with PETG now work just as well as PLA.

Thank you for all the work you put into this, it's excellent!

Cheers

Thank you for the hint, I have corrected it.

Hey guys,

I have seen that the cooling down from printing temperature (60°C) to room temperature is the main driver for the deviation in my printer. Therefore I added a feature to compensate this effect.
If you want to compensate the temperature shinking I would not mess up with the calibartion of your printer. Instead use the recomended scaling factor in the excel sheet to scale each model individually in the slicer.

I noticed my prints are not fitting together correctly and I have attached my results below I am currently printing the 2 piece

Printer: JGAURORA A5
Slicer: Simplify 3d

Your results are actually normal for FDM printers. As any kind of manufacturing process has a tolerance and a design needs to concider these tolerances. If parts printed with your printer do not fit together I would say that the problem is the design and not your printer. Typically I foresee a gap of 0.2mm between parts if I want a tight fit and 1mm if they should not touch.

Noticed round holes were printing as ovals... Thought maybe my steps got somehow miss aligned... Printed your thing... It only wants me to adjust my Y by .006 steps and leave X the same... But it's also saying X has a deviation of .08 and Y of .19 which tells me circles will be ovals and dimensions will be off... If it were shrinkage I would assume those numbers would be more uniform... No clue what the problem is but at least now I can scale the model in X and Y to compensate

I'm running Cura 3.6.0 on a modded He3D H500
Thank you for the model

You mean a standard deviation or the mean deviaion? It sounds like your printer is well calibrated but you have some random effects in y. Check for some mechanical defects on the y axis. If you upload the excel file I may tell you more.

New to 3D printing and don't quite understand how to use the mean wall thickness deviation number in this calibration tool and whether I'm doing things right, so figured I'd post to 1) document as requested by lukeskymuh and 2) see if anyone wants to weigh in one how I'm printing so far for accuracy and precision and provide any comments and suggestions they would like to offer.

From what I'm understanding, my stepper motors are better calibrated so I can achieve more PRECISE parts. My printer just prints parts slightly larger than it is supposed to at the moment. It seems like I should also be able to compensate for the remaining wall thickness deviation, but I have some learning before I can get there.

Printer: Ender 3 Pro
Slicer: Cura v3.6.0
Nozzle size: 0.4mm
Layer Height: 0.2mm
(calibrated extruder previously, and running at E98.30. Left flow at 100% for now, but will be reading up to see how I need to adjust that)

Before Adjustment Deviation:
X = 0.29
(Mean Wall Thickness Deviation 0.33)
(axis steps = 80)
Y = 25
(Mean Wall Thickness Deviation 0.28)
(axis steps = 80)
z = 0.04
(Base Deviation 0.00)
(axis steps = 400)

AFTER Adjustment Deviation:
X = 0.03
(Mean Wall Thickness Deviation 0.34)
(axis steps = 80.37)
Y = 0.09
(Mean Wall Thickness Deviation 0.29)
(axis steps = 80)

The second Z print is still going so I'll update it when it finishes.

Thanks for the data.

The mean wall thickness can have different reasons like vibrations and over/under extrusion. Check the quality of your prints for ghosting and overextrusion and the diameter of your filament. The value alone only indicates that there might be something wrong.

Great tool! When I do my x-axis measurements, I notice, for any measurement, it was slightly larger at bottom than through rest of height.. .. like a slight puddle. I just measured at the middle of the piece and got good measurements. Is this an indication of anything correctable? Using Ender 3 and Cura.

It could be a mechanical issue, or just a coincidenence. If the effect is below 0.1mm it is the limit of the technology. If smaller than 0.3mm it probably can be tuned. If larger than 0.3mm the ist probably an mechanical problem. Overloaded belts are a typical cause.

Attached is poor photo. You can see slight bulge just above bottom chamfer. It adds .1 - .25 depending on which test point it is measured. It only does it along the Y-axis. I am attaching the Excel file also. I think I am happy with test results but am so new as to not know how good is good enough.

"Good enogh" is driven by your application. If you printing something artistic this normally doesn't matter. If you have mechanical applications I would try to stay below 0.1mm standard deviation and +-0.2mm max deviation because it will be very hard to get much better values. Currently I am collecting some values to detrmine what is typical for FDM.
Basically as lower the value are as better parts will fit, mechanisms will work and you have to forsee less gaps in your design.

The bulge is interesting. As the Z indicates that you have an constant offset in Z, I could imagine that it is linked to you first layer height. maybe try to increase nozzel distance (z offset) by 0.22mm.

Your Y is axis looks good but the x has a large deviation. You might compansate this by changing the steps in x but first I would check for mechanical issues (belt tension/damage?)

Won't a z-offset make the first layer extrude too far above the build plate to stick? Sorry if this is a dumb question. I'm new to this. Also.. by "max deviation" do you mean the highest value in any of the lines of the table? Can you explain meaning of "no offset" in the table? If there is documentation of this stuff, I'll go to it and not ask as many basic questions.

Results attached ;-)

Thanks a lot for this huge work !

Printer (Ender 3) and sicer (simplify3D) are noted on the X page)

Thanks, impressive result for the y axis. Was the printer calibrated? if not I assume the heatbed was cold, as I would expect a lager effect due to the temperature change.

Well here are my numbers. Not fully sure what Im looking at but its obvious that X and Y have a negative deveation but is it out of range? I do have issues with getting parts that are supposed to join togeather to fit properly. Ie. hole is to small for peg and am trying figure out why. I still have to recheck my Esteps to see if Im extruding to much. I posted the XYZ table images here https://www.thingiverse.com/make:611079 I also attached the excel to this comment.

Z
Standard deviation [mm] 0.12
base deviation [mm] -0.18

Y
Standard deviation [mm] 0.35
mean wall thickness deviation [mm] 0.14

X
Standard deviation [mm] 0.25
mean wall thickness deviation [mm] 0.22

Calibration cross

It means that your printer prints parts to small. If you fit only parts together printed with the same printer and direction it won't have an impact but a part printed in x will be nearly 1mm smaller than a part printed in z. This is in fact a quite large deviation. You can compensate it by changing the steps per mm for each axis (see the green cells). But i would first check some other causes: make sure the slicer has not scaled the part belts are well tensioned. But i would guess that just the pulley on the steppers is slightly to small. (Due to manufacturing low quality)
Your standard deviation could be smaller. So you will probably have problems printing parts with tight tolerances.
This is ok for a printer at the lower price segment. If you need more accurate printing: replace belts and pulleys and make sure that the frame is really stiff.

Comments deleted.

This is a grat design but I have a big problem with my calibration, hope you can help. I posted a make with values I get.

After calibrating the tension on my HEVO CoreXY I get a 20x20 cube I tried this calibration cross and after finetuning the steps/mm I still don’t get good accuracy.

Why would the “mean wall thickness deviation” be positive on X and negative on Y?
Same goes for the Multiplier calculation. According to this math I am overextruding on X and underextruding in Y.

The part feels good and even to the touch on the X and Y walls, beautiful bottom and top layer.

Printed 45* rotated in PLA. Acc of 2000mm2/s, jerk of 1200mm/min. Duet3D. Chinese 0.4mm nozzle on e3D V6 original, Nema17 high torque 0.9 on x and y from e3d. Calibrated Bondtech BMGextruder. Igus(metal casing) 8mm with aluminum rods on X. LM10UU on Y. Steel reinforced Gt2 belts. Precision Piezo Orion on Z.
55cm Capricorn bowden. Pressure advanced enabled on latest Duet firmware and tested with same result disabled.

Sliced with diffrent settings in latest Cura. 0.4mm line width or 0.48mm. 0.2 layer height, infill 80mm/s, 40mm/s inner wall and 20mm/s outer wall. 2 top, 2 bottom, 2 walls, 10% infill. fill gaps between walls and print thin walls enabled or disabled.

Same problem if sliced in S3D with similar speeds and settings but 0.48 line width.

What do you mean with not a good accuracy? Realistic typical standard deviations are around 0.1mm.
The “mean wall thickness deviation” is affected by many effects. If the value is below 0.15mm you can ignore it. It is probably just the limit of accuracy of your printer.
What is your "Mean relative deviation"?
If you whant you can upload the excel sheet.

Hi! I've posted screenshots of the Excel on my "I've made one".
My standard deviation is under 0.1 and I'm happy with it but the mean wall thickness deviation is +0.31 on X and -0.2 on Y.
What causes this? I am building the printer and wan't to get it as accurate as possible.

Thats really a interesting effect. My guess is that there are two effects and one acts only in one axis. Like: underextrusion and vibrations in X. Maybe check your belts and frame. Do you have ghosting effects on some prints? Your stepper calibration is perfect.

It was the steel reinforced belts. When I took them off they where crooked. Replaced with glasfiber reinforced and its much better. Will be even better when I get my e3d kevlar reinforced.

Thanks for the file! I'm not really sure if I need to make any changes based on my results. Most of my measurements were larger than they were supposed to be but changing my steps made the 5mm measurement differences even larger and it told me to move stuff the other way. So, I went back to stock (100, 100, 400). You guys are better at this than I am so I would appreciate any help you could give me.

Anet A8 with hulk braces and Cura 3.4.1

SD X = 0.09
SD Y = 0.16
SD Z = 0.09

You have already very small deviation values. So no need to change the stock values.
Therefore your printer is alredy nealy optimal and the values proposed by the excel sheet should be very similar to the stock values. Due to small inacuaries of the measurement it will always propose slightly different values. These inacuacies are the (realistic) limit of your printer. Don't be disapointed, these are quite good values.

If not: send a screenshot of the excel sheet.

Thanks! I was hoping I would find some problems. I've had difficulties printing multiple part models (like things with nuts/bolts) where the tolerances were too tight. I guess fixing individual prints is part of the fun of this hobby.

To tight tolerances are a problem indeed. I designed this mainly to have realistic numbers for tolerances so that no cleaning is necessary for mx designs. Fixing is something needed only for bad designs.

First, thank you for putting this together! This is outstanding and in my opinion, the only calibration tool that makes sense. I'm a little confused by the directions. Do I need to print the cross twice? Looks like one print in the layout in the STL, then a second print with the model rotated clockwise 90 degrees?

Also, I have my extrusion multiplier calibrated to 0.90. Is that going to throw off anything with this test?

Thanks Dreylok. You need only to print it once. Thats the reason it is a cross. In a first version (see "remix from") you had to print it twice.
The advantage of this aproach is that it is independant from the wall thickness (and therefore extension multipier) itself. But the "mean wall thickness offset" give an indication if you are extruduing to much or not enough. But this can be affected by other effects also. So I sould not calibrate is based on this value only.

Thank you for the quick reply! I knew you could get both measurements from one print, but the X and the Y tabs in the spreadsheet both say to print it in different orientations.

This is an amazing piece of work. Thanks again!

Oh thats from a old version. Thank you I will correct it.

Where are the spreadsheets? I didn't see them in the files I downloaded

They should be there. Click on thing files.

There are two spreadsheets, with the same same and same upload date :D Which one is the right one please?

Sorry for the late response. Both where the same, I uploaded it twice by accident. Now there is only one.

Thank you for a very useful and genius thingy, my printer was close but now spot on :)

Thank you fot this great procedure !
I made new pieces with a french version of the excel sheet and a complete tutorial for french users on my channel
https://www.thingiverse.com/rio31/designs

Great. It is good to see that it is usefull for many.

Hi,
Thank you for comment. I have a question, because I'm not an 3D printer expert. On another french forum, an expert said it is a wrong procedure because we should NEVER modify the steps/mm on the 3 axis X, Y, Z, but only calculate them for ever with the physical parameters of the printer : belts, motors, poulies ...) We have only to modify the steps/mm of the extruder for taking account the properties of the filament.

What do you think ?

For me , it is true the axis steps/mm are calculated the physical parameters, but, as for any mechanical system, we are never sure of the précision of this parameters, for example the size and characteristics of the belts which can vary during the movement.
For that, I think it is useful to adjust the physical printer parameters thru he firmware with your procédure.

Is it wrong ??

He is right and the theoretical steps per mm based on calculations are more accurate then calibrations with cubes. See also the video of 3D Tom linked.

But this test and the aproach in the excel sheet is much more accurate than these cubes.

So:
For good printers this test is a confirmation that it is accurate or perform minimal corrections due to manufacturing tolerances.
For not so good printers or you can improve the accuracy somehow with a calibration.

I printed and ran the calibrations on my printer, which is basically stock (less than 1 month old) with some safety additions, belt tensioners X,Y.

However, I have a question about the first (yellow) block on the spread sheet. Am I supposed to enter my own number in that block on the X,Y and Z sheets? If so, where do I find those numbers? I've only been 3D printing for less than a month but having great fun. I really like your approach to calibrating these machines. Thanks for any help-Tom

Here are my specs:
Filament: TacTink-1.75 (measured .070") Yellow
Glass bed - hairspray
Room temp 78F
Humidity 45%
Bed 60C
Extruder 190C
Extruder not calibrated
Cura 14.5 mostly default settings.
Nozzle .4

The axis setting (steps per mm): I you whant to change the setting than you need to enter our current settings. It is not needed to calcuate the accuracy / standard deviation. For many printers it is not needed to change this setting secially if the value of the standard deviation before and after correction (cell c9 and d9 for x) is similar.
You can get it when you send the g code M501 to your printer or in most of the firmware menus ou can find it also.

Can you post your achived standard deviations? (Cell C9 for x, c8 for y and z)

Thanks for the speedy reply.........

The cell C9 for X=0.17
cell C8 for Y=0.08
cell C8 for Z = 0.13

I tried to send my file filled in (of your spread sheet) earlier but the website doesn't accept them......maybe they want a screen shot.

These are normal values for a calibrated printer. I would say it is not worth to ajust the steps for your printer.

Thank you so much.......that puts my mind at ease. Have a great evening my friend :)
Tom

My test with a stock Anet A6:

Standard deviation is 0.14 for x and 0.13 for y.

Forgot, thank you very much for your work.

I performed this test.

Standard deviation is 0.13 for x and 0.08 for y.
Standard deviation for z is 0.03.

Printer is Anet A6 with Marlin 1.1.5 and various upgrades (belt tensioner for x and y, cable chains, y axis frame braces, y axis belt holder, better quality bearings for x and y axis, auto bed leveling, mosfets for bed and hotend).

Slicer used is Cura 2.6.1.

Thank you, this is very nice. I'll post my results soon

@lukeskymuh:
I updated my "I made this" with more information specific to my machine. Good job on this one; very helpful.

I printed this and measured. Standard deviation is 1.24 for both X and Y.

Printer is the Anet A8

Slicer is the Slic3r Prusa Edition 1.36.2

Thanks for a very useful tool.

It told me I need to change my steps from 100 to 99.97 so I probably don't think it needs adjusting ;-)

99.97! Nice, thats how it should be. But the standard deviation is quite high. Check if the measured values are in the right cells and wich values have the largest deviation.

Thanks. I will when I get home tonight.

One thing I wondered was in your spreadsheet for X & Y tabs it shows the same image. Am I supposed to turn the print through 90' and measure again? Because of the images I simply used the measurements for X & Y...

Also, on my printer, M501 did nothing much - I had to use M503 which then reported 100 steps/mm for both X & Y.

I had a typo in the top 25mm measurement column - I'd entered 20.10 when it should have been 25.10...

Now I get a standard deviation of 0.22 and am suggested to change steps from 100.00 to 100.34