Getting Better Dimensional Accuracy - Calibration

[Torgeir 242]

Yes, the easy thing first.. 👍

 

I've also seen a strange nozzle delivering a little more on one side..

 

Well, good luck with testing..

 

Torgeir

[my3DBr 4]

Woooowww @Torgeir, you just said something very very interesting.

You see, I installed the 3D solex kit matchless of heat block and the nozzle.

I noticed something when the printer does the initial purge of filament - it don’t fall down 90° normally. It travels like to the side first. As you know this nozzle has two internal channels.

Could it be it? 
Unfortunately I started to notice accuracy problems before changing it, recently. But never worried too much before.

I will send come pics.

[Torgeir 242]

Hi @my3DBr,

 

Well, the nozzle in general is important -as many other parts of the printing process.

I do not know much about "double/multi" channel nozzles, but for sure they are made to improve heat transfer to the filament flowing through the nozzle and this in order to print faster..  Higher speed will always degrade precision by some degree.

 

I'll think that if you want to have precision, you have to use the original nozzles and use the advised settings -"engineer setting".

 

How did the test go?

 

Thanks

Torgeir

 

 

 

 

 

[my3DBr 4]

Hello @Torgeir.

I was having more intensity at work this week, so less time for testing. I deed printed a cube at 40mm/s; 0,15 layer if I remember correctly ( the "normal"mode). No good, still having a big difference between Y and X - in this case X 19.9mm and Y 20.31mm - the overall size has got smaller as expected, but my concern is similarity between X and Y.

I'm baffled because I did measured the axis steps very accurately and they are ok, extremely ok - a needle glued in the printhead and a very good professional scale at the bed, moved the printhead with the scroll button in the printer. Everything pristine. All through 150mm. Sooooooo it's difficult  to assume hardware problems. I was hopping for it to be inaccurate.

I began to send emails to Bondgear to try to solve the feeder problem and I am trying to compensate this deviation between axis resizing the X axis at Cura. I am thinking about print one cube with Prusa Slicer but can't find a good GCode start.

[my3DBr 4]

I have news.

When adding 1,9% size to X, all sizes of the cube are equal. Amazing 20.36 exactly for both.

When subtracting 1,9% size to Y, sizes are exact X 20.00 and Y 20.03.

This shows I have a problem with the Y axis. But why when I run tests of steps they are ok? 

Now I have to print bigger files to see if this is a magical number.

Edited August 14, 2021 by my3DBr

[my3DBr 4]

Hello everybody.

It happens that those % diferences are not magical numbers. this diference works for 20mm cubes but not 40mm.

What I did noticed is that Y axis is 99,8% of the time 0.4mm bigger, no matter the size. I am still waiting for a crimper tool to arrive so I can change the stepper motor for the Y axis, but I will try for now to reduce the print in Cura for 0,4mm is all sizes to see what happens.

[my3DBr 4]

Hello again!

So it seems I nailed it. Removing -0.39mm from Y axis in every print at any size solved the problem.

I am really really curious to know what could be the reason for it. Broken firmware? I am using the last stable version of Tinkergnome - witch is not the latest version. Try the latest? Try the last oficial firmware for the UM2E+? Any ideas?

[SandervG 1,521]

Hi @my3DBr, thank you for your message and keeping us posted. I'm happy to read you managed to solve the problem. I understand you have done a few modifications to your Ultimaker, like bondtech feeder and a different nozzle - right? When you measured all the steps with a needle, the steps seemed to be quite accurate. But when you make a print you get an inaccuracy in Y. In that case, I would probably suspect it has something to do with the extrusion train? Not saying your modifications are causing it, but perhaps there is a relation. 

 

0.39mm is also almost the size of a nozzle, and you mentioned before that it looked like when you ooze it is being pushed out to the side? If i remember correctly, with a 'normal' nozzle, when a prime does not go straight down but to the side, this is often an indication that your nozzle is dirty. But when the nozzle has internal components they might behave differently. 

 

From testing we have seen that the pressure on the print head from the filament in the bowden tube is not universally the same, it is different on X then it is on Y. Perhaps with your Bondgear (Bondtech?) this is amplified, combined with your nozzle and allows for some over extrusion on Y? 

 

[my3DBr 4]

2 hours ago, SandervG said:

Hi @my3DBr, thank you for your message and keeping us posted. I'm happy to read you managed to solve the problem. I understand you have done a few modifications to your Ultimaker, like bondtech feeder and a different nozzle - right? When you measured all the steps with a needle, the steps seemed to be quite accurate. But when you make a print you get an inaccuracy in Y. In that case, I would probably suspect it has something to do with the extrusion train? Not saying your modifications are causing it, but perhaps there is a relation. 

 

0.39mm is also almost the size of a nozzle, and you mentioned before that it looked like when you ooze it is being pushed out to the side? If i remember correctly, with a 'normal' nozzle, when a prime does not go straight down but to the side, this is often an indication that your nozzle is dirty. But when the nozzle has internal components they might behave differently. 

 

From testing we have seen that the pressure on the print head from the filament in the bowden tube is not universally the same, it is different on X then it is on Y. Perhaps with your Bondgear (Bondtech?) this is amplified, combined with your nozzle and allows for some over extrusion on Y? 

 

Ok Sander, I will provide the nozzle change.

[bladerunner2205 0]

Hi guys, im having exactly the same issue with our new s5 with x and y differences. Im also using the printer for jigs, mostly cmm holding for measuring and x and y have a 0.4mm difference. Its not a scale issue ( steps/mm) and im struggling to find a solution. Printer is new and all original although we have several nozzles to hand all of which new. My main issue is the 0.4 mm difference between X and Y and no settings I can find to adjust as line width or horizontal scaling work in all axis at the same time! So please post up your findings and results and I'm all ears on possible fixes. Ive been printing at home with a £300 4 year old wanhao printer ive modified and calibrated and right now its running rings around the s5 when it comes to mechanical accuracy!!! Im wanting to fix the issue rather than try scaling the model in x vs y as that quite often will not work with any surface thats not 100% vertical etc. To say I'm a little disapointed with a £6k printer thats nearly 0.5 mm out of size is an understatement but I know ultimaker do make good stuff and I'm still hopeful its an easy adjustment as they seem to have built the printer well and its all square and trammed correctly with what seems pretty good steps per mm etc. But I'm definitely missing something and this difference is the same whatevwr material im printing with. I do add a bit of cooling for pla and especially tough PLA as the default cooler is not great with the lower print temps and this helps on sharp corners such as the cube but the x and y are still 0.4-0.5mm different!

[SandervG 1,521]

Hi @bladerunner2205, welcome to the community! What kind of objects have you been trying to print and measure? Could you perhaps share some photos? Let's perhaps just start at the beginning and verify the measurements. I'm not suggesting you're measuring it wrong, but it happened more than once in the past that someone made a print and overlooked the 'elephant foot' at the bottom which made it seem much larger as it is. 

 

Are X and Y both off by the same? 

 

And you say you have several nozzles to hand. Do you mean nozzles or print cores?

 

Thanks! 

[bladerunner2205 0]

Hi SandverG.  And thanks for your input. I've  done a few fixtures from 250mm long straight edges to internal bores and cylinders from 11 mm to 40mm. So i did a few benchys to look at the usual print settings temps and cooling performance and these seem to work best with basic plain PLA now. Then I started to print 20mm cubes to adjust again the usual speed and extrusion and print order priority settings. I'm making jigs for QA purposes so have to hand a good range of measuring equipment and having done 3d prints for a few years I'll always ignore things like elephant foot,seams,layer height zits and corner over shoot as thats going to be giving false information. Rather Ill dial the printer in for its media and printing enviroment first and then start looking at actual sizes internal and external. What im getting is a 0.4mm difference between x and y on a 20mm cube but also 0.4mm difference on an internal bore of say 40mm. So its not a step count or id expect that to increase the error in proportion to the print size. This seems pretty well dialed in from the factory tbh but Im not away how the end user could calibrate the steps on the s5 anyhow. Layer lines and size across the different heights all seem within 0.1mm at 3 different points , so Im a bit stuck on how to adjust tbh! Ive changed things like line width to 0.35 for a 0.4 print core. Oh and by different nozzles to hand I mean several print cores. 4x 0.4 aa and 2x 0.8 aa as well as a 0.4bb all new and Im only using genuine ultimaker fillaments such as pla, tough pla, petg, tpu and the pva types. Hope this helps 🙂 no photos at the moment but can take some sample shots of 20mm cubes and fixtures etc tomorrow. I hope the OP doesnt mind Ive jumped on this thread but its the closest Ive come to finding the same symptoms being discussed. Print core nozzles are clean and fillament leaves the nozzle vertically. I did manually check the bed for level this morning just so its not only relying on auto leveling to sort out any run out ( which it does rather well tbh!) 

Edited September 8, 2021 by bladerunner2205
Spelling mistake

[FRabe 5]

@bladerunner2205, are the X and Y axes square to each other? I took delivery of three S5 bundles last year, and two of them had issues with squareness. This could be affecting the sizes depending on how out of square they are. This can occur in initial shipping.

 

thing:3044638 on thingiverse is what I used to remedy this, and after adjusting the horizontal expansion and horizontal hole expansion in the slice, I have got down to +-0.1mm on X and Y, which after reading online is about as tight as you can get with FDM tech.

 

Another thing to check would be belt tension, just incase. https://support.ultimaker.com/hc/en-us/articles/360011529340-Check-the-tension-of-Ultimaker-S5-short-belts is a good guide for this.

 

Would love to continue to hear how your efforts go. Cheers.

[my3DBr 4]

For now I am having troublesome hours trying to square the axis. When printing squares, the size of the back of the print is different from the front side. I've already made 2 or 3 alignments for now....the tools were made from PETG, which is a bit flexible. Will try printing them with PLA.

Resizing Y size for 0,39 mm less solved the problem for now.

Still fighting, I am not giving up the fight for now. I have to win the printer...

Tiresome though.

[my3DBr 4]

On 9/8/2021 at 4:43 PM, bladerunner2205 said:

Hi SandverG.  And thanks for your input. I've  done a few fixtures from 250mm long straight edges to internal bores and cylinders from 11 mm to 40mm. So i did a few benchys to look at the usual print settings temps and cooling performance and these seem to work best with basic plain PLA now. Then I started to print 20mm cubes to adjust again the usual speed and extrusion and print order priority settings. I'm making jigs for QA purposes so have to hand a good range of measuring equipment and having done 3d prints for a few years I'll always ignore things like elephant foot,seams,layer height zits and corner over shoot as thats going to be giving false information. Rather Ill dial the printer in for its media and printing enviroment first and then start looking at actual sizes internal and external. What im getting is a 0.4mm difference between x and y on a 20mm cube but also 0.4mm difference on an internal bore of say 40mm. So its not a step count or id expect that to increase the error in proportion to the print size. This seems pretty well dialed in from the factory tbh but Im not away how the end user could calibrate the steps on the s5 anyhow. Layer lines and size across the different heights all seem within 0.1mm at 3 different points , so Im a bit stuck on how to adjust tbh! Ive changed things like line width to 0.35 for a 0.4 print core. Oh and by different nozzles to hand I mean several print cores. 4x 0.4 aa and 2x 0.8 aa as well as a 0.4bb all new and Im only using genuine ultimaker fillaments such as pla, tough pla, petg, tpu and the pva types. Hope this helps 🙂 no photos at the moment but can take some sample shots of 20mm cubes and fixtures etc tomorrow. I hope the OP doesnt mind Ive jumped on this thread but its the closest Ive come to finding the same symptoms being discussed. Print core nozzles are clean and fillament leaves the nozzle vertically. I did manually check the bed for level this morning just so its not only relying on auto leveling to sort out any run out ( which it does rather well tbh!) 

Any news?

Here I am still struggling with the same innacuracy.

I don’t know what else to do.

I bought all the moving pieces again from the same shop - belts and pulleys, the hole thing, and that will be my last try.

 

[my3DBr 4]

After a lot of tinkering, I took radicals decisions.

Spoiler: problem solved.

I took back all the old (and a little screwed up) pulleys (even buying ones with the exact same specifications, the original ones are bigger in diameter), re aligned everything up, the problem with Y bigger tan X persisted.

I took my dial gauge and started measuring steps. I realized why I did not noticed deviations before: they were little but present.

20mm in X were 19.88 and in Y 19.60. There you are. No escuses now: in practice all the schematics and mathematics were not valid. No "pulleys with X size and Y dents and belts with Z length equals N steps, so no deviation".

I took all the possible measures possible: firmware upgrades, all the mechanics revisited (dismantled everything up for 2 times). Changed nozzles a few times.

It was time for steps/mm reckoning. Done. Perfect squared cubes (although 20.6mm, witch I'll manage with horizontal expansion).

I am very sorry for you engineers and Ultimaker makers, but I had to do it. Changed de steps/mm for X and Y.

And I really don't know why:

A) I took me so long to realize it was just it

B) still believe in science and do understand all the logical mathematics involved - but....that's it.

 

What I can tell you is that is more valued for me to have a machine working than obeying engineering proud. I have 2 ender 3 that never gave me that problems, for a little percentual of the price. My ultimaker HAD TO WORK properly, period.

Thank you for your time and help. I expect this long story to hep others.

 

So, the flowchart for uneven X and Y prints

1-check for belts, pulleys. 

2-check for software/firmware problems

3-all the above are relatively simple, so don't be scare: go for measuring steps WITH THE RIGHT TOOLS, NOT WITH PRINTS if the above did not solved.

 

Be safe.