another UMS5 print accuracy problem thread

[ganadyne 2]

Hello,

 

I've been using a Prusa i3 mk3s for about 6 months making custom IC shipping containers and decided to upgrade to an S5 a little under a month ago. I've had issues with interlocking prints being oversized, which is pretty well documented around this forum. Most people recommend simply using a negative horizontal expansion value to bring the print size closer to the .STL size. While I have some success using a -0.1mm horizontal expansion value for outer dimensions, my model has some fine 0.4 mm wide lines on the surface which get completely removed by using this horizontal expansion value. I have to say, this is very disappointing. I will echo the many other reports I have seen of people running into XY dimensional accuracy issues with this printer and say that I am very surprised a Prusa mk3s produces much more accurate prints with less work. At this point I would not recommend anyone purchase the S5 if dimensional accuracy is important to them.

 

It seems like a software or firmware fix would be appropriate to solve this issue; does UM plan to address this issue, or are there any plugins that will adjust outer and inner walls only and leave single wall features alone?

 

Thank you,

 

Gabriel

[Cloud-Forge 3]

Having the same issue here. I'll have to find the right settings. I guess It'll take a little more work to achieve the accuracy I need.

Edited September 17, 2020 by Cloud-Forge
written on the spur of the moment, wasn't necessarily constructive

[SandervG 1,521]

Hi, we're sorry to hear that you are struggling to reach the level of accuracy with your Ultimaker that you are looking for. Perhaps it would help if you could share some specific elements that you need help with, instead of 'it is not good and not accurate'. 

 

We have an engineering profile that should give you pretty good results, and plenty of features in Cura that allow you to print outer walls before inner walls, or the other way around. There is horizontal expansion you can use, and speed settings you can optimise for optimal results. 

 

Plenty to configure and I'm sure with the right settings you can reach the accuracy you are looking for. 

[Cloud-Forge 3]

58 minutes ago, SandervG said:

Hi, we're sorry to hear that you are struggling to reach the level of accuracy with your Ultimaker that you are looking for. Perhaps it would help if you could share some specific elements that you need help with, instead of 'it is not good and not accurate'. 

 

We have an engineering profile that should give you pretty good results, and plenty of features in Cura that allow you to print outer walls before inner walls, or the other way around. There is horizontal expansion you can use, and speed settings you can optimise for optimal results. 

 

Plenty to configure and I'm sure with the right settings you can reach the accuracy you are looking for. 

 

Yeah sorry about that.. I agree it is not a good start. But I was feeling a bit let down after my few first prints.

The handling of the machine and the startup was close to flawless, so I had really high expectations for the printing results.

 

Now, I just got the printer less than a week ago, so I guess I'll have to find the right settings.

 

To answer your questions:

I use PLA tough (Ultimaker).

 

 

I used the following settings I have found on an other thread about accuracy (I did left jerks as default) :

Line width: 0.4
Wall thickness: 1.2
Top/Bottom thickness: 1.2
Speeds: 40
Jerks: 30
Horizontal expansion: -0.03
walls: 3 

 

Also I use Engineering profile with 0.1 height.

 

Now the pieces fit better,

but I still struggle to screw them together.

After I clean the burr line.

 

What are the right settings for tough PLA for best accuracy  ?

 

Thanks.

 

 

 

Edited September 17, 2020 by Cloud-Forge

[SandervG 1,521]

Thank you, by heart I think those settings you have listed are pretty much the same that are in the engineering profile. Without taking any specific details about your model into account, default engineering profile should provide you with the best dimensional accuracy. Now if your model has any distinct features you want to take some additional settings into account. 

 

What is the difficulties with screwing? Do you have difficulties of driving the screw in? I imagine you want to have some friction to ensure a tight fit, but if it is too much you could look into the 'Hole horizontal expansion' feature. Similar as horizontal layer expansion, you can apply a factor to each hole in your model. 

[Cloud-Forge 3]

Thanks you.

 

After some tests, trials and errors, I can now print whit the accuracy I need !

 

The settings I use are the one above with jerks at 20, retraction at every layers. 

I use the engineer profile at 1.5.

Now my prints don't expand. 

The settings that help is Horizontal Expansion : -0.16

(For Ultimaker PLA Tough and Ultimaker PLA)

 

Hope it help others as well.

 

I got scared for a while,

but the problem was the guy in front of the printer not the machine.

 

 

Edited September 18, 2020 by Cloud-Forge

[Smithy 1,144]

2 hours ago, Cloud-Forge said:

but the problem was the guy in front of the printer not the machine.

🙂 - let's better say it was a mix of a lot of options in Cura with difficult requirements.

 

Thanks for the feedback, and happy that it is now working for you!

[ganadyne 2]

Hi All,

 

Here is an update: 

 

The rounded corner issue is due to the bowden feed system and can be remedied by using the correct print settings. This is pretty well documented around the forum and elsewhere online.

 

We had issues printing fine single wall details (0.5mm wide single layer line features on a surface) where the printed lines had extremely variable widths. I suspect this is also caused by the bowden feed system not regulating pressure in the print head well. We remedied this issue by using a 0.25mm print head which was adequate to get the single wall feature accuracy we needed.

 

In an attempt to fix the various and hard to understand dimensional accuracy issues which made interlocking parts not fit together, we meticulously modified and test printed our model file one line width at a time so that it would print patterns that do fit together. After this was accomplished and some test prints had the correct printed dimensions, we rotated the model 90 degrees on the print bed only to find that the model has systematically shifted its dimensions. We get a distortion of up to 0.75mm between X and Y oriented prints for a dimension that should be approximately 100mm using the ESD PETG required for our application and tech support recommended print settings, and a systematic 0.5mm distortion using toughPLA with Ultimaker engineering profiles. Printing the same part on different parts of the print bed also results in variations in dimensional accuracy, but we have not done extensive characterization of this effect.

 

I suspect the significant distortions between the X and Y axes were causing a lot of our headache when trying to tune an interlocking print model on the S5.

 

After doing a lot of test prints and debugging with Ultimaker support, they have determined that our printer is working as intended. For a comparison, we have printed over 600 of these interlocking parts on the Prusa we used as a proof of concept machine with few problems and predictable model file vs measured print dimensions. We wanted to upgrade to a printer with easily replaced modular components and better build quality, but these tolerance issues are enough to dissuade me from moving our printing program onto the S5 platform. So, I am returning the S5 and taking a hit on return shipping and a restock fee.

 

There is a lot to like about the S5 - the modular print heads, large print area, robust cable management, nice enclosure and air manager system, etc. If you do not need very precise and accurate prints, it may serve you well and live up to it's reputation. Unfortunately that isn't the case here and we did end up spending a lot of time and money with little to gain.

 

As an afterthought, I do wonder if printers with large build area tend to have more XY accuracy problems generally. There is another thread here where users with a lot of Ultimaker experience claim the smaller format printers have better accuracy and I imagine the S3 and S5 have a lot of design similarities.

 

Gabriel

Edited September 18, 2020 by ganadyne
just added an afterthought

[jwsocool 0]

I'm having the same issues as described above.  Same parts different areas of the bed different results.  I'm very unhappy and wish I would have selected a different printer.  From reading the forum I believe that the S5 is not worth the extra money.  There is something wrong with the design.  I have to do a lot of testing using the scaling tool with the part in a fixed orientation.  My friend has a 600 dollar printer at home that performs better at dimensional accuracy.  I don't recommend this printer to anyone.  The x axis is always smaller than the y.  Turn the part 90 degrees the x is still smaller.  Two identical parts two different sets of results.  I'm talking about .5 mm differences.  I want to make prototypes that turn into production.  I can't be customizing the models just to get a decent print.  The Material Station and Air Station work well but the printer is below average.  I can't count on getting a good print and have to print the same part multiple times to get a result that is good enough.  Again stay away from the S5.

[Cloud-Forge 3]

How long have you got the printer for ?

Because my initial reaction was about yours,

but after some trials and error I can now print very accurately. 

Cura is really a deep piece of software and the S5 a good machine. IMO.

As such they take some time to be set right.

 

At the difference of a cheaper printer that might be good at first, but hard to evolve upon. that's my experience anyway.

[jwsocool 0]

A few weeks. 

 

My primary requirement is to producing working prototypes before committing my designs to traditional fabrication.  It's not a good workflow for me to create models for printing only.

 

I'm printing a test part 50 mm x 50 mm x 25" tall.  No mater the material and using the using Ultimaker profiles. The X is always smaller than the Y.  The only solution is to play around with the scale tool in one axis (~.7%) to get a square part.  It is very consistent  It's firmly my belief that the factory calibration of stepper steps is incorrect.  Using scaling in X by .7% and the other expansion settings I've been able to make better parts in the center of the bed.  I also can't print multiple parts of the same item as they will never match unless I keep the same orientations.  Again I'm very disappointed but need to move on.  On balance I like the printer and am resigned to scaling every part in one direction.  I worked with support and although they were friendly and helpful there is not much they can do about a poor factory set-up aside from the hassle of returning the machine.  I write this mostly to warn others to not consider the S5 if you expect good dimensional reliability.  From my experience and others on the forum this seems to be a accurate assessment (including a co-worker with a 600.00 printer at home).  The Material station and Air Station work well and are a great features.  The belts are tight as are all the shafts.  I realize the materials make a difference as do the settings.  I've been using measuring tools since 1977 starting out as Tool and Die Maker, machinist, cnc machinist and now a design engineer know full well that I can't apply the same standards to a 3D printed machine.  I'm not measuring in the corners but in the most stable part of the print.  I do believe that I with a square part I should get better matching results (.35 mm over 50mm).

[SandervG 1,521]

Hi, 

 

Thank you for your comments and sharing your experience. A small offset can be expected, depending on the material, size of your print or other custom configurations you might be able to influence the accuracy slightly. 

 

On 12/27/2020 at 5:54 PM, jwsocool said:

I do believe that I with a square part I should get better matching results (.35 mm over 50mm).

 

I believe that too, it sounds like it might not operating to its fullest capabilities. 

[Jasonjason 6]

The answer to prove it’s the stepper calibration is simple.  Increase the size of the square- say 250by 250mm.  If the error increases proportionally, it’s a stepper calibration issue.

if not start looking elsewhere,  like where the start point for each later is... this can be a big effect, as the first bit feeds out more, and you ended up with wider extrusion.