S5 Pro destroyed itself

[richardstaley 0]

Does anyone know if the S5 is supposed to have some sort of failsafe in case of clogging so that the machine doesn't tear itself apart when the filament sticks on the nozzle?

[IRobertI 521]

It does not. The S7 has an added sensor that can detect if the door opens up to help a bit with this.

[chekkefr 2]

The bed camera on most printers would normally do that (check the model looks like it should at ever stage of the build and do 'spagetti detection'). It's trivial to write a computer vision script to detect that (an undergrad could write one in a day or two) and the s5 has a bed camera but ultimaker haven't implemented it for some reason (unlike pretty much every other manufacturer).

 

One of many aspects about ultimakers that I've found frustrating.. 

Edited March 22, 2023 by chekkefr

[ahoeben 2,012]

5 minutes ago, chekkefr said:

The bed camera on most printers would normally do that

"Most printers" don't even have a camera.

 

  

4 minutes ago, chekkefr said:

It's trivial to write a computer vision script to detect (...)

 

I don't think you know what the word "trivial" means.

 

I know that there are printers and services that use trained models to detect failed prints. Training such a model may be "trivial", but getting the required amount of relevant data to feed the model is not.

[chekkefr 2]

4 hours ago, ahoeben said:

"Most printers" don't even have a camera.

 

  

 

I don't think you know what the word "trivial" means.

 

I know that there are printers and services that use trained models to detect failed prints. Training such a model may be "trivial", but getting the required amount of relevant data to feed the model is not.

 

So I'm a 4th year computer science PhD specialising in making assistive technology for the blind using computer vision so when I say it's trivial, it's trivial. 

(Note I got these pictures from google, we have eduroam so connecting to them locally isn't something I'm willing to spend time doing. So i cant confirm they're from the s5 camera but the consistent position makes me think they are). 

So you'd do edge+contour detection on the nozzle (area marked in red) or even template matching (you can use basic cv really doubt you need a neural net for something this basic, if you do roboflow is very easy and fast to prototype on). Adjust the cropping area as the print head moves using the g-code. Check for blur on the camera (as in second image). You can detect if the nozzle is jammed, generating spagetti or extruding into the case as above. You can also confirm that the print bed is cleaned and previous prints removed (again using some basic cv) before starting a new one (another reason we dont allow jobs to be sent to the printer remotely). You could perform input shaping to ensure faster and better quality prints by tracking the movement of the print head depending on the max fps of the camera. All that is very light weight and could comfortably be run on device or through cura. As long as you have access to the video stream I feel very confident a second/third year CS undergrad could write and debug that code in hours if not a day. If they couldn't I'd start asking questions. There are also endless open source and even github repos which implement something no doubt far more robust.

That's what i mean by trivial (for a coder or anyone with any background in cv). I am genuinely disappointed that this has devolved into some bizarre youtube style argument. I hope you start taking feedback on board and catch up to the rest of the field again. 

 

Edited March 22, 2023 by chekkefr

[JollyOrc 0]

I can only give my sympathies, had the same thing happen to me a year ago. The good news: It is mostly fixable, when you apply a lot of patient prying and gently heat the blob around parts to get it off.

 

Take apart the head to get to the different parts easier. The main difficulty will be getting the hinge work nicely again. It might keep a tendency to fall open at unopportunate times, which is the main culprit for this issue AFAIK.

[Dustin 175]

5 hours ago, chekkefr said:

That's what i mean by trivial (for a coder or anyone with any background in cv). I am genuinely disappointed that this has devolved into some bizarre youtube style argument. I hope you start taking feedback on board and catch up to the rest of the field again. 

I understand your frustration.

But please do not attack our Community Members its uncalled for.
Until my response not a single UltiMaker employee has responded to this thread, everyone talking here are Community Members and UltiMaker 3D Printer owners just like yourself.

You are more than welcome to share your feedback with UltiMaker directly here: https://support.makerbot.com/s/contactsupport

Additionally to my knowledge (sorry its limited im not a programmer, my background is in engineering) there is not enough processing power built in to the S-Line machines to support local "spagetti detection". Additionally speaking with my personal network of 3D Printing users and Enthusiasts. Everyone that has been using 3rd party solutions for this.. has been mostly hit and miss.. the Ai detection is just not good enough yet even among the paid 3rd party solutions and results in many false reports leading most of them to disable or abandon the detection completely because it was more issue than just letting a print fail.

So it very much is not Trivial as noted by ahoeben

[chekkefr 2]

2 hours ago, Dustin said:

I understand your frustration.

But please do not attack our Community Members its uncalled for.
Until my response not a single UltiMaker employee has responded to this thread, everyone talking here are Community Members and UltiMaker 3D Printer owners just like yourself.

You are more than welcome to share your feedback with UltiMaker directly here: https://support.makerbot.com/s/contactsupport

Additionally to my knowledge (sorry its limited im not a programmer, my background is in engineering) there is not enough processing power built in to the S-Line machines to support local "spagetti detection". Additionally speaking with my personal network of 3D Printing users and Enthusiasts. Everyone that has been using 3rd party solutions for this.. has been mostly hit and miss.. the Ai detection is just not good enough yet even among the paid 3rd party solutions and results in many false reports leading most of them to disable or abandon the detection completely because it was more issue than just letting a print fail.

So it very much is not Trivial as noted by 

Ok I misread your response. I’ll reply tomorrow. 
 

Edited March 22, 2023 by chekkefr

[ahoeben 2,012]

@chekkefr, I did not mean to descent to youtube-level comments, but I can see how you could read mine as such. I will try to do better next time, but I think for now enough has been said on the matter.

Edited March 23, 2023 by ahoeben

[richardstaley 0]

On 3/22/2023 at 8:57 AM, JollyOrc said:

I can only give my sympathies, had the same thing happen to me a year ago. The good news: It is mostly fixable, when you apply a lot of patient prying and gently heat the blob around parts to get it off.

 

Take apart the head to get to the different parts easier. The main difficulty will be getting the hinge work nicely again. It might keep a tendency to fall open at unopportunate times, which is the main culprit for this issue AFAIK.

Unfortunately, it sheared the tabs that hold the lower fan hinge onto the printhead.. and tore the wires to the capacitive sensor board..