Notifications
Clear all

Prusa i3 mks3+ resonance profiles (input shaper)  

Page 1 / 2
  RSS
MarschallMarc
(@marschallmarc)
Active Member
Prusa i3 mks3+ resonance profiles (input shaper)

I just did some resonance measurements on my klipperized but other than that stock i3 mk3s+. Semi stable table, printer sitting on a very heavy stone plate, well-tensioned belts (on the high side). The table is no issue, I got other printers measured on the same one with very good values.
As I have not seen any graphs so far I thought I'd just drop it here for future reference. Not exactly impressive...

Posted : 11/10/2021 9:12 am
Hanzov69 liked
Dan Rogers
(@dan-rogers)
Noble Member
So fix it!

My MK3S is nearly silent, and my MK3S+ is a whisper.  What's your excuse?

Posted : 11/10/2021 9:27 am
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:
funny guy

In addition to your aggresivness, you show an impressive lack of understanding, what I just showed you.
This has absolutely nothing to do with a printer being silent or loud.

Posted : 11/10/2021 9:45 am
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:
maybe eat a snickers

If you are not familiar with technical stuff, there is no need to be that aggressive. Either skip the stuff you do not understand or try to google (there are more than enough keywords in my post and in the plots) or even ask.
On a somehow related topic: does this forum have a block or ignore function for specific users?

Posted : 11/10/2021 9:55 am
Neophyl
(@neophyl)
Famed Member

No the forum does not have a block or mute function as far as I am aware. Sometimes I wish it did. 

The problem is that you are the only person I know that looks like they are using klipper with a prusa. I am aware of klipper as I’ve toyed with building a Voron 2.4 as my next printer and would probably try klipper on that. 
Personally without doing a lot more research the graph is meaningless, what does a good one look like compared to that etc. You might be lucky and find one person but the vast majority will just be using standard prusa firmware on mostly stock machines. 

Posted : 11/10/2021 10:46 am
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:
RE: Prusa i3 mks3+ resonance profiles (input shaper)

The thing is, this graph is valuable information for those who can read it and I could not find one for the i3 so far. All I wanted to provide is information for those looking for it in the future. While I used klipper to obtain the data, it is not limited to using the information with klipper. It gives general information about the mechanical capabilities (and limits) of my i3 (which is build and seated well), with a good chance of more general applicability to other i3. For example, it shows that reducing the print head weight will most likely allow faster accelerations. It is all in there.
I would have been happy to explain what you can read from it as well, just not when directly attacked like that.
I come from a background in science and when I obtained data I think of being interesting for others, I usually provide them to the public, which does not mean, that everybody in the world has to be able to make use of it.
edit: https://github.com/PrusaOwners/prusaowners/blob/master/Klipper.md

This post was modified 2 months ago by MarschallMarc
Posted : 11/10/2021 10:56 am
karl-herbert
(@karl-herbert)
Illustrious Member
RE: Prusa i3 mks3+ resonance profiles (input shaper)

Hello and welcome to the Prusaforum!

Interesting topic...

I would be interested in the topic of input shaping, as I have a Duet2 board installed on one of my MK3 and have already experimented with it there. One of the next projects is also a MK3 on which Klipper runs so far (original Einsy Rambo without MMU2), only there I have no experience on the subject of input shaping. Which software did you use for the evaluation, how did you implement the corrections in Klipper according to the table and which experiences did you make with the corrections?

As I said, I have currently only been able to test this on a Deut2 with the DWC gui. It's certainly possible to achieve improvements in the printed surface quality. However, the topic may be new territory for many, but it's an additional possibility for fine-tuning.

Thanks!

wbr,

Karl

Posted by: @marschallmarc

I just did some resonance measurements on my klipperized but other than that stock i3 mk3s+. Semi stable table, printer sitting on a very heavy stone plate, well-tensioned belts (on the high side). The table is no issue, I got other printers measured on the same one with very good values.
As I have not seen any graphs so far I thought I'd just drop it here for future reference. Not exactly impressive...

 

Statt zu klagen, dass wir nicht alles haben, was wir wollen, sollten wir lieber dankbar sein, dass wir nicht alles bekommen, was wir verdienen....
Posted : 11/10/2021 11:50 am
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:

Which software did you use for the evaluation, how did you implement the corrections in Klipper according to the table and which experiences did you make with the corrections?

I used only the inbuild software of Klipper. They provide a python script using NumPy to analyze and mathplotlib to visualize the results. It is really just one command you need to run.
Of course, you need to add the adxl345 first. The Klipper documentation has a guide on how to do this which you can follow (of course there are many other ways too).
The difference with and without input shaper is always huge. IMHO, this is a game-changer, I use it on every printer.
The Prusa is not a printer that sees a lot of work from me, it is more sitting on the sideline, but I now at least was able to up the accelerations to 5000mm/s^2 (but it creates visible smoothing) which is a great time saver.

Posted : 11/10/2021 2:06 pm
Jimbo70
(@jimbo70)
Trusted Member
interesting

I would like to use Klipper instead of the Marlin. Is it difficult to convert?

...
Posted : 11/10/2021 4:18 pm
karl-herbert
(@karl-herbert)
Illustrious Member
RE: interesting

To install Klipper you have to change bootloader and FW. Setting and configuring the parameters takes a lot of time and patience. You can use templates (especially the Prusa FW settings), but in the end you have to find the individual optimal settings for each printer. Some users in German-speaking countries including myself are already working with it. But I am constantly optimizing. Input shaping is a special chapter. So far I have documented most "How to install Klipper" processes, but only in German.

I have not yet specified a graphical user interface. But for now I work with Octoprint and a special plugin.

Klipper is something for hobbyists and tinkerers - not a finished product, as well as the Prusa FW.

wbr,

Karl

Statt zu klagen, dass wir nicht alles haben, was wir wollen, sollten wir lieber dankbar sein, dass wir nicht alles bekommen, was wir verdienen....
Posted : 11/10/2021 7:17 pm
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:
RE: Prusa i3 mks3+ resonance profiles (input shaper)

I would like to use Klipper instead of the Marlin. Is it difficult to convert?

It depends. If you use the prusaowners klipper fork and follow their docu you can make the switch in 2 hours.
If you are familiar with everything you can use something like fluiddpi and set it up in 3h.
If you have no idea about it and want to go from scratch, it can be a very long journey.

 

To install Klipper you have to change bootloader and FW. Setting and configuring the parameters takes a lot of time and patience. You can use templates (especially the Prusa FW settings), but in the end you have to find the individual optimal settings for each printer. Some users in German-speaking countries including myself are already working with it. But I am constantly optimizing. Input shaping is a special chapter. So far I have documented most "How to install Klipper" processes, but only in German.

I have not yet specified a graphical user interface. But for now I work with Octoprint and a special plugin.

Klipper is something for hobbyists and tinkerers - not a finished product, as well as the Prusa FW.

wbr,

Karl

I did not have to change the bootloader on my Einsy Rambo (I bought my i3 mk3s+ from Prusa, so no clone). I was able to flash it via USB without any preparation. It seems like there are different revisions out there, as others reported to need to flash a bootloader first.
Prusa put a lot of work into finding good standard settings for Marlin and you can use those (and add something like 0.045 PA) in klipper as well. This way your printer will already work as good as it does with Marlin settings wise (and some things better others different). It is an inherent property of 3d printers that optimizing settings for individual machines gets better results. 

Klipper is something for hobbyists and tinkerers - not a finished product, as well as the Prusa FW.

I very much disagree on that one. It is in fact much better in a farm environment than Marlin is and allows very easy central management.
The difference simply is that nobody prepared klipper for the i3 with the same dedication as Prusa prepared Marlin. The code base of klipper is also much more professional than the Marlin base. 

My goal was not to convince anybody here of klipper, I really just wanted the graphs to be available.

This post was modified 2 months ago by MarschallMarc
Posted : 12/10/2021 8:32 am
Jimbo70 liked
Jimbo70
(@jimbo70)
Trusted Member
thanks for the information

I do not have the background to get started so its sounds like many long hours for me. I think i will wait till it is more mainstream.

Your charts are interesting but way over my head. Even though your charts are in hertz it seemed apparent to me that you were not talking about audible sound.

...
Posted : 12/10/2021 12:19 pm
karl-herbert
(@karl-herbert)
Illustrious Member
RE: thanks for the information

I am a mechanic and have not much idea about programming and electronics. Nevertheless, I have started to work on the project Klipper. For this I built another MK3 and first of all I got a lot of information from the internet. The opinions, instructions and experiences to Klipper are very different and contradictory, which can be somewhat difficult for a beginner. It took me a weekend to get the printer up and running, but there's still a lot of tweaking to be done. In any case, it is exciting and you learn a lot. However, I did not touch my original, 4 years old first Mk3/MMU2. It runs very well in its original condition. In the next few weeks I will start test prints and compare the results with the original MK3. With some experience in programming everything is certainly easier, but it is also possible to carry out such a project without any special previous knowledge. You can't break much and the original Prusa firmware is quickly reinstalled.

wbr,

Karl

Statt zu klagen, dass wir nicht alles haben, was wir wollen, sollten wir lieber dankbar sein, dass wir nicht alles bekommen, was wir verdienen....
Posted : 12/10/2021 4:51 pm
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:

 

Posted by: @karl-herbert

 You can't break much and the original Prusa firmware is quickly reinstalled.

A very important fact indeed

Posted : 13/10/2021 6:47 am
Stefan
(@stefan-3)
Eminent Member
what do the two diagrams tell?

As I have not seen any graphs so far I thought I'd just drop it here for future reference. Not exactly impressive...

I assume these diagrams show at which frequencies the entire printer structure has resonance frequencies?

So what do the two diagrams show ? The axes are labeled identical in both diagrams, the diagram title does not tell me the difference...

But in both diagrams the solid lines have peaks at very different frequencies?

And what are you not very impressed about?

Posted : 13/10/2021 10:05 am
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:

I agree, the plots are not very well described if you do not know them (input shaper users would do so). Might be worth to do submit a pullrequest for better plot titles.
Those are resonances on two different printer positions when shaken at specific frequencies (which also means different accelerations, as the amplitude stays the same).
The first one is with x-movement (measured on the print head), the second one is with y-movement (measured at the bed).
Why are they not impressive? Well, they show that the printer cannot handle a lot of acceleration (peaks are already at quite low frequencies) and even with input shaping slight ringing will still be there. It also shows movement in z-direction while "shaking" in x or y, which means the whole printer is vibrating a lot. The stock firmware does not allow a lot of acceleration, and this graph shows why this is the case.
Still, input shaping significantly improved my print quality in this case.

Posted : 13/10/2021 6:51 pm
JensH.
(@jensh)
New Member
Input Shaper

Hello MarschallMarc,

interesting topic. I have some issues with InputShaper at my Prusa and would like to compere some results.

is there a way to contact you? Im am not allowed to send you am message here, why ever. Discord maybe?

Posted : 28/10/2021 7:35 am
MarschallMarc
(@marschallmarc)
Active Member
Topic starter answered:
sure

jens.hubenthal, you can go to the RatRig discord to find me, I am pretty active there.
There will soon be news about Klipper on Prusa from our side as well, making it much easier to set up in future.

Posted : 28/10/2021 7:40 am
Helmuth
(@helmuth)
Active Member
RE: I thought it was interesting

I was taken aback by the initial aggressive response, I thought the data was interesting, even if I don't understand it properly.

Also looks like you have already 'fixed it' using Klipper/Shaper, so took the post to be a provision of information. I do thank you for posting it, and would like some expansion.

Some questions:

  • Can you explain the Y axis- what does it represent?
  • Why are there two graphs, what's the difference between them?
  • What speeds were involved, and does changing speeds markedly change resonance peaks?
  • What's the 'real world' impact of this, do you think? If it's sound, low frequency stuff in the <200Hz range is not audibly irritating. Any sound my printer makes, it's really the higher frequencies that I find irritating. I'm thinking this might be more relevant to print quality, albeit at a very level so fine as to below my threshold of interest.

My quick and dirty takeaway from this, which could be totally off base, is that most resonance comes from X and Y movements, which isn't surprising (as I would assume for starters that Y movements are lots of mass, and Z movements are simply slower because of the lead screws)

Posted : 29/10/2021 2:54 am
Helmuth
(@helmuth)
Active Member
Ah, most questions already answered

By subsequent posts. Thanks for that follow up.

 

Posted : 29/10/2021 4:32 am
Page 1 / 2
Share: