Marlin - Input Shaper - Improve Your Prints - Chris's Basement - 2023
HTML-код
- Опубликовано: 2 июл 2024
- Today we continue Input Shaping Series with Marlin Firmware.
00:00 Introduction
01:37 Overview
02:48 Slicing Your Test
06:53 Firmware
08:52 Baseline Print
10:25 Calibration Print
13:50 Frequency Number
18:24 Results
20:55 Conclusion
Helpful Links:
🔗🔗🔗🔗🔗🔗🔗
Marlin Input Shaper:
marlinfw.org/docs/gcode/M593....
Helpful videos:
🎞🎞🎞🎞🎞🎞🎞
Marlin 2.0 Video:
• Marlin 2.0 On 8-Bit Bo...
SUPPORT THE CHANNEL BY CLICKING THESE LINKS!
E3D Affilate Link:
e3d-online.com/?aff=32
OpenBuilds - Great CNC and 3D Printer Parts (aff)
openbuildspartstore.com/?ref=...
Slice Engineering - Mosquito Hotends (aff)
www.sliceengineering.com/?p=r...
Wham Bam Systems Flexible Beds (aff)
bit.ly/2HoS1cV
Amazon (aff) [US DE GB CA IT ES FR JP]:
www.amazon.com/?tag=chrisbase...
Banggood(aff) ban.ggood.vip/7vm3
Aliexpress(aff) s.click.aliexpress.com/e/UrRbCm8
Buy me a coffee to keep me going!
buymeacoff.ee/Wc0DouFbl
You can support my work on Patreon
/ brotherchris
One Time Tips, You Can PayPal Me:
www.paypal.me/ChrisRiley3D
----------------------------------------------
Find Me Over On Twitter!
----------------------------------------------
Twitter : / chrisriley3d
FTC Disclaimer: A percentage of sales is made through Affiliate links
As an Amazon Associate I earn from qualifying purchases. 
Наука
Glad to see Marlin has implemented this, Klipper and Marlin help each other to improve and the community benefits.
I agree Dave. Working together in the name of science.
I must say that this is - for me personaly - much better explain video than same topic seen on Teaching Tech channel. Thank you! 👍
Glad it was helpful! Thanks for watching
Chris the only person in the universe that explains and we understand ❤
Thank you!
That's true for me too 🤞... Thank You Chris 🤝 and keep up
Thank you very much, this is far out the clearest tutorial for input shaping with marlin.
Thank you!
Thank you for the informative videos! Don't understand how you don't have 10x your current number of subscribers.
Thank you!
Fantastic explanation, as always. You clearly have a great understanding of the concepts. Thanks!
Thank you!
Oh so perfect, I just rebuilt a Anet A8+ and upgraded to the latest Marlin. Left input shaper off as I was going to wait for a video like this... THANKS!!!
Woo Hoo! I love it when a plan comes together! Thanks for watching.
As usual, great content Chris. I guess I was wating for this video to change my printers from 1.9! Brilliant.
Glad it was helpful!
Always the best and complete videos step by step, thanks.
Thanks for watching!
Thanks, Chris! Going to give this a try on my SV06 before tackling the Klipper change.
NICE! Let us know how it goes!
Very nice you rock Chris
Thanks!
thank you for the video, ended up eddeting the -Professional Firmware for 3D Printers- and it freaking works, you made me happy!
Great to hear! Thank you!
THx M8 for you wisdom !!!! Very good Videos
Glad you like them! Thanks
Awesome video as usual Chris thank you very much!! 🥳
Thank you!
@@ChrisRiley big improvement!!!
Great video and content as always Chris :)
Glad you enjoyed it
Glad you enjoyed it
Interesting to test, thanks
Thanks for watching!
NICE! Th3d have a calculator at there homepage.
Great explaind!!
Thank you!
Great, thanks for sharing about that topic, I use a Kobra 2 Anycubic with Marlin Firmware :) 👍
You're welcome, thanks for watching!
Chris: You're my go-to guy when I make any changes to my printers. Recently I installed Marlin 2.1.2 on both my printers and a BTT SKR Mini e3 V3 board in one of them. I've got everything working well, but now that Marlin offers so many options and Cura gets more and more complicated I worry about conflicting settings.
Some day I'd like you to do a video about how to configure (or turn off) the jerk settings in Cura, linear advance, input shaping, coasting, ironing, and all of the other nice bells and whistles which may or may not play well with one another. Do some of these options replace others or do some of them compliment each other? I'm a little confused and I'm not sure I have really optimized my printer. Or perhaps I should say, I may have optimized a particular feature but now I worry about negating it by making the wrong selection in my slicer, adding the wrong g code to my start script, or causing some other conflict.
Thank you! That's a really good idea. It would be good to know which setting conflict with which ones. In Prusa Slicer they have a way to ignore lots of them, CURA should really do the same. I will look at it further.
Thanks Chris! It will be interesting to see what advantage (if any) Input shaper has on CoreXY machines (that's all I have).
Yeah it will. Thanks for watching
Is it going to be weird seeing this video embedded on the same M593 webpage that is featured in the video? I know I always find it a little weird to see the G-code pages that I composed in videos and articles. Anyway, great overview, and I look forward to seeing how the next version of Marlin (2.1.3) is received, with a new Fixed-Time motion system and new Input Shapers. I'm working hard to get that together and out soon, a release that is meant to improve upon all we did for 2.1.2 and be more reliable and bug-free than any previous version. The whole Marlin community has been going above and beyond, and I gotta say I'm really proud and grateful to everyone who has helped us to keep moving forward, and at higher speeds with less ringing!
Thank you Scott
Now I REALLY need Creality to release their Marlin config for the Ender 5 s1!
Lets hope a lot of companies take advantage.
Really well done, thank you. Does turning linear advance back on after testing and using you new x,y numbers mess anything up?
It shouldn't but if you are testing a really high accel rates, you might see some inaccuracies on you prints.
What I would also like to see is slicers avoiding a resonance frequency on solid infill. Now when you have a square that has to be filled at 45 degrees with a corexy, the head will first move back and forth very quickly as near the starting corner the lines it needs to lay down are pretty short, but as you get closer to the center of that square the lines get longer and hence the frequency of the head moving back and forth is lowered. With the correct length of the infill lines this will hit a resonance frequency and cause unnecessary shaking of the printer, but as the lines further grow, this shaking drops again. With more complex geometry, the area hitting this frequency can be very annoyingly large. This is not the shaking input shaping is avoiding or has to avoid. The better method here would be if the slicer just adjusts the speed slightly based on the length of the infill lines to avoid the resonance frequencies. Even speeding up could help here, just avoid hitting that duration of straight line that matches the resonance frequency.
That's not how this works. Just lower your acceleration during infill, that does exactly what you describe... slowing down shorter movements because it's accelerating slower and won't reach high speeds in the small infills. However, then you're also negating all the benefits of having a corexy.
@@csdstudio78 No, if high acceleration would be the issue, then higher frequency (shorter lines) would be worse, but that is not the case. This really is a case of hitting a resonance frequency at just the right line length. Of course, if you use lower acceleration, you can make things so that you never hit that frequency because even very short infill lines will cause a lower frequency than the resonant one, but I don"t have a CoreXY to print slower than my Ender3.
That's an interesting thought. Feeding some of these values for specific machines back into the slicer could be very handy.
Great Video as usual, can you please do a video on the Duet solution please.
Great suggestion! I'll look into it!
Fantastic video Chris! Thanks for all those guides, you're helping people a lot.
Does Input shape need a powerful motherboard for working correctly? I just got a SKR 1.4 + TMC2209 and don't know if this will run this new feature.
Thanks is advance!
Thank you! Any 32bit board now a days will run it. Your board will work great!
@@ChrisRiley Hi Chris. I also have an SKT 1.4 Turbo with TMC2209, I have followed your great directions but unfortunately i run into a memory over run issue, i looked for a solution but it looks like its an A architecture limitation in the board, there is an article on this but my reply gets deleted if i add the link.
Thanks Chris. I only compile marlin maybe 2 times per year. It’s such a pain to re-learn how to use vscode. Might be easier if you’re printer is a standard configuration. If anything drives me from marlin to Klipper it will be vscode.
You're right, if you don't use it all the time it's a struggle. Hopefully we can help make it a little more straight forward.
Been using the 2225´s from GH Enterprises with Smart Shaper for marlin pretty decent so far, but tbh in fine detail I still have to print slow to avoid ghosting.
Agreed, I see the same thing a lot.
@@ChrisRiley agreed as well.
Hey Chris, whats up? Cant wait to hear your feelings on the Mk4 i must honestly say it was a bit of a let down considering the 5 years between that and Mk3. Also cant believe you now have 63K subscribers, your doing good :) Best regards Espen
Thanks, I'm looking forward to the live stream!
Chris, do you have a list of the M codes used in 3d printing. I don't need them often so I forget them. Thank you Great video.
You're welcome! I still have to go back to the REPRAP list from time to time. marlinfw.org/meta/gcode/
@@ChrisRiley Thanks.I will download. I also fount the Marlin Help which explained each M code. Thanks a bunch11
Whats a good rule of thumb for calibrating the acceleration and jerk? Should we do that before or after input shaping?
They all go hand in hand. It's hard to give a go value because they are going to change as you want to move faster and faster. On an i3 machine I like to start with 3000 accel and jerk value of 8. Then start ramping up accel from there.
@@ChrisRiley Oh right so im probably being a bit conservative with my coreXY acceleration being around 2500 for X and 2000 for Y.... Im using junction deviation at the moment and I think I need to increase that for sure.
Как всегда Крис круче всех! Ни кто не объясняет лучше. Собираешься ли ты сделать обзор на новый Prusa MK4? Очень хотелось бы видеть обзор от тебя.
Находится в пути!
I really hope that Sovol will add input shaping to the SV06. I've tried compiling it myself (Unfied 2 firmware for the SV06), but didn't work properly. I really don't like customizing Marlin and compiling it to flash on the printer. 😅
That would be cool!
Do you recommend turning off junction deviation when tuning input shaper? Do they impact each other? Should it be used with input shaper or would classic jerk better?
Both JD and CJ and going to work fine with IS. I personally think CJ is easier to use and I don't see a lot of difference with either one.
Excellent tutorial, Chris :)
I'm running the professional firmware by Mriscoc and within the advanced menu of the firmware I have an option called "step smoothing"
Do you know if this would have an effect on the input shaping function?
I've followed your tutorial, but I'm not seeing any changes in the ringing with input shaping enabled or disabled, even with the correct M593 code in the layer change section in prusaslicer.
Thanks! Interesting, not sure what step smoothing is referring to, so I'm not sure. I have seen where a printer doesn't so much different if the print head is really light and there isn't much ringing to start with.
Thanks Chris;
Do you know if it is possible to edit the Prusa Mini config file?
You can edit the .ini file, but I wouldn't know what syntax to use. Good luck!
before the 2.1.2 we were following another type of calibration and had to calibrate damping too. but now there is nothing about damping calibration. i was getting better results for both axis when i was using 0.10 damping
Good call out, I want to go back and try to dial it in further when different models become available.
I was really hoping to implement this in my MK3S+.. but not really good enough to compile firmware
I would be really hard I think to get this all working on an 8bit board.
awesome video as always 😄
@ChrisRiley: i know it's an older board, but can you please make a video on getting the MKS Sbase 1.3 working with the RepRap Discount Full Graphic Smart Controller?
there are some extra steps, to get it working... there are some notes in the marlin 2.0 example config, but it's not exactly a tutorial... you are so much easier to understand.
I'll look into it!
@@ChrisRiley yaay 😁
So, whether to go back to Marlin with IS and PA or keep my Klipper setup!
What is the point of going back to Marlin if you have a klipper setup already? A screen?
@@RusakovLA I liked the screen on my printer!
🙂👍
@@sargfowler9603 You can get a screen that connects to the pi if you are running klipper...or just use a phone/tablet. I sort of think that going back to Marlin would be a step backwards, but everybody has different needs/wants/preferences.
MeldMerge is a great tool to find text file differences. Wirks in Windows, Mac, Linux.
Thanks for the tip!
It's probably a good idea to make sure your belt are correctly tensioned before doing this, just in-case they are worn slack as I presume that if you do this and then realise that your belts are stretching then you are going to have to do it all again.
Thanks for the tip!
Yeah, I enabled input shaping in firmware and my printer would not boot (or rather got into continuous boot loop).
Just commenting out the #define for input shaping made it work again
Repeatable
Damn, what MCU is your printer running?
@@ChrisRiley It's a 2560 Mega, RAMPS
@@NickBR57 wondering if it filled up all the memory to an unsafe level.
@@ChrisRiley Marlin Auto Build reports 7274 of 8192 bytes RAM used and 196972 of 253952 bytes Flash so it doesn't look like memory.
I did notice a games menu the other day (!!! Why would I want games on a printer ???) so shall look for that and remove it to free up more...
...
Now shows 7150 RAM used and 190162 Flash.
But even before, there should have been plenty.
Thanks for the suggestions though Chris 😃
@@NickBR57 It looks like you have enough, but it wouldn't hurt to try. Search this to shut off the games, // Frivolous Game Options Also, to really seal the deal, uncomment this line, //#define DISABLE_M503 it saves a lot just as a test.
Do you know if its possible to cut enough things to get IS to work with the BTT SKR 1.4 Turbo?
Keep getting errors about not enough ram.
Or is the SKR 3 in my future?? heh
Thanks for the videos, they have helped me immensely...
1.4 should have more than enough memory to use every, no sure what is causing the issue.
@ChrisRiley thanks for the reply. It's been a while since I did anything with the firmware, I prolly screwed something up. Thanks again, love your videos...
Have you looked into the new Arduino GIGA R1 WiFi for running Marlin?
No, but I'm interested.
Remember to rotate the part by 45 degrees if you're on coreXY
Thanks!
Do you have linear advance enabled at the same time, or do the 2 together cause issues?
You can do both at the same time, because one can help the other out, but they are tricky to get in tune together.
Am I correct in thinking that you turned back on Auto Cooling and Acceleration Control once all the testing was complete? Does Leaner Advance stay off?
Correct, I did turn them back on. Linear advance can be turned back on, but it will probably need to be tunned now that input shaper is on.
The input shaping tests always fail on my printer, it reaches a certain layer and then my Y no longer moves, but X goes back and forth creating a large blob. Do you have any ideas on how to fix this? I've experimented and it seems to be related to the M593's D value? D0.9 will print, D0.8 is likely to fail, D0.2 always insta-fails. I'm kinda stumped, as having the D value high results in the print having barely any difference at all from input shaping
Interesting issue, I have never seen that. I'm not exactly sure how the damping value works with this. Sorry not much help here.
Would you tune linear advance before or after input shaping?
Klipper recommends input shaping first.
I agree. Because it helps you get the harp corner shape and then linear advance can reduce the blobs.
Otherwise, you'll tune linear advance to a too high value and then have cutoffs on your corners
IMO all the calibration should be without LA as in video, but then recalibrate LA and use it to get the final comparison (benchys in this case).
Although LA and IS are totally different, they achieve one task - get improved quality without sacrificing speed.
👍🙂
Is input shaping dependant on speed? If I tune for say 100mm/s and get my settings dialed in, will I need to retune different values for say 150mm/s or 60mm/s?
After you get it dialed in, the speed really shouldn't matter. The faster you go, the worse the ringing might be, but the resonance frequency shouldn't change. If you do increase speed, if might be worth another test just to see if it can be improved.
how do you open 2 instaces of marlin so you can have both on at the same time? i tried to do it and it wouldnt it always replaced one with another
I always use notepad ++ with the compare plugin. Easier for me to see the config files side by side.
Does the firmware's extruder Feedrate affect the performance of Input Shaping?
That is a great question. I would think that it does, but I have never really tested it.
Doesn’t the print itself add mass to the bed? Is there a way to compensate for that at all?
That is a great point and it does for sure, I would think we could calculate that somehow.
Hi, i have a question, after i have my ideal input shaping frequency, can i restore my K=0 value to my ideal K for linear advance? Or both things are incompatible
So when you find your IS value. Run the LA calc as the same speed and pick what value looks best.
Can you still use linear advance with input shaping? Or do they not play well with each other?
You can use them both together, but they will impact each other so you have to tune for it. I would get IS dialed in first, then tackle LA.
@@ChrisRiley cheers mate
Chris, you never discussed the dampening like I was talking about in my previous comment on your other videos. Why didn't you cover that?
Honestly, I don't fully understand it, I will get it on the next video.
@@ChrisRiley You and me both! My guess is the dampening is how it speeds up and slows down almost like jerk, but I honestly have no idea. I am almost to the point where I am going to leave Marlin behind and just put my efforts into learning Klipper.
So you're trying to find it's resonance frequency to negate it?
Correct.
Why would you enable LA in Prusa? It should already be on once you have set it up.
Prusa slicer for the Prusa profiles actually change the LA value all through the print based on certain factors. Also, other Prusa machines use the same filament profiles, so resetting it at the start of every print just to be safe can be useful.
Adding G92 E0 throws an error message;
"G92 E0" was found in layer_gcode, which is incompatible with absolute extruder adressing
Remove it from the layer gcode section, you might not need it any longer.
I've seem really cheap "Portable Input Shaper's" or Aliexpress. If the idea that you attach it to the print-head, run with the printer, download the data it collected and then input the data in to Marlin, Klipper etc. or am I misunderstanding the concept of it ??
I haven't seen those, but I'm guessing yes, it would plot out the data you would then input as a CSV file.
I have an unmodified MK3S. I have a Raspberry Pi 3B. I want to do the same and improve the print quality . What do I need to do? Can I just rewrite the firmware without modifying the printer?
It would be hard to do this on a stock MK3 as it's an 8 bit board. It might not be able to keep up to use IS.
Dumb question, why is the y crossing the x axis and the y crossing the x axis? I mean, when I move y the bed goes form front to back not side to side. Is this part not oriented wrong?
Right, so, they're switched because we are looking for vibration cause by direction change. The part is made to introduce quick stop and starts.
How would you do this on a prusa mk3s?
The only way at this point would be to flash it to Marlin. Even then, I'm not sure the 8bit board achieve benefit from IS.
I have a BTT mini E3 V3.0 . Without input shaping being enabled Marlin compiles fine using Platformio which says that is has only used 12.2% or RAM and 50% of flash. When I enable INPUT_SHAPING_X and INPUT_SHAPING_Y and try to compile it I get the error "firmware.elf section `.bss' will not fit in region `RAM'", which suggests it's too big for the RAM. But surely there should be plenty!! Any one any ideas how to fix this??
Wow, that doesn't sound right for sure. Not sure what might be taking up all the memory or if this is just a bug.
@@ChrisRiley It's weird just how many times you make Marlin changes look easy and when I do the same Marlin becomes a pain! I think I need to get on the forums to try to solve this.
Great video thanks, but thats some seriously subtle changes on your prints. I cant see any difference at all on my version. On your I can sort of see what you are talking about but wow its the sort of thing that needs a magnifying glass and angle the light just right and even then its tough to see.
Totally agree, we are getting down to fine details here. If you have a really light print carriage, you might not see any issues at all.
Cool to see an implementation. Cool to try but I’m scared of doing stuff like this for fear of bricking a part.
I was thinking about these sort of things last few days watching mk4 vids and mention of IS. Really need to do it live on the fly which is why you need the accelerometer. But that would be more applicable to the extruder on corexy. Than a bed slinger. As it has two axis it monitors. Ideally metering x and y and z on a slinger would be tricky to do live.
On the live checking as the parts get placed they get heavier. So if you filled a build plate to max it will have different weight than in first few layers, causing different harmonics.
The other thing I was thinking about as spool get lighter the moment on spool holder will reduce so that would cause the extruder to change relative to y axis. So there would be an inherent variation over the 200+ mm. Technically all prints would bend however we don’t see it. Be interesting to see the impact on z when a 1kg spool is hanging off frame with an offset vs one in line with frame. If it affects print quality or creates a bend.
I just tied running my prusa at 999% last night - first layer was great and then it went pear shaped. Killed print as a big blob fell off extruder. IG guess first layer is 20x 10 so 200mm/s but the 2nd layer is 60x10 do 600mm/s the filament didn’t really adhere look like a hairy scrotum. I got it run at 300% on some parts but time of print didn’t reduce that much. Handy to know if you’re in a rush for something. You can increase speed - did notice more ringing though…
Hey, thanks for the comment. You are right, active monitoring would really be interesting. Knowing some of the folks I do, I'm sure someone will try and work that into a design.
Does this work on 8bit boards?
I speny only 1 evening with Anycubic Trigorilla 8bit, and its have issues with memory. After disabling some features to freeup memory it flash board with no errors. After enabling in menu Input Shaper with tmc2208(standalone) just become crazy and Marlin making reboot. Ithink I need to play with frequency
I tried it with ramps on a 2560 mega and a4988 drivers. With IS enabled I had some crazy stepper skipping.
Chris could you please try it on an 8bit board.
Great video. Keep up the great work 👍
I can give it a try and see if we can get enough memory freed up to make it work.
Some here mentioned weird motion. That might be fixed now in 2.1.x and bugfix-2.1.x but don't quote me on that. We'd love to get more feedback at the GitHub project so we can sort out any remaining issues with this feature on AVR and get those fixes into the next release. It is a pretty optimal implementation, but quite new, so we haven't seen every edge case. Thanks for your comments!
@@ScottLahteine Hi Scott. Many thanks to you and all the selfless developers for keeping the Marlin project running. I think I'll try the latest bugfix and see if there's a difference; I last attempted Marlin IS about a month ago, so things may have been corrected since then. Scott, is there an easier way for noobs like myself to report bugs and suggest features other than github? I've just gotten comfortable with configuring the firmware and understanding the structure and would love to contribute, but github is a bit intimidating.
Im not use to the background of your video looking this nice.
Lol, thanks!
Great video but way to advanced for a common man ( me )
Hang in there. Thanks for watching!