Thanks Eddie. I hope word spreads and you get more subscribers. There are a lot of people out there with advanced CoreXY machines that can use this information your putting out.
stumbled over your post on hackaday, thanks for the nice video! my results an now WAY better. my results for the ender 3 v2 are x=37.24 // y=37.52, not as stiff as your printer but still good, tomorrow i'll try tuning my pressure advance
Purely anecdotal reference point, but setting input shaper on my Voron 2.4 and while it's difficult to measure accurately I believe I'm getting about 37.5Hz on X and Y. Reprinting the part with input shaper enabled right now. Very helpful video by the way. Thanks.
Hey Mark! Those are reasonable values, though in general at least on Vorons the Y should be a bit less than the X. I do think it's worth getting the accelerometer--for a few dollars you do get more accurate data (and quicker data), I've struggled a bit with the test prints some in the past with the calculated Y frequency being a bit "off".
@@eddietheengineer Thanks, I actually just finished the test print with those values and it is pretty incredible. The accelerometer is for another day LOL, I'm still recovering from the 2.4 build process. Just for reference, I used the 3D printed belt tension gauge that was posted on Discord to ensure my belt tension was close to equal and I believe it's working pretty well. More stuff to work on, but this is pretty amazing so far.
I printed half my voron parts with a mk3s+ with the included voron PIF slicer presets and didn’t have an issue. The ringing was surprisingly minimal. I didn’t slow the accel and stuff down till I was half finished printing the parts. But it didn’t make a huge difference in the print quality. I was surprised to say the least 😂
holy shit i can't wait to try. i can do 2200-2500 already and i have minimal ring at 2500. i wonder how far i can go with this! edit: 6000 at 100mm/s external no ghosting! but in irl models that ren't perfectly aligned cubes it's more realistically 5300 at 70mm/s outside! THIS IS HUGE!
Thanks Eddie! Can't wait to try this on my ender 3. I hope you have a video for the accelerometer as they seem cheap to buy if they give better quality.
Do you leave max_accel_to_decel in the config after that or do you rmove it. Also my guess is that you would lower the accel after that for regular print like 3 or 4K
the thing I'm confused about is I have now set a speed in my printer config file, but if I reprint the ringing tower and enter the commands as stated the speeds goes past the maximum set.
This is a great question! It's a bit complicated to really answer for sure, but in general things will change. Without inputshaper, your printer does tend to overshoot a bit, for instance, when printing infill. Input shaper will compensate and the toolhead may not overshoot anymore, or it's possibly it slightly undershoots. This would mean you would need to increase the infill overlap setting on your slicer to reflect how things changed.
@@eddietheengineer hey! i tried it. my ghosting is gone even at 6000 and i did a test print at 5300 perimeters and 6000 outside with 70mm walls and 100 external if I remember, could still break the 0.15mm one like i always do! tho on the actual model test some sides had SLIGHT ghosting. I ordered an accelerometer to try the digital route for fun! but even with my 10m initial calibration this is amazing! now i'm worried my printers screws are gonna get lose D: some parts are making noise at such speeds now xD
Hi, I did the test print. but mine doesn't even have a single ringing to it. Already check the slicer doesn't have any additonal/adaptive control. should i do anything for the input shaper?
Hmm, that's a good question! Are you sure the acceleration was set to 7000 and your print speed was 100mm/s? With 2 bottom layers, 2 perimeters, no infill, and no top layer my test print is right around an hour and 15 minutes.
the results with 7000 are really good, with which value did you end up in daily printing? is a high value of 7000 unhealthy for my ender 3 v2? or will this slowly kill my little baby? thanks again, this is a gamechanger
I'd double check to make sure that the gap in the ringing test print isn't too large. If you're willing to purchase a ~$5 ADXL345 sensor, it will not only tell you the best input shaper parameters, but will give you a suggestion on what your max acceleration should be!
I’m honestly not sure about S-curve and PA-remember input_shaper isn’t S-curve. Dmitri is still working on S-curve and it hasn’t been merged in yet. I did see improvements with PA, I just needed less than before 👍🏼
Dario another great question! Yes, the resonance frequency will change with varying belt tension. My initial test was around 25Hz and then I realized that the belts were a bit too loose. After tensioning the belts some more it was closer to 35Hz
How much time do you save by printing at those higher accelerations? Also, given that really tall thin prints above 200mm vibrate worse as you print higher, how does this affect your calibration? I can imagine it will change with height? I have printed tall thin (2mm) structures around 360mm tall and the oscillations at the top are much more expanded than those around the middle.
It’s a good amount! It’s both higher accelerations and faster print speeds. I know we did some calculations a while back but if I remember correctly it’s maybe around 25% reduction in print times?
It may have changed, but last time I checked the command had to be sent while a print is running-the tuning tower command will end once the print ends. If you send it before the print begins I think it doesn’t work-I may be wrong though! Also, you can send the tuning tower command multiples times and it won’t be an issue since the test sends a fixed value per Z height based on the command. If you have any other questions let me know!
@@eddietheengineer Btw. You mentioned accelerometer in the video. Would or could this be used to find the x,y resonance frequency values instead of running a print?
Jay Lee github.com/KevinOConnor/klipper/blob/work-adxl345-20200730/docs/Measuring_Resonances.md that’s the latest! It’s on a separate branch than master. Dmitri also has a newer test for it, but it’s still early
Thanks Eddie. I hope word spreads and you get more subscribers. There are a lot of people out there with advanced CoreXY machines that can use this information your putting out.
Wow - this is some amazing work. Subscribed you. Lots to learn
Same.
stumbled over your post on hackaday, thanks for the nice video! my results an now WAY better. my results for the ender 3 v2 are x=37.24 // y=37.52, not as stiff as your printer but still good, tomorrow i'll try tuning my pressure advance
Don't you need to set all "Acceleration Control" in Prusa slicer to 0 for this to work?
Yes! Either 0, or you can set it to a very high value greater than the acceleration you are testing input shaper at.
Purely anecdotal reference point, but setting input shaper on my Voron 2.4 and while it's difficult to measure accurately I believe I'm getting about 37.5Hz on X and Y. Reprinting the part with input shaper enabled right now. Very helpful video by the way. Thanks.
Hey Mark! Those are reasonable values, though in general at least on Vorons the Y should be a bit less than the X. I do think it's worth getting the accelerometer--for a few dollars you do get more accurate data (and quicker data), I've struggled a bit with the test prints some in the past with the calculated Y frequency being a bit "off".
@@eddietheengineer Thanks, I actually just finished the test print with those values and it is pretty incredible. The accelerometer is for another day LOL, I'm still recovering from the 2.4 build process. Just for reference, I used the 3D printed belt tension gauge that was posted on Discord to ensure my belt tension was close to equal and I believe it's working pretty well. More stuff to work on, but this is pretty amazing so far.
@@marklandsaat3696 That's awesome! I completely get the exhaustion--it's a lot of work to get a printer completely built and printing well!
I printed half my voron parts with a mk3s+ with the included voron PIF slicer presets and didn’t have an issue. The ringing was surprisingly minimal.
I didn’t slow the accel and stuff down till I was half finished printing the parts. But it didn’t make a huge difference in the print quality. I was surprised to say the least 😂
This is awesome 🙌
holy shit i can't wait to try. i can do 2200-2500 already and i have minimal ring at 2500. i wonder how far i can go with this!
edit: 6000 at 100mm/s external no ghosting! but in irl models that ren't perfectly aligned cubes it's more realistically 5300 at 70mm/s outside! THIS IS HUGE!
Thanks Eddie! Can't wait to try this on my ender 3. I hope you have a video for the accelerometer as they seem cheap to buy if they give better quality.
I'll definitely do a follow up with that!
Do you leave max_accel_to_decel in the config after that or do you rmove it. Also my guess is that you would lower the accel after that for regular print like 3 or 4K
You can comment out that line once you’re done with the test! It’ll default back to 1/2 of your max acceleration
Thanks Eddie, well explained
Glad you found it helpful! 😊👍🏼
Awesome Video!
the thing I'm confused about is I have now set a speed in my printer config file, but if I reprint the ringing tower and enter the commands as stated the speeds goes past the maximum set.
Thank you for sharing.
There is strictly no speed loss (assuming your acceleration is identical)
Looks good Eddie. Do you think this is a safe test for the Voron Switchwire? I picked up one of those accelerometers so I might text that way too
Yes! Definitely. You’ll have to test on the toolhead for X, then do a test on the bed for Y 😊
@@eddietheengineer thank you!!
this is very nice))) need to try it)
Have you tested part accuracy like tolerances etc at this speed?
This is a great question! It's a bit complicated to really answer for sure, but in general things will change. Without inputshaper, your printer does tend to overshoot a bit, for instance, when printing infill. Input shaper will compensate and the toolhead may not overshoot anymore, or it's possibly it slightly undershoots. This would mean you would need to increase the infill overlap setting on your slicer to reflect how things changed.
@@eddietheengineer hey! i tried it. my ghosting is gone even at 6000 and i did a test print at 5300 perimeters and 6000 outside with 70mm walls and 100 external if I remember, could still break the 0.15mm one like i always do! tho on the actual model test some sides had SLIGHT ghosting. I ordered an accelerometer to try the digital route for fun! but even with my 10m initial calibration this is amazing! now i'm worried my printers screws are gonna get lose D: some parts are making noise at such speeds now xD
Can I use input shaping and pressure advance?
Dario yes! I do. I recommend redoing the pressure advance tuning after you add the input shaping-I was able to reduce my PA from 0.08 to 0.04
Thanks. I had the same question.
how can i handle this option in Marlin Firmware?
Hello! Unfortunately Marlin does not support Input shaping
where can i donwload the ringing_tower.stl ?
github.com/Klipper3d/klipper/blob/master/docs/prints/ringing_tower.stl
Here you are!
Hi, I did the test print. but mine doesn't even have a single ringing to it. Already check the slicer doesn't have any additonal/adaptive control. should i do anything for the input shaper?
Hmm, that's a good question! Are you sure the acceleration was set to 7000 and your print speed was 100mm/s? With 2 bottom layers, 2 perimeters, no infill, and no top layer my test print is right around an hour and 15 minutes.
the results with 7000 are really good, with which value did you end up in daily printing? is a high value of 7000 unhealthy for my ender 3 v2? or will this slowly kill my little baby? thanks again, this is a gamechanger
I'd double check to make sure that the gap in the ringing test print isn't too large. If you're willing to purchase a ~$5 ADXL345 sensor, it will not only tell you the best input shaper parameters, but will give you a suggestion on what your max acceleration should be!
@@eddietheengineer thanks for the quick answer, i'll invest this 5 bugs for sure ;)
Amazing!
Ohh a question... We should turn pressure advance back on but using this with S_Curve is probably overkill. I mean use one or the other correct?
I’m honestly not sure about S-curve and PA-remember input_shaper isn’t S-curve. Dmitri is still working on S-curve and it hasn’t been merged in yet.
I did see improvements with PA, I just needed less than before 👍🏼
This cleaned up my Y axis ringing perfectly but didnt touch my X axis at all. I am still doing some testing but any idea why that would be?
Hey did you solve your issue? input shape is not effecting my y axis at all
what happens if the belts lose tension? does resonance change and should everything be recalibrated?
Dario another great question! Yes, the resonance frequency will change with varying belt tension. My initial test was around 25Hz and then I realized that the belts were a bit too loose. After tensioning the belts some more it was closer to 35Hz
How much time do you save by printing at those higher accelerations? Also, given that really tall thin prints above 200mm vibrate worse as you print higher, how does this affect your calibration? I can imagine it will change with height? I have printed tall thin (2mm) structures around 360mm tall and the oscillations at the top are much more expanded than those around the middle.
It’s a good amount! It’s both higher accelerations and faster print speeds. I know we did some calculations a while back but if I remember correctly it’s maybe around 25% reduction in print times?
You said AFTER you press print you have to execute the tower command. Isn't this put in just before you start printing
It may have changed, but last time I checked the command had to be sent while a print is running-the tuning tower command will end once the print ends. If you send it before the print begins I think it doesn’t work-I may be wrong though! Also, you can send the tuning tower command multiples times and it won’t be an issue since the test sends a fixed value per Z height based on the command.
If you have any other questions let me know!
@@eddietheengineer i think you have to run it before. In the instructions step 4 is execute the tower command. Step 5 is print the model.
Try it and let me know! I’ve always executed tuning towers after the print starts and has success but if it works both ways that’s great too ☺️
@@eddietheengineer Btw. You mentioned accelerometer in the video. Would or could this be used to find the x,y resonance frequency values instead of running a print?
Jay Lee github.com/KevinOConnor/klipper/blob/work-adxl345-20200730/docs/Measuring_Resonances.md that’s the latest! It’s on a separate branch than master. Dmitri also has a newer test for it, but it’s still early