S-curve in marlin 2: seems like a huge quality improvement.



  • @warbunnies said in S-curve in marlin 2: seems like a huge quality improvement.:

    @burtoogle said in S-curve in marlin 2: seems like a huge quality improvement.:

    I have a flex3drive on my Kossel XL which is pretty light and it still suffers from ringing. I can reduce the ringing by reducing acceleration but then I get fat corners. So it would be nice for me if this alternative acceleration scheme is implemented and it does actually perform as well as people say it can. I'll certainly be trying out any betas if/when they appear.

    What acceleration are you using? My k800 xl did not have much issue at 1400mm/s acceleration. Now I'm considering jumping up a lot higher... it just prints so darn good with marlin 2.

    I'm currently using 3000 and putting up with the ringing. I prefer some ringing to having fat corners.



  • @warbunnies said in S-curve in marlin 2: seems like a huge quality improvement.:

    This is what I expect from a delta with a decent extruder: https://imgur.com/a/PAoU2

    The model you are printing is curved so the low acceleration probably doesn't really matter. You can see some rounded corners on the support but it's hard to tell in that image how bad they are.

    Anyway, I am happy enough with the flex3drive in terms of print quality.



  • I'd love to see photos showing the improvements made from S-curve acceleration.



  • @burtoogle said in S-curve in marlin 2: seems like a huge quality improvement.:

    @warbunnies said in S-curve in marlin 2: seems like a huge quality improvement.:

    This is what I expect from a delta with a decent extruder: https://imgur.com/a/PAoU2

    The model you are printing is curved so the low acceleration probably doesn't really matter. You can see some rounded corners on the support but it's hard to tell in that image how bad they are.

    Anyway, I am happy enough with the flex3drive in terms of print quality.

    Yes I agree. That comment wasnt about s-curve. BenDiesel was talking about using a different extruder and I was arguing that the flex drive was just a drop in quality from a normal bowden/direct drive.



  • @bot

    What photo would you like to see? Marlin 2 with & without s-curve or smoothieware vs marlin 2? I'm a bit busy this week but I can set up the firmware and make something during my lunch breaks.



  • @warbunnies I'd be more interested in a video of it in motion comparing S-curve and non, all else being equal.



  • Videos would definitely be better, but often lack the perfect angle and lighting to show the differences in extrusion.

    Yes, photos showing before and after, and indicating which speed/accelerations were used and possibly even print times.

    We need to compare the time spent and quality of the prints. If the prints are of the same quality, but they're faster with s-curve, that's an improvement. If they're better looking with s-curve but with no speed benefit (as would be evident with actual print time) then s-curve still has a benefit. Etc.



  • Great to know that's on dc42's whatchlist



  • @bot

    I'll do my best. I've only done video editing once and only have my phone... so I'll probably stick to pictures. But I got the firmware baked at work today so I should be able to make some simple calibration cubes after work. I upped my acceleration to 3000 so it should force the ringing to be more visible. I'm not sure I'll have time to check print times but will definitely try.



  • @warbunnies don't bother with video editing take your smartphone and push it to YouTube 🙂



  • Okay I'm doing more testing but I am honestly a bit confused with my initial results...

    With an acceleration of 3000:

    Turning on the S-curve feature does not seem to do anything. Anything!
    The prints are so similar that the camera on my phone cannot pick up any difference & asking friends to blindly rate has produced indeterminate results.

    Even more perplexing is that both prints took almost exactly the same amount of time. The s-curve was faster by 5 ish seconds over a 1hr50ish min print. If anything I would have assumed it would have been slower because the acceleration is not constant...

    They also produced the same amount of noise.

    My conclusion is the the marlin firmware (with or without s-curve) is just better than smoothieware. 1400mm/s use to be a hard cap for my printer but the 7000mm/s acceleration im testing now suggests I can tune my printer much higher. I'll let you know if the 7000mm/s produces artifacts worth taking pictures of. I may also try increasing the jerk value to see if it has any effect.

    Also my small delta may be a poor candidate to test these features. The effector is less than 100g total & its a relatively small machine 260mm D x 361mm H & overkill structure so there's just nowhere near enough weight moving around to tell if the s-curve is helping. If it was easy to do/ i had more free time, I would stick it on my large delta 325mm D x 680mm H with the added weight of a flying extruder and see if the s-curve helped. If anyone has a heavy effector and some free time, I'd love to see if it does anything.



  • @warbunnies said in S-curve in marlin 2: seems like a huge quality improvement.:

    My conclusion is the the marlin firmware (with or without s-curve) is just better than smoothieware.

    I probably shouldn't comment because I've never been hands on with smoothieware on a delta. However, I have used Marlin 1.6.6 and above quite a bit on Delta's, and from reading about smoothie... well... I believe your conclusion is probably spot on.



  • @danal

    I actually think I found the issue. The name was recently changed from #define BEZIER_CURVE_SUPPORT to #define S_CURVE_ACCELERATION

    & whoever did that missed 1 part of the code that still used #define BEZIER_CURVE_SUPPORT

    That's what you get for beta testing. I'm trying the firmware with both defined and should actually be using the s-curve feature.


Locked
 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.