For the past few months (as long as I've had the DuetWiFi), I've been getting really bad jerking when my corexy machine rounds curves. It happens at any speed, but it affects surface finish and turns into a terrible grinding noise at higher speeds (above say 60mm/s). This doesn't happen with any straight moves, diagonal or not. It only appears to happen between line segments, no matter what the curve radius is.

I have acceleration set to 2000mm/m^2. My usual jerk setting is 300mm/m, but I tried playing with it down to 10mm/m and up to 10,000mm/m, at which point I skip steps of course. It doesn't seem to affect the jerkiness in any appreciable amount.

It happens with both slic3r and simplify3d, with sliced models and the primer skirt.

I haven't payed too much attention to it because I only started to go for high speeds recently, but now that I know I can get 150mm/s, this jerk is killing me and my surface finish.

Does anyone have any ideas?

Edited 2 time(s). Last edit at 01/24/2017 08:57PM by Masterjuggler.

It sounds to me that the model you are printing has too few line segments making up the curves.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I just tried printing a 15mm cylinder with a face count varying from 10 to 60 in increments of 10, and I played with the speeds to see if that affected the grinding sound. For the most part, I kept it set to 60mm/s.

The 10 and 20 count cylinders didn't sound grindy, but each corner definitely jerked. Once I hit 30 to 40, it sounded like someone using a wood rasp on steel. It was less pronounced at 50, and even less so at 60, but was definitely still there.

On the worst face counts, 30 and 40, I tried varying the speed. Dropping below 30mm/s pretty much stopped it. Ramping it up to 150mm/s I realized the speed didn't go above around 60mm/s, so I started to play with jerk again.

I guess I wasn't testing it properly before, because jerk now definitely affects the grinding noise. For all previous tests I kept it set to 300mm/m. I have a direct drive with a large 76oz/in motor, so I kept it low. Bringing it up to 600mm/m reduced the noise considerably. 900mm/m almost eliminated the grinding noise at any speed and any face count. Going higher didn't change the noise at all.

I then tried cylinders at 5mm, 10mm, and 30mm. The 30mm wasn't any worst than 15mm. 10mm had more grinding than 15mm, and 5mm more than 10mm. I found the sweet spot for jerk is at 1000mm/m. Again, higher didn't yield better results.

So I've found one solution to the problem, but a 17mm/s instantaneous speed change seems excessive to me with a very heavy gantry. I'll have to do some corner ringing tests though to see if it's really as bad as I'm imagining.

So... how do you feel about working on dynamically varying jerk based on corner angle?

I don't see how dynamically varying jerk would help. Basically, if the product of the requested speed and the sine of the corner angle is lower than the configured jerk, the head will not slow down to take the corner (that's an approximation, but not far off). Whether the jerkiness sounds rough or smooth will depend on whether it excites a resonance in the mechanics. With a heavy print head, that's more likely.

Edited 1 time(s). Last edit at 01/25/2017 04:09PM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I'm working on taking the motor off the gantry and running a dremel flex shaft from it to a 25:1 worm drive on the extruder. I'll see what that changes if I can get it work. I don't really want to spend $90 on a flex3drive, but I know that this is an upgrade I want. I don't want to have to go back to bowden again, especially because I'm experimenting with flexible filament.

@ Masterjuggler. Your acceleration setting looks really low compared to what I'm using - wondering if that might be the problem?. You say in your OP that they are 2,000mm/min^2 which (divided by 60) would be 33 mm/sec^2 . Or is that a typo and it should read 2,000 mm/sec^2?

For info, on my heavy corexy (the Y carriage (including the x carriage) weighs in at 1,350gms and I have jerk set to 600mm/min, accels set to 1,200mm/sec^2, and max speeds set to 70,000 mm/min for X and Y. HTH

---

[somei3deas.wordpress.com]
[www.youtube.com]

Yes, I did mean 2000mm/s^2, thanks for pointing that out so I could clarify.

I don't actually know how much my gantry weighs. I'll have a chance to weigh it properly in a few weeks when I take it apart to replace the slowly failing PLA brackets with something stronger. When I made the printer, the crappy (to put it nicely) QU-BD Twoup I was using was only capable of PLA lol.

I bumped the acceleration down to 1200mm/s^2, and it definitely makes a difference. I'll have to play with it more over the next few days and see how it affects my jerk settings. Thanks for the tip.