Pressure advance jitters on small moves



  • Hi,

    Im testing pressure advance atm (RFF 1.19.2), with a short bowden and 0.2 k factor I can see some small but noticable benefits.

    One thing I just noticed is that even on very small zigzag moves, the extruder goes crazy and jitters back and forth. I saw a similar description here (https://www.duet3d.com/forum/thread.php?id=1999#p26172 point 2). Is there maybe a way to keep the small k factor but define a certain minimum move length?

    Cheers



  • That's what I'd like to see too but it seems that some users see benefits when doing arcs. That being the case then it has to be applied to small segments I guess. My feeling is that pressure in the system builds up as a function of the length or time of a particular move. Therefore with very short moves, there isn't enough time for pressure to build up so pressure compensation shouldn't be needed for short moves. I think what screws it up is instantaneous speed change or jerk as it is called. What this does is effectively chain a sequence of short segments together so that pressure builds up over the whole sequence and therefore needs compensating for at the end.

    TBH I don't fully understand how the algorithm works but I think it applies a compensation to the acceleration of the extruder at the start and end of a move. If that's so, then maybe increasing jerk should limit the acceleration and deceleration and therefore limit the the effect of pressure compensation? Dunno. Every time I try to figure out the relationship between acceleration, jerk and pressure advance, I end up with a headache. No doubt David will be along soon to put us right.



  • I had similar issues on small details, when I reduced my extruder jerk to a very low value. Might be worth playing around with it.



  • @deckingman

    Therefore with very short moves, there isn't enough time for pressure to build up so pressure compensation shouldn't be needed for short moves.

    Yeah thats what I thought as well. On top of that I notices that it actually stopped extruder and I also heard some extruder skipping. I can imagine that it helps in some cases, but there should be an option nevertheless, maybe something for the firmware wishlist.

    @DjDemonD At what values did you start and how much did you reduce to "make it work"?



  • I'm seeing the same effects. I tested with a value of .05-.15, but have ultimately turned it off b/c I don't like watching my poor extruder jitter like that.


  • administrators

    My extruders jitter on small zigzag moves, but they run quietly and it hasn't done them any harm. One is a Titan, one is an Ormerod extruder with printed gears, and one is a Chinese ungeared extruder. OTOH if I increase the extruder jerk beyond about 400 then the Titan in particular starts sounding bad. As long as you don't have extruder jerk set too high, the jittering happens smoothly.



  • @sungod3k:

    @deckingman

    Therefore with very short moves, there isn't enough time for pressure to build up so pressure compensation shouldn't be needed for short moves.

    Yeah thats what I thought as well. On top of that I notices that it actually stopped extruder and I also heard some extruder skipping. I can imagine that it helps in some cases, but there should be an option nevertheless, maybe something for the firmware wishlist.

    @DjDemonD At what values did you start and how much did you reduce to "make it work"?

    I honestly can't remember. I think it was an E jerk of just 2-3 I was using, which with pressure advance on was causing noticeable slow-down around small details, I increased it to 10 and it became less of an issue.



  • with the nimble the original suggestion recommended a jerk of 1 which really slowed everything to a snails pace increased to 40 and it became usable again.

    Doug



  • @dc42:

    My extruders jitter on small zigzag moves, but they run quietly and it hasn't done them any harm. One is a Titan, one is an Ormerod extruder with printed gears, and one is a Chinese ungeared extruder. OTOH if I increase the extruder jerk beyond about 400 then the Titan in particular starts sounding bad. As long as you don't have extruder jerk set too high, the jittering happens smoothly.

    The extruder jerk may be the cause of mine. It is pretty high. I will lower it to 400 and test again. OTOH, with my jerk being higher, it seems like it is just highlighting the jitters. Wouldn't the jitters still exist with a lower jerk since the commands would still be being sent to the extruder motor?

    When testing this last night I was printing "Phil A. Ment", Matter Hackers little astronaut, and the jitters were pretty constant. They seemed to occur on all permitters, even the longer smoother ones.


  • administrators

    The jitter will occur between any pair of moves that require the print head to decelerate. That's the nature of pressure advance. If it's happening on curves, that is a sign that your XY jerk setting may be too low.



  • @dc42:

    The jitter will occur between any pair of moves that require the print head to decelerate. That's the nature of pressure advance. If it's happening on curves, that is a sign that your XY jerk setting may be too low.

    My XY(Z) jerk is at 2000.



  • @Dougal1957:

    with the nimble the original suggestion recommended a jerk of 1 which really slowed everything to a snails pace increased to 40 and it became usable again.

    Doug

    Yes, that might have been it come to think of it, well remembered. I'm using 40 too, no issues now. Did a 1 hour print on a caliblock with the nimble on the kossel xl, I use very minimal pressure advance with the nimble, the idea being just to account for the twist in the cable before it generates movement at the extruder.



  • @DjDemonD:

    @Dougal1957:

    with the nimble the original suggestion recommended a jerk of 1 which really slowed everything to a snails pace increased to 40 and it became usable again.

    Doug

    Yes, that might have been it come to think of it, well remembered. I'm using 40 too, no issues now. Did a 1 hour print on a caliblock with the nimble on the kossel xl, I use very minimal pressure advance with the nimble, the idea being just to account for the twist in the cable before it generates movement at the extruder.

    What is your pressure advance setting, may I please ask? I'm moving to a nimble as soon as it gets delivered.

    Also, DjDemonD, thanks for your mount design for the nimble on the smart effector. I'm planning on using that.


Locked
 

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