E acceleration has to be limited to to E jerk/PA

[oliof]

@dc42 dropped this absolute truth bomb in discord today:

E acceleration has to be limited to the value of E jerk divided by pressure advance

I really would like to have this enshrined in the documentation.

Also, it would be great if the firmware warned on acceleration exceeding this limit, and allowing for an E acceleration setting that autocomputes accel by dividing jerk by PA.

[ctilley79]

@oliof

Can you give an example with typical jerk settings for a geared direct drive extruder motor?

[oliof]

@ctilley79 my Orbiter 1.5 on an EVA3 print head (roughly 60mm between feeder gear and melt zone) I used jerk 300 at a pa of 0.02 to 0.03, so 10,000 to 15,000 max accel with the limits above (my setting was at 6,000 accel, so well within limits).

[ctilley79]

@oliof Gotcha. I have my E jerk set ridiculously high at like 3000, so I'd never hit any limit with acceleration. It was my prior understanding is that for PA, you need high jerk values for E. I guess I need to reevaluate that.

[oliof]

@ctilley79 my jerk may be too low. Your understanding is correct.

[oliof]

@ctilley79 look, @Phaedrux dropped this other thing the other day: E Jerk can be very high, between 3000 and 6000:

https://forum.duet3d.com/topic/33948/very-low-acceleration-during-printing/6?_=1698255872645

[droftarts]

@oliof I’ll check with @dc42 and update the advice on M566 jerk for extruders, and acceleration. In the old days before PA it was usually set quite low, to avoid skipping or stripping the filament during advance or retract, but this seems to have changed with PA, with a knock on affect on extruder acceleration.

Ian

[oliof]

@droftarts thanks! Please don't forget to also update the configurator to provide sane values. I know it errs on the side of conservative values, but it should not be completely outside the ballpark mentioned in documentation (machine specific values notwithstanding).

[gloomyandy]

@oliof I know this topic is old, but I have a quick question about the assumed units in this discussion.

Normally jerk is given in mm/min but acceleration in mm/s/s. In the example above @oliof has a jerk of 300 (which I assume is 300mm/min) and with a PA of 0.02 this allows for an acceleration of up to 15000. However doesn't the jerk value need to be converted to mm/s before that calculation? If you do that you get a much lower acceleration limit of only 250mm/s/s! Even with the higher jerk values suggest above of 3000mm/min the limit is "only" 2500mm/s/s.

So my questions are:

1. Are the jerk values quoted above in mm/min?
2. should the jerk value be converted to mm/s before dividing by PA to give the acceleration limit?

[oliof]

@gloomyandy the values I quote are as used in my config.g, so yes there is a unit confusion error here. Which would put massive limits on e acceleration (although it is not entirely clear to me what happens if E acceleration exceeds the limit of E jerk/PA).

EDIT: OR the limits are to be computed as above and the unit confusion is irrelevant. One for @dc42 to clarify.

[droftarts]

@oliof Belatedly, I've added this advice to the Pressure advance page, and to M572. I also added the note about high extruder jerk to M566.

Ian

[gloomyandy]

@droftarts Did you resolve the question I raised above (https://forum.duet3d.com/post/327871) about the units used to set jerk and acceleration?

Perhaps @dc42 could comment?

[droftarts]

@gloomyandy thanks for reminding me, and it seems you are correct. I asked @dc42, he says:

Still correct but express jerk in mm/sec.

I asked about setting extruder jerk as in @oliof 's example (with extruder jerk set to 300mm/min / 5mm/sec), which would significantly limit acceleration to between 5 / 0.02 = 250mm^2 to 5 / 0.03 = 166mm^2 max acceleration. He said:

Set extruder jerk quite high to avoid acceleration being limited during printing moves, as long as this doesn't result in skipped steps. Extruder acceleration can be set higher if desired and that value will be used for retract and reprime moves.

So I think setting extruder jerk high as @Phaedrux suggested, to between 3000mm/min (50mm/sec) and 6000mm/min (100mm/sec), which should be okay on most extruders as it's really there to avoid limiting axis acceleration, the calculation becomes:
50 / 0.02 = 2500mm/s^2 to 50 / 0.03 = 1666mm/s^2 (for 3000mm/min)
100 / 0.02 = 5000mm/s^2 to 100 / 0.03 = 3333mm/s^2 (for 6000mm/min)

I don't quite know how this translates into the real world, ie if these speeds are even close to being reached. I guess you generally tune PA for the speed you're printing at (roll on conversations about non-linear/adaptive PA), and if you're printing really fast you'll notice odd extruder behaviour. I guess if you need to increase extruder acceleration to cope with PA at high speed, you'll need to increase jerk too?

Ian

[dc42]

Bear in mind that the purpose of the extruder jerk limit is to avoid the extruder skipping steps. So don't set it so high that you get skipped steps when using PA!

[droftarts]

@oliof @gloomyandy After discussing this with @dc42, I think I've misunderstood how extruder acceleration is limited under PA. The jerk / PA limit on acceleration is applied automatically by RRF, so the user doesn't need to set jerk or acceleration manually to take account of PA. The user sets extruder speed, jerk and acceleration at whatever the extruder can achieve in extruder-only moves (like retraction). The user only needs to know that acceleration is limited when PA is applied, to jerk / PA, so may have consequences such as slowing down or stuttering during printing moves, or skipping steps. This, at least, makes the documentation simpler! Which I'll now update, again...

Ian