Vertical lines vs. geared extruders

gloomyandy

I wonder how well microstepping works at these lower speeds and relatively high motor loading particularly with the use of these smaller motors? I seem to recall that there is a relatively large reduction in torque/holding power at the intermediate microstep positions. Perhaps the actual movement is not as smooth as you might expect?

o_lampe

@gloomyandy
Worst thing would be, if all the torque pulses from various sources come into sync and add up.
Filament-Tsunami

Luke'sLaboratory

@zabana

Sure. I can certainly print more at .2mm layers. I typically print at volumetric limits, 25mm3/s for Jessie Filament with this setup, meaning if I go lower layer heights, the toolspeed will be significantly higher.

My printer these prints were off is a Stablebot core. Mechanics are Corexy. This is an injection molded orbiter with the 20mm extruder. Running .35A rms, x64usteps. Hotend is a slice magnum skeeter with .6mm vanadium nozzle. Klipper. 7k accel. Pressure advance enabled (tho this should only matter for the corners).

hackinistrator

i think the bondtech style extruder gears are the main cause for those issues .
the tensioning mechanism on bondtech style extruder effects the backlash . more tension = gears fully engaged . less tensions = gears are separated , causing backlash .
too much tension and you're chewing the filament .

CCS86

@jens55 said in Vertical lines vs. geared extruders:

@ccs86, my BMG LGX uses 405 steps/mm and it has a 3:1 gear reduction build in and I agree that the steps are not easily visible although in a print you might still see an overall pattern. I think an extruder without gear reduction is around 120 or so steps per mm which is considerably coarser than your 709 steps/mm.
Here is my math .... please double check and correct me where/if I have made an error.
1 step = 1/120 mm of 1.75 mm filament
filament is (1.75 /2)^2 x 3.14 = 2.4 mm^3 volume of filament per mm.
Since each step is 1/120mm then each step is 0.02 mm^3 of filament
Printed filament (0.4x0.2) is 0.08 mm^3 volume per mm laid down
So we lay down 4 mm of filament for every full step of the extruder.
At a micro stepping rate of 16 we lay down 0.25 mm of filament or 4 microsteps per mm laid down.
This is certainly visible in itself but you have to keep in mind that each step has a flow curve associated with it which makes the individual steps more visible. Then you put all of this into a largish printed area (let's say 100mm x 100 mm) and even though the eye might have an issue looking at the individual line and see the flaws it is remarkably more accurate to see a pattern in a larger area and you will see this pattern.

So to sum it up and even if my math is incorrect, I was going under the assumption of a standard extruder where the motor turns the gear that feeds the filament and you would see the steps. When reduced in a 3:1 gear reduction like the BMG, steps are much less visible. At 709 steps/mm
the end result would be even smoother.

I think your math is correct, depending on your microstepping. I am pretty sure that 709 step/mm on the Sherpa mini is considering x16 stepping, not full steps of a 1.8* motor.

CCS86

@hackinistrator said in Vertical lines vs. geared extruders:

i think the bondtech style extruder gears are the main cause for those issues .
the tensioning mechanism on bondtech style extruder effects the backlash . more tension = gears fully engaged . less tensions = gears are separated , causing backlash .
too much tension and you're chewing the filament .

What issues?

Backlash in Bondtech gears only affects the idler's ability to drive the filament. Especially at lower flow rates, there is essentially no demand for the idler to drive. Only once the primary gear starts to slip on the filament is the idler gear truly loaded.

droftarts

@ccs86 I agree. Low tension isn’t the problem, it’s high tension that causes cogging issues in the Bondtech-style gears. And as @gloomyandy (I think) pointed out, the lack of torque at some microsteps causes the movement to be delayed, then catches up as the micro stepping gains more torque towards the full step (or wherever there is more torque), which creates a pulsing in the extrusion.

I had a look on the Bondtech website to try and find correct tensioning advice, but couldn’t find any.

Ian

CCS86

@droftarts said in Vertical lines vs. geared extruders:

@ccs86 I agree. Low tension isn’t the problem, it’s high tension that causes cogging issues in the Bondtech-style gears. And as @gloomyandy (I think) pointed out, the lack of torque at some microsteps causes the movement to be delayed, then catches up as the micro stepping gains more torque towards the full step (or wherever there is more torque), which creates a pulsing in the extrusion.

I had a look on the Bondtech website to try and find correct tensioning advice, but couldn’t find any.

Ian

That's a very interesting theory. Basically, regardless of micro-stepping setting, you have a periodic torque output, depending on distance from the closest full step. This would cause a similar (but inverse) plot of positional error of the extruder, given a constant filament force. If this is true, I can only see more mechanical reduction and/or smaller step angle to improve the situation.

hackinistrator

@ccs86 said in Vertical lines vs. geared extruders:

@hackinistrator said in Vertical lines vs. geared extruders:

i think the bondtech style extruder gears are the main cause for those issues .
the tensioning mechanism on bondtech style extruder effects the backlash . more tension = gears fully engaged . less tensions = gears are separated , causing backlash .
too much tension and you're chewing the filament .

What issues?

Backlash in Bondtech gears only affects the idler's ability to drive the filament. Especially at lower flow rates, there is essentially no demand for the idler to drive. Only once the primary gear starts to slip on the filament is the idler gear truly loaded.

you never truly know which side (primary or idler) is actually loaded and driving the filament with this set up .
if gears are fully meshed , it can be that the idler is the one loaded and driving the filament .

i dont think its the microstepping torque spread issue , as i already tested extruders with 2 set-ups , one with bondtech style (not original , so there might be quality issues) and one with simple idler .
the simple extruder wins .
even prusa had many quality issues in mk3 since the move to bondtech .

CCS86

@hackinistrator said in Vertical lines vs. geared extruders:

you never truly know which side (primary or idler) is actually loaded and driving the filament with this set up .
if gears are fully meshed , it can be that the idler is the one loaded and driving the filament .

I don't agree. The primary gear is always driving. Torque input comes in through the primary gear and that gear is always engaged in the filament. How could it not be driving?

The idler gear is always in contact with the filament too. So, it is either out of contact with the main gear and idling, or in contact and also driving.

hackinistrator

@ccs86 lets say the the gear mesh is perfect - no backlash .
in this case , why do you think that only primary gear is driving ?
both spin at the same speed , it does not matter which one is directly connected to the motor .
in ideal world and perfect filament torque will be split 50/50 in this case .

CCS86

@hackinistrator said in Vertical lines vs. geared extruders:

@ccs86 lets say the the gear mesh is perfect - no backlash .
in this case , why do you think that only primary gear is driving ?
both spin at the same speed , it does not matter which one is directly connected to the motor .
in ideal world and perfect filament torque will be split 50/50 in this case .

Read my post again.

o_lampe

@ccs86
I've changed my 'Sherpa Mini Clone' to adjustable gears as you proposed. Will test it today...also allowed my stepper to be adjustable with the spur gear. That wasn't possible before, because the center hole wasn't oval.
I also beefed up a few strategic points, but that's not the point here.

zabana

@o_lampe HI!

How did it work?

o_lampe

@zabana
Surprisingly well!
I have some ringing, I try to dampen with input shaping, but that's another thread.
In the meantime I print at 90-100mm/s and 300-400mm/s travel speed. The result is 'sharper' than the same parts printed on my CoreXY. (shorter belts?)
I have an idea, how to drop the motor about 20mm to improve effector stability. But, that's again another thread.

oliof

@o_lampe your sherpa mini clone reminds me of the superfly extruder and the Sailfin extruder... a lot of minimalized setups around now.

Nuramori

@o_lampe if you want to drop it another 20mm or more, you can drive the outer edge. You can look at mine on how I did it, if you want.

edp drive

o_lampe

@oliof
It's actually @mrac1 's design, modified by @garyd9 and now modified by me. The offspring is the sherpa mini.

@Nuramori I want to use a rotating orbit gear. But without the star-gears.

hackinistrator

this guy is doing great analysis of this issue
https://www.youtube.com/watch?v=dL6u0UwPJOQ

it looks like he also thinks (like me) that this comes from direct drive dual gear extruders . i'm waiting for the next vid to see his conclusion .

bot

@hackinistrator Interesting video. I suppose we have to wait for the second video to find out why this is caused by what he says it is? I was watching with captions on and no sound, so maybe I missed some type of evidence that was presented?

Not an attack. The theory seems quite plausible, I'm just genuinely curious as to if any more info is out there or in the video that I missed.